How digital transformation shapes the performance of manufacturing firms: a study of China Open Access

In recent years, digitalization has been widely acknowledged by both academia and industry as a key driver of the Fourth Industrial Revolution (Plekhanov, Franke, & Netland, 2023). Since the early 21st century, major world economies have introduced strategies to promote digital economic development. For instance, in 2001, Japan launched the “e-Japan” strategy; in 2009, the United Kingdom released the “Digital Britain” plan; and in 2013, Germany introduced the concept of Industry 4.0 at the Hanover Fair, which has been actively implemented since. Similarly, China has prioritized integrating information technology with the real economy, elevating the digital economy to a national strategy and continuously refining its top-level design and policies (Sun, Fang, Li, & Wang, 2024).

These trends reflect the increasing global consensus on digitalization as an inevitable driver of economic transformation (Bi, 2024). Both policy frameworks and theoretical studies emphasize the advantages of digital transformation in boosting economic development (Wang, 2023). However, despite its potential, digital transformation poses significant challenges for micro-enterprises. According to the “2021 China Corporate Digital Transformation Index Research” by Accenture, only 16% of Chinese companies achieve meaningful results from digital transformation efforts. A McKinsey report even indicates that the failure rate of digital transformation for typical enterprises is close to 80%. The economic consequences of digital transformation are pivotal, as they influence the confidence and motivation of enterprises to pursue digital change (Niu, Wen, Wang, & Li, 2023).

As the cornerstone of China’s economy, the manufacturing sector faces numerous challenges and opportunities in the digitalization process. The critical questions are: How does digital transformation impact the operational performance of manufacturing enterprises? What are the underlying mechanisms driving this impact? In-depth exploration of these issues has become a major focus for both industry and academia.

The relationship between digital transformation and corporate performance remains a topic of debate in current research, with no consensus reached. Many scholars argue that digital management implementation enhances competitive advantages (Benner & Waldfogel, 2023), stabilizes export performance (Li, Shao, & Wang, 2015a; Li, Lu, Mittoo, & Zhang, 2015b), improves supply chain efficiency (Zhou & Wan, 2017), strengthens financial outcomes (Jeffers, Muhanna, & Nault, 2008) and drives steady organizational growth (Cui, Jiao, & Zhang, 2013). Based on a resource-based view framework, Nwankpa and Roumani (2016) found that active digitalization significantly boosts corporate performance. Similarly, Zeng, Zheng, and Li (2018), using data from 214 Chinese automotive manufacturing firms, demonstrated that digital technologies improve the stability of sustainable development performance. He and Liu (2019), analyzing data from A-share listed companies in China between 2012 and 2017 and confirmed that transitioning from physical to digital significantly enhances corporate performance.

Conversely, other studies have highlighted mixed or negative outcomes. Brynjolfsson and Hitt (2000) found that information technology adoption does not always positively affect performance, often showing no significant correlation or even a negative impact, leading to the “Information Technology (IT) Productivity Paradox” introduced by Robert Solow. Lou and Xue (2011) observed that enterprise information systems could initially reduce profitability but tend to benefit performance in the long run. Qi and Xiao (2020) revealed that challenges in integrating digital technologies with existing resources and operations often result in negligible short-term performance gains. Liu, Yan, Zhang, and Lin (2021), using data from China’s “Two Integration Pilot Zones,” identified a U-shaped relationship over time between production efficiency and digital transformation.

This study, based on real data from 2,346 manufacturing firms listed on the Shanghai and Shenzhen stock markets from 2011 to 2021, examines the impact of digital transformation on the business performance of manufacturing enterprises. The primary contributions are as follows: First, this research innovatively proposes a “Technology-Organization-External Network” analytical framework, making a theoretical contribution to understanding the mechanisms through which digital transformation affects business performance in manufacturing firms. Furthermore, the study investigates the dynamic marginal effects of digital transformation on the operational performance of Chinese manufacturing enterprises and analyzes the influence of key external factors, such as digital infrastructure, financial development levels and industry competitive intensity, on the enabling effects of digital transformation.

The remainder of this study is organized as follows. Section 2 reviews the existing literature and proposes the research hypotheses. Section 3 introduces the empirical model and data. Section 4 presents the empirical estimation results. Section 5 examines the underlying impact mechanisms. Section 6 further explores the dynamic marginal effects and the influence of external factors. Section 7 summarizes the main findings and provides policy implications.

Enterprise digital transformation has become a focal point of academic research. Many studies have concentrated on the indirect effects of digital transformation on innovation capacity and total factor productivity (Fang & Liu, 2024). However, existing research has yet to fully elucidate the underlying mechanisms through which digital transformation impacts operational performance. To address this issue, this study innovatively proposes a “Technology-Organization-External Network” impact mechanism framework.

In recent years, the rapid development of new-generation digital technologies has made digital transformation a central focus for both academia and industry. However, a unified understanding of the concept of digital transformation has yet to be established and its definition and scope continue to evolve. Table S1 summarizes the definitions and interpretations of digital transformation proposed by various scholars, highlighting the role of new-generation digital technologies in this process. These technologies, represented by the internet, artificial intelligence, big data, cloud computing and blockchain, have been key drivers of the digital economy (Tangwaragorn et al., 2024). For example, with the application of big data technologies, data empowers the production, sales and research processes on the supply side and, within the context of enterprise-customer co-creation, facilitates value creation through precise demand forecasting and generation. The integration of these technologies not only optimizes internal corporate operations but also enhances interaction with the external environment, fostering a more collaborative ecosystem. New-generation digital technologies are not merely updates or advancements in technology; their impact extends to multiple dimensions of enterprises, including organizational structure, business processes, production methods, business models, value creation and industrial ecosystems. Therefore, the understanding of digital transformation should not be limited to efficiency improvement; it should also include the disruptive changes it brings to internal operations and industry structure.

A significant amount of research has explored the relationship between digital transformation and firm performance, but the conclusions remain inconsistent. Some studies suggest that digital transformation may lead to an “IT investment trap,” thereby reducing firm performance, or its impact on firm performance is not significant (Vu, Van Nguyen, Nhu, & Tran, 2024). On the other hand, a larger body of research has revealed the multidimensional positive impacts of digital transformation on firm performance. Digital transformation facilitates technological advancement and strategic transformation (Qi & Xiao, 2020), breaks down traditional organizational boundaries (Yoo, Henfridsson, & Lyytinen, 2010), reshapes market competition patterns (Xiao & Qi, 2019), improves internal resource allocation efficiency (Zhang & Li, 2022) and enables cost optimization (Yao et al., 2022; Zhao, Wang, & Li, 2021; Zhao & Huang, 2022), thereby enhancing firms’ competitiveness (Hu, Chen, & Qiu, 2021).

Through the integration of digital technologies, digital transformation contributes to reshaping firms’ operations and value creation processes. Based on the alignment between manufacturing processes and business models, digital transformation in manufacturing enterprises can generally be categorized into three modes: (1) Digital value-added service transformation, which involves developing new services based on digital technologies, enhancing enterprise value and customer experience and increasing market share (Zhang, Ma, Pang, Xing, & Wang, 2023), (2) Digital integrated manufacturing transformation, which achieves intelligent production through the integration of cyber-physical systems, significantly improving production efficiency and resource utilization and reducing costs (Zhang, Wu, Li, & Chen, 2024) and (3) Digital dual value transformation, which leverages internet platforms to integrate customer data, optimize supply chain management and market forecasting and enhance response speed and market agility (Guo, Ke, & Zhang, 2023). These three modes also reflect the potential impact of digital transformation on performance improvement across different types of manufacturing enterprises.

Based on these findings, the first hypothesis is proposed:

The mechanism through which digital transformation affects the operational performance of manufacturing enterprises remains unclear. Most studies focus on the technological and organizational aspects of digital transformation and pay less attention to the impact of external networks (Chen & Xu, 2023). To shed light on the “black box” of the impact of digital transformation and gain a comprehensive understanding of the mechanisms, this paper further proposes the “Technology-Organization-External Network” impact mechanism framework, as shown in Figure 1. The technological and organizational pathways are typically regarded as internal mechanisms. The external network refers to the collective relationships between the firm and external economic entities, including relationships with upstream suppliers, downstream customers, investors, government departments and others.

1. Mechanisms at the technological level

Enhancing innovation effectiveness. Digital technology is considered a “general-purpose technology,” which generates substantial economic returns by stimulating research and development (R&D) innovation, integrating production and manufacturing processes and giving rise to new and efficient technologies and models. Innovation can be categorized into various types, including product innovation, process innovation, organizational innovation and business model innovation (Liu, Zhou, & Liu, 2019) and it undergoes stages such as opportunity recognition, concept development, product design, process design and commercial production (Li, Liu, & Han, 2022).

Digital transformation helps to improve the efficiency of firms in all stages of innovation. In the stage of opportunity identification, enterprises enhance the efficiency of uncovering market demands and innovation opportunities through digital transformation by leveraging big data analysis technologies. Enterprises cultivate agile organizational capabilities within the R&D departments, enabling timely tracking of market dynamics and reducing inefficient R&D investments.

From conceptual development to the commercial production stage, significant human and capital investment is required. Throughout this process, the progression from identifying business opportunities to the establishment of innovation plans requires iterative validation, careful deliberation and simulated calculations. The application of digital technology substantially reduces communication costs. The utilization of systems such as Computer-Aided Design, Computer-Aided Engineering and Computer-Aided Manufacturing provide substantial convenience to R&D personnel, significantly shortening R&D cycles and reducing development costs. By integrating historical manufacturing data with the data market, enterprises can effectively reduce trial-and-error time and material costs, thus enhancing the efficiency of R&D innovation. Although there are phenomena such as “patent bubble” (Lin, Ding, & Chen, 2025), the support of digital technologies—such as big data analytics and computer-aided systems—enables enterprises to identify innovative opportunities, optimize trial-and-error processes, reduce the cost of experimentation and integrate external innovative achievements. This enhances their ability to transform innovative outcomes into improved operational performance.

Based on the analysis, the following research hypothesis is proposed:

Enhancing labor productivity. From a technological perspective, the digital transformation of manufacturing enterprises is fundamentally a form of directed technological progress that responds to changes in relative factor prices (Tang, Li, & Xia, 2022). Related research has indicated that the application of digital technology represents a labor-biased technological advancement (Bai & Zhang, 2021). On the one hand, digital transformation in manufacturing companies can directly enhance labor productivity by simplifying work processes, using automation tools and programs to replace human labor in highly repetitive and complex tasks and transforming organizational forms and structures. On the other hand, the “technological unemployment” brought about by digital transformation increases a company’s human capital, thereby indirectly boosting labor productivity (Tang et al., 2022). Economic theory dictates that labor is one of the core input factors in business production and operations. With input levels remaining constant, an improvement in labor productivity can significantly enhance economic output and increase a company’s operational performance. Moreover, an increasing number of manufacturing companies are transitioning into digital value-added services. They are integrating digital services or products into physical products. Digital products and services have extremely low marginal costs, approaching zero. Under the influence of economies of scale, this raises labor productivity and consequently enhances a company’s operational performance.

Based on the above analysis, the following research hypothesis is proposed:

1. Mechanisms at the organizational level

In terms of organizational management, digital transformation brings about structural optimization and efficiency improvements. If a company’s digital transformation remains confined to the technological level, it is challenging to achieve success (Jiao, Yang, Wang, & Li, 2021). Digital transformation not only reshapes business processes but also integrates organizational innovation as an entry point, combining technological innovation to drive value creation (Xia & Lou, 2018). Therefore, digital transformation should encompass comprehensive changes, including value-creation methods, business processes, organizational structures, talent configurations and more.

Specifically, digital technologies promote the flattening of organizational structures (Qi & Xiao, 2020), enhancing management efficiency by accelerating information flow and simplifying departmental structures. Enterprises adopt platform-based management approaches, breaking down departmental barriers and enhancing cross-departmental collaboration (Liu & Liu, 2022). The application of digital management systems replaces some supportive, functional roles, streamlining departments and improving management efficiency. This leads to further optimization of resource allocation and enhances the dynamism and flexibility of processes.

In terms of labor structure, digital transformation introduces intelligent assets, enhancing employees’ digital skills and self-management capabilities, thereby reducing training and coordination costs. Additionally, the emergence of “virtual organizations” breaks down traditional organizational boundaries, saving management costs and achieving efficient collaboration. Companies can integrate digital products, digital services and value data into their products, forming data-service-product bundles (Chen, Huang, & Liu, 2020). Through digital platforms, they strengthen sales channels, optimize production processes and enhance product premium capabilities.

Based on the above analysis, the following research hypothesis is proposed:

1. Mechanisms at the external network level

Reducing supply chain concentration. Enterprise digital transformation has strong externalities, causing disruptive changes not only within the organization but also a profound restructuring of the industry chain (Liu, 2006). First, the application and upgrading of digital information technologies drive enterprises to transition from traditional linear models to networked organizational structures (Li, Li, & Zhou, 2020a). This shift has decentralized upstream enterprise sales channels and broadened channels for raw material procurement, thereby supporting the expansion of transaction channels, simplification of supply chain processes and reduction of supply chain management costs. Second, digital transformation enhances enterprises’ innovation capabilities, encouraging them to expand into higher-value segments and high-value-added areas of the value chain, thereby altering the internal distribution dynamics within the value chain. Simultaneously, through platform-based business models, digital transformation amplifies consumer heterogeneity, generating a “long tail effect” and fostering the emergence of integrated manufacturers. These developments further influence supply chain concentration.

Supply chain concentration refers to the extent to which enterprise partners are concentrated in the upstream and downstream segments. Currently, existing research on supply chain concentration and its impact on a company’s economic outcomes has produced two entirely opposite conclusions. From the perspective of bargaining power, supply chain concentration weakens a company’s bargaining power at both ends of procurement and sales (Tang, 2009), resulting in profit exploitation (Zhao & Li, 2023) and increasing financial risk (Itzkowitz, 2013). Conversely, from the perspective of operational efficiency, supply chain concentration can effectively reduce transaction costs and optimize production scheduling and inventory management by promoting information sharing and business collaboration, thereby enhancing overall operational efficiency (Flynn, Huo, & Zhao, 2010; Chen & Wang, 2014; Ak & Patatoukas, 2016).

This paper posits that excessive supply chain concentration is a manifestation of a company’s external resource dependency. It can lead to power imbalances among partners and influence the decision-making and autonomous resource allocation of the dependent party (Zhao & Li, 2023). In the digital context, a company’s reliance on external resources diminishes, information flows more smoothly and the search costs for alternative suppliers and potential customers are significantly reduced. Thus, digital transformation helps reduce supply chain concentration while retaining the advantages of high supply chain concentration, such as enhanced bargaining power and reduced risk of supply chain disruptions. Diversified procurement fosters supplier competition, providing companies with higher-quality products and services and multiple sales channels help companies maintain robust business performance.

Based on the above analysis, the following research hypothesis is proposed:

Enhancing market recognition. Digital transformation enhances corporate operational transparency and information efficiency, alleviating information asymmetry with investors. This allows financial institutions to more accurately assess enterprise creditworthiness and business prospects using advanced algorithms and big data analysis (Zhang, Lin, & Zhu, 2021), thereby making it more likely to invest in enterprises with high credit ratings (Segal, Shaliastovich, & Yaron, 2015). Digital transformation enterprises convey positive developmental signals and enhance market expectations for their value growth by disclosing digital-related information in their annual reports. As part of digital transformation, digital marketing boosts brand visibility and investors often make investment decisions based on easily accessible information. This further enhances investor interest and willingness to invest in digital transformation enterprises (Zong, Li, & Dai, 2020). In general, higher capital market recognition implies that enterprises can more easily secure financial support, providing a strong impetus for their development and resulting in higher operational efficiency and better performance.

Based on the above analysis, the following research hypothesis is proposed:

Enterprise operational performance (denoted as performance). To comprehensively evaluate enterprise operational performance, this study adopts a multidimensional approach based on financial indicators, referencing the work of Li et al. (2021). Specifically, the selected indicators include: return on equity, representing financial profitability and earning capacity; capital accumulation rate, reflecting the ability for capital growth and development; debt to asset ratio, indicating the capacity to utilize external funds and manage debt repayment; and total asset turnover ratio, capturing sales capacity and asset utilization efficiency. These indicators and their descriptions are detailed in Table S2. The selection of this composite indicator system is based on the following considerations: (1) these financial ratios are among the most commonly used, widely accepted and data-accessible measures in corporate financial analysis, (2) they collectively cover four critical dimensions of performance—profitability, growth, solvency and operational efficiency—and have been validated in previous studies as representative indicators (Zhu, Wu, & Sun, 2018) and (3) the correlations among these indicators are relatively low, which helps to reduce information redundancy. To avoid the bias introduced by subjective weighting, the entropy weight method is employed to assign objective weights to each indicator. The resulting weighted composite score is used as a summary measure of a firm’s operational performance. For ease of interpretation and comparison, the final composite score is linearly transformed and multiplied by 100.

Digital transformation index (denoted as dt). Enterprise digital transformation is a complex and systematic endeavor. To comprehensively measure the level of digital transformation and objectively reflect the current state of digitalization among firms, this study adopts the Digital Transformation Index constructed by the China Securities Market and Accounting Research team in collaboration with East China Normal University, as part of the China Listed Firms Digital Transformation Research Database. The index evaluates digital transformation across multiple dimensions, including firm-level strategic leadership (34.72%), technology-driven initiatives (16.20%), organizational enabling (9.69%), digital outcomes (27.13%) and digital application (8.84%), as well as macro- and meso-level environmental support (3.42%). In addition to textual mining of annual reports, the index also incorporates firm-level input data on digital talent recruitment, digital capital investment and digital technology patents. These resource-intensive inputs help to mitigate concerns about reliance on strategic disclosures, as they reflect tangible commitments rather than purely rhetorical positioning.

1. Enterprise organizational management efficiency (manage). This variable reflects a company’s overall operational efficiency. It is measured as

This formula captures the efficiency of organizational management by considering how management expenses relate to total revenue.

1. Labor productivity (labor). Digital transformation has significantly enhanced labor productivity, particularly through automation, which reduces labor input in tasks like tracking and data management during production. Since this study focuses on manufacturing enterprises, it emphasizes the impact of digital transformation on productivity within production processes. Labor productivity is measured as

This measure accounts for production-specific productivity gains driven by digital transformation.

1. Innovation effectiveness (innovation). Innovation effectiveness is often gauged through metrics such as R&D investment, patent applications, granted patents, or the proportion of capitalized R&D investment to total R&D investment. To provide a more objective measure of real-time innovation performance while accounting for patent-granting periods and bureaucratic delays in authorization, this study employs:

This approach captures the innovation performance of companies effectively and in a timely manner.

1. Supply chain concentration (supply). Supply chain diversification reflects a company’s ability to manage and maintain stable procurement and sales channels. The breadth of the supply chain is calculated as

where, $Psuppliers$ represents the proportion of annual purchases from the top 5 suppliers and $Pcustomers$ represents the proportion of annual sales to the top 5 customers. This measure balances procurement and sales channel diversity, reflecting the company’s supply chain resilience.

1. Capital market recognition (recognition). A firm’s attractiveness to the capital market largely depends on its visibility and recognizability among external investors. Analyst coverage and media exposure can enhance information transparency and mitigate information asymmetry, thereby facilitating access to external financing. In this study, the sum of analyst coverage and media coverage is used as a proxy for a firm’s capital market recognition, capturing the extent to which it is identified and recognized in the capital market.

This measure captures the degree of external recognition and interest in the company, which influences its financing ability.

To examine the net effect of digital transformation on the operational performance of manufacturing firms, this study draws on relevant literature and incorporates several control variables that may influence operational performance into the model (Dai & Fang, 2024; Sui, Jiao, Wang, & Wang, 2024). These control variables are specified as follows: (1) Enterprise ownership (state). The types of enterprise ownership primarily include state-owned enterprises, privately owned enterprises and foreign-funded enterprises. In this study, a dummy variable is used to represent enterprise ownership, with state-owned enterprises coded as 1 and non-state-owned enterprises as 0. (2) Enterprise age (age). Enterprise age is defined as the period from the establishment of the enterprise to the time of the study, typically measured in years. This variable can be accurately calculated by referencing the enterprise’s date of registration. (3) Board size (board). Board size reflects the total number of board members, as reported in the enterprise’s annual report. (4) Independent director ratio (indep). This ratio indicates the proportion of independent directors to the total number of board members. (5) Top 10 shareholders’ holding ratio (top10). This ratio represents the combined shareholdings of the company’s top 10 shareholders in relation to the company’s total equity. (6) Dual leadership (dual). This variable indicates whether the positions of chairman and general manager are held by the same individual. The variable is assigned a value of 1 if the positions are combined and 0 otherwise. (7) Capital intensity (capital). Capital intensity is measured as the ratio of the enterprise’s total assets to its operating income. (8) Enterprise’s growth rate (growth). This is defined as the ratio of the difference between the current period’s main operating revenue and the previous period’s main operating revenue to the previous period’s main operating revenue, used to assess the enterprise’s growth capacity over a certain period. (9) Government subsidies (subsidy). This variable is calculated by comparing the total amount of government subsidies to the enterprise’s total operating income and assessing the impact of government support on the enterprise. (10) Income tax rate (tax). The income tax rate reflects the ratio of the enterprise’s actual income tax expenses to its pre-tax profit, serving as an important indicator of the enterprise’s tax burden. (11) Total assets (asset). This study uses a firm’s total assets as a proxy for firm size in order to capture its potential impact on operational performance.

In order to examine whether digital transformation of enterprises can promote the improvement of performance in China’s manufacturing industry, this paper constructs the following fixed-effect model:

where, $Controlc,i,t$ is the control variable vector, $μi$ represents the individual firm-specific fixed effect, $α0$⁠, $α1$ and $αc$ are the coefficients to be estimated, $σt$ denotes the time-specific fixed effect, and $εi,t$ represents the random disturbance term. Considering the possibility of backdoor listings and industry changes, industry fixed effects are included in the model. The study places particular emphasis on the coefficient $α1$⁠, which represents the impact of digital transformation on firm performance.

To dissect causal chains, this study investigates the indirect influence of the independent variable (dt) on the dependent variable (performance), specifically through the conduit of the mediator variable (M). The mediation testing model is presented as shown in equations (7)-(8). Additionally, this paper examines the interaction effects of $dt×$ M to explore potential moderating roles (Preacher, Rucker, & Hayes, 2007), as evidenced in equation (9).

where, $Mi,t$ represents the intermediary variables, which are empirically tested as innovation effectiveness (innovation), labor productivity (labor), organizational management efficiency (manage), supply chain concentration (supply) and capital market recognition (recognition). The coefficients of particular interest are denoted as $β1$⁠, $λ2$ and $τ3$⁠.

This paper selects manufacturing companies listed on the Shanghai and Shenzhen stock markets from 2011 to 2021 as the initial research sample. The data sources are as follows: corporate financial data mainly originates from the China Stock Market & Accounting Research (CASMAR) database. The Corporate Digital Transformation Index is sourced from the CASMAR database’s subrepository on corporate digital transformation. The Enterprise Digital Application Index is obtained from annual reports published by listed companies, utilizing text analysis to compile the frequency of relevant keywords. Corporate patent data are obtained from the National Intellectual Property Administration. Data related to a company’s human capital is sourced from the Wind database. In our initial research sample data, we undertook the following data preprocessing steps: First, we excluded the samples of companies classified as special treatment (ST), *ST, or particular transfer (PT). Second, we removed samples with missing values for relevant variables. Then, we excluded samples of companies that went public after the year 2020. Finally, we obtained panel data for 2,346 manufacturing companies, resulting in a total of 19,733 sample observations. To mitigate the impact of outliers on the results, this paper applied winsorization to all continuous variables at the upper and lower 1% levels. Some variables, such as firm age and board size, are log-transformed to reduce right-skewness in their distributions. Table 1 reports the descriptive statistics of the variables used in this study.

Table 2 presents the regression results of the baseline estimations. In column (1), only the digital transformation index (dt) is included as the explanatory variable. Building on this, columns (2) through (5) progressively add more firm-level control variables to mitigate potential endogeneity arising from omitted variable bias. In all models, year, firm and industry fixed effects are included to control for unobserved heterogeneity at the temporal, individual and industry levels. The results from columns (1) to (5) consistently show that the coefficient of the digital transformation index (dt) is significantly positive at least at the 5% confidence level. Specifically, in column (5), after accounting for a comprehensive set of control variables, a one-unit increase in the digital transformation index is associated with a 0.995 increase in firm performance. This provides strong evidence that digital transformation has a positive impact on the performance of manufacturing firms, supporting research hypothesis H1. These findings are consistent with the conclusions of Barba-Sánchez, Meseguer-Martínez, Gouveia-Rodrigues, and Raposo (2024), who argue that digital transformation and its antecedents can enhance firms’ IT capabilities, thereby promoting performance improvement. This underscores the critical role of digital transformation in driving better operational outcomes.

To address potential endogeneity, this study follows the approach of Zhao et al. (2021) by using “per capita telecommunications traffic (telecom)” as an instrumental variable (IV) and applying the two-stage least squares (2SLS) estimation method. Improvements in telecommunication infrastructure serve as an important external driver of firms’ digital transformation, implying a potential correlation between the two. Per capita telecom data traffic can be regarded as an exogenous factor because it is mainly influenced by external infrastructure development and macro-level policies, without directly affecting firms’ operational outcomes. We argue that telecom mainly influences firm performance by enhancing macro-level communication conditions and promoting digital transformation, while its potential effects through alternative channels remain relatively limited. Our model controls for firm, industry and year fixed effects, along with multiple firm-level control variables, which helps mitigate potential confounding effects from other transmission channels (e.g. transaction costs, market access, etc.). Since the IV is constructed from macro-level data while firms are observed at the micro level, spatial autocorrelation concerns are substantially reduced. The inclusion of multiple fixed effects further absorbs and controls for common macro-level shocks, thereby partially alleviating the potential spillover effects of telecommunications infrastructure on neighboring firms.

Table 3 presents the estimation results for both the first and second stages of the IV approach. As shown in column (1), the first-stage results indicate that improvements in macrotelecommunication infrastructure significantly promote firms’ digital transformation. The validity of the instrument was confirmed by a series of tests. The p-value of the Kleibergen-Paap (KP) rk Lagrange Multiplier (LM) statistic is below 0.01, rejecting the null hypothesis of under-identification at the 1% significance level and thereby confirming the relevance of the instrument. The Kleibergen-Paap rk Wald F statistic exceeds the 10% critical threshold proposed by Stock and Yogo (2002), indicating that the instrument passes the weak identification test. In column (2), the coefficient of digital transformation (dt) is positive and statistically significant at the 1% level. These findings suggest that even after addressing endogeneity, digital transformation in manufacturing has a significant positive impact on firm performance, further supporting hypothesis H1.

In this study, the IV estimate is greater than the fixed-effects (FE) estimate obtained from the baseline regression, which can be explained as follows. The IV approach focuses on “marginal firms” that are significantly influenced by the IV, i.e. firms whose digital transformation is primarily driven by improvements in telecommunications infrastructure rather than internal capabilities. Compared to highly digitalized firms, these marginal firms are often in the early stages of digital transformation, where the benefits are more pronounced. For example, initial improvements such as basic internet access or the implementation of preliminary digital processes can lead to substantial productivity gains. Consequently, IV estimates predominantly capture the “from-zero-to-one” transformation effects for these marginal firms, while FE estimates reflect an average effect across all firms, including those that are already highly digitalized and experience diminishing marginal returns. As a result, the IV estimate is larger than the FE estimate.

To enhance the persuasiveness of our research findings, this study conducted several robustness tests.

The essence of corporate digital transformation is a form of organizational strategic management behavior that encompasses multiple levels from strategic planning to technological implementation. To measure and evaluate digital transformation more precisely, it can be broken down into different indicators. In our research, we used the digital transformation index’s digital strategy leadership index and digital application index to replace the digital transformation index in our regression analysis. The regression results are presented in columns (1) through (5) of Table 4. In column (1), the regression coefficient for digital strategy leadership (strategy) is 0.144, but not significant. In column (2), the regression is performed on the lagged digital strategy leadership (l.strategy), resulting in a coefficient of 0.375, significant at the 5% confidence level. Compared to the results in column (1), the effect is more pronounced and its significance is higher, indicating that the positive impact of digital strategy leadership on a company’s operational performance is persistent and exhibits a certain degree of timeliness. In column (3), the regression is conducted using the company’s digital application level (application) as a replacement for the digital transformation index. The digital application level indicator is derived from the frequency of appearance of relevant keywords related to digital applications in the company’s annual reports, excluding the “management analysis and discussion” section. Considering the inherent randomness in annual report disclosures and their potential bias in measuring the degree of digital application, the maximum continuity operation is applied to the measurement of the digital application level over time. Specifically, when a company’s digital application level index decreases in a particular year during the sample period, the maximum value for that company in any subsequent years within the sample period is used to replace the values for those years, resulting in a more realistic digital application level index (appmax). In column (4), appmax is used as a replacement for application in the regression. The results in columns (3) and (4) demonstrate that the level of digital application significantly enhances a company’s operational performance.

Considering that a company’s operational performance is a composite index derived from various financial indicators and that different calculation methods can yield different results, we reevaluated a company’s operational performance using Principal Component Analysis (PCA) and conducted a replacement test for operational performance measured using the previous Entropy Method. Column (5) in Table 4 presents the results of the reevaluation of corporate performance using the PCA method, denoted as performpca. The coefficient of dt is 0.180, which is significant at the 5% level. This indicates that even after changing the method for calculating corporate performance, the conclusions of this study remain valid.

As mentioned earlier, the digital transformation index includes multiple keyword frequency indicators derived from text analysis of annual reports disclosed by companies. While the index constructed through machine learning-based text analysis provides a more accurate reflection of a company’s actual digitalization status, in the era of digitization, companies often strategically disclose their internal digitalization situation to access external resources. Research has shown that some companies may have suspicions of opportunistic and strategic hype in their disclosure of “Internet Plus” information (Zhao, Chen, & Cao, 2020). Additionally, certain industries have natural and close connections with digital businesses, and the frequency of disclosing related keywords in annual reports is higher. To eliminate this strategic behavior of companies and the inherent attributes of industries, the following measures were taken in the study: (1) Exclusion of companies listed on the Sci-Tech Innovation Board and the Growth Enterprise Market (GEM). Companies on these boards are mostly high-tech enterprises or closely related to the Internet business model, and they are more inclined to use digital transformation-related keywords in their annual reports. (2) Exclusion of companies listed after 2011. Companies that went public after 2011 have experienced rapid advancements in information technology since their initial listing, leading to a greater inclination to incorporate digital technologies into their day-to-day operations. (3) Exclusion of companies in the computer, communications and other electronic equipment manufacturing industry (industry code 39).

Table 5 reports the results of the regression analysis after excluding the effects of strategic business behavior and industry attributes. In column (1), the sample of companies listed on the Sci-Tech Innovation Board and the GEM was removed. The results show that the regression coefficient of the digital transformation index remains significantly positive at the 1% confidence level. In column (2), the sample of companies listed after 2011 was further removed based on column (1). The results show that the regression coefficient of the digital transformation index remains significantly positive at the 1% confidence level. In column (3), the sample of companies in the computer, communications and other electronic equipment manufacturing industry (industry code 39) was further removed based on column (2). The regression results indicate that the digital transformation index continues to significantly enhance a company’s operational performance at the 5% confidence level. This suggests that after eliminating the effects of strategic business behavior and industry attributes, the conclusions of this study remain robust and reliable, confirming H1.

In real-world situations, companies with better business performance often have more discretionary funds for information technology construction and digital transformation. In other words, there may be a phenomenon of reverse causality. To address this question, we extend the observation period for the impact of digital transformation on the performance of manufacturing firms. Specifically, the digital transformation variable (dt) is lagged by one period and two periods, and the regression model is re-estimated. The results are presented in Table 6. The regression results show that, regardless of whether the explanatory variable (dt) is lagged by one period or two periods, the degree of digital transformation remains significantly positive at the 1% level in influencing the performance of manufacturing firms. This indicates that digital transformation has a cumulative and reinforcing effect on the performance of manufacturing firms. These findings further support the robustness of our results.

The previous baseline regression results have already revealed the positive impact of digital transformation on the performance of manufacturing enterprises. To further explore how digital transformation affects corporate performance, this paper examines research on five specific impact channels from three dimensions: technology, organization and external networks.

Table 7 presents the results of the mechanism test for enhancing innovation effectiveness. Column (1) shows that enterprise digital transformation significantly increases innovation output at the 1% confidence level. In this specification, the model controls for firms’ R&D expenditure (rdspend), including wages paid to R&D personnel. In column (2), the coefficient of innovation performance (innovation) is −0.037, significant at the 5% confidence level. This suggests that innovation performance exerts a crowding-out effect on current operational performance, as innovation activities may temporarily divert resources from other operational functions, negatively impacting short-term performance.

To investigate whether enterprise digital transformation enhances a firm’s ability to convert innovation outcomes into operational performance and mitigates the negative effects of innovation activities, column (3) introduces an interaction term (dt×innovation). The interaction term’s coefficient is significantly positive at the 5% confidence level, indicating that as the level of digital transformation increases, the crowding-out effect of innovation on current performance diminishes. At higher levels of digital transformation, firms can significantly shorten their R&D innovation cycles, allowing innovation outcomes to positively contribute to current performance. This demonstrates that digital transformation facilitates the effective conversion of innovation outcomes into operational benefits.

Overall, the results in columns (1) through (3) show that enterprise digital transformation not only enhances innovation performance but also improves the ability to translate innovation outcomes into operational performance, thereby positively influencing overall firm performance. These findings confirm that enterprise digital transformation strengthens both innovation effectiveness and its operational conversion, providing strong support for H2.

Table 8 reports the results of the impact mechanism test for improving labor productivity. In column (1), labor productivity is regressed on the digital transformation index. The results indicate that digital transformation significantly enhances labor productivity at the 1% confidence level. The results in column (2) show that the regression coefficient of labor productivity is 1.188 and is statistically significant at the 1% confidence level. This indicates that, statistically, labor productivity significantly improves firm performance, which is consistent with theoretical expectations and the overall conclusion.

As discussed in the preceding theoretical analysis, digital transformation enhances firm performance by improving labor productivity. Therefore, the positive impact of digital transformation on firm performance is expected to be more pronounced for firms with lower labor productivity than for those with higher labor productivity. To further examine this mechanism, Column (3) includes both labor productivity and the interaction term between labor productivity and the digital transformation index (dt✕labor) in the regression. The results show that the coefficient of the interaction term is −0.494 and is statistically significant at the 10% confidence level. This indicates that the positive effect of digital transformation on firm performance is more significant among firms with lower labor productivity, providing further evidence that digital transformation improves firm performance through enhanced labor productivity. Therefore, hypothesis H3 is supported.

Table 9 reports the results of the examination of the impact mechanism of digital transformation on enhancing organizational management efficiency. In column (1), the coefficient of dt is significantly positive at a 5% confidence level, indicating that digital transformation enhances organizational management efficiency (manage). In column (2), the coefficient of manage is 7.015, with statistical significance at a 1% level, suggesting a significant positive correlation between organizational management efficiency and enterprise performance. Generally, under unchanged conditions, higher organizational management efficiency is associated with better enterprise performance.

If the digital transformation of manufacturing firms improves firm performance by enhancing organizational management efficiency, it is expected that the impact of digital transformation on firm performance would be more pronounced and significant for firms with lower management efficiency compared to those with higher management efficiency. To test this hypothesis, column (3) includes both management efficiency and the interaction term (dt✕manage) in the regression. The results show that the coefficient of the interaction term is significantly negative at the 5% confidence level. This suggests that, holding other factors constant, the same level of digital transformation has a more substantial positive effect on the operational performance of firms with lower management efficiency. This finding further confirms that digital transformation enhances firm performance by improving organizational management efficiency, thereby supporting hypothesis H4.

Table 10 reports the empirical results of the impact mechanism testing of digital transformation on reducing supply chain concentration (lnsupply). As shown in column (1), enterprise digital transformation significantly reduces the level of supply chain concentration. In column (2), the coefficient of lnsupply, at a 1% confidence level, is significantly negative. This implies that supply chain concentration has a negative impact on enterprise performance to a certain extent. If digital transformation strategies aimed at reducing supply chain concentration lead to improved enterprise performance, we can anticipate that these strategies will be more effective in enterprises with more concentrated supply chains. To test this, To test this, this study incorporates the interaction term (dt✕lnsupply) between supply chain concentration and digital transformation into the model, as shown in column (3). The results show that the coefficient of the interaction term is 2.473 and is statistically significant at the 1% confidence level. This indicates that digital transformation significantly enhances performance in enterprises with high supply chain concentration, confirming our initial expectations. Thus, it can be concluded that enterprise digital transformation positively affects enterprise performance by reducing supply chain concentration, thereby confirming research H5.

Wu, Hu, Lin, and Ren (2021) found that digital transformation significantly increases stock liquidity. In the capital market, due to the presence of the hot pursuit effect, analysts often tend to focus on targets with strong liquidity and the trading activity of stocks also positively attracts media coverage. Therefore, stock liquidity positively impacts a company’s capital market recognition. On the other hand, stock liquidity is also influenced by various uncontrollable factors, such as market sentiment. To confirm the direct impact of corporate digital transformation on capital market recognition, the study controlled for stock turnover rate (turnover).

As shown in column (1) of Table 11, the coefficient of digital transformation (dt) is significantly positive at the 1% level, indicating that digital transformation effectively enhances firms’ visibility in capital markets. The result in column (2) shows that improved capital market recognition (recognition) has a positive impact on firm performance, with significance at the 5% level. In column (3), the coefficient of the interaction term (dt✕recognition) is negative and statistically significant at the 1% level. This suggests that digital transformation has a stronger effect on improving the performance of firms with lower market recognition by facilitating more effective access to external financial resources. These empirical findings are consistent with expectations and provide support for hypothesis H6.

To gain a more comprehensive and in-depth understanding of the mechanisms, this study further conducts an empirical analysis of the dynamic marginal effects of digital transformation over time in Chinese manufacturing enterprises, as well as the external conditions that influence its effectiveness.

Figure 2 depicts the dynamic marginal effects of the digital transformation index on enterprise performance over time. It is evident that the marginal impact of digital transformation on enterprise performance shifted from negative to positive around the year 2015, displaying an increasing trend thereafter. This is an intriguing finding as it coexists with two phenomena during the sample period, namely the “IT Productivity Paradox” and the “IT Dividend.” This study suggests that the reason for this phenomenon lies in the fact that digital transformation is fundamentally a long-term investment project dependent on external connections. In the early stages, digital transformation was primarily focused on information management. However, due to the immaturity of digital technologies, limited data conversion capabilities and inadequate digital infrastructure, data silos emerged among enterprises, hindering value transmission and sharing. Additionally, demand-side e-commerce platforms have not yet become mainstream, limiting the potential for value creation through digital marketing. Consequently, early investments in digital infrastructure resulted in low returns and in some cases, enterprises fell into an “IT investment trap.”

This unfavorable situation was transformed through the critical role played by the government in digital infrastructure development. Initiatives such as the “Broadband China” strategy in 2013, the “Informatization and Industrialization” integration standard pilot in 2014, the introduction of the “Internet Plus” action plan and the release of “Made in China (2025)” in 2015, all exemplify the government’s commitment to fostering the rapid development of digital infrastructure. These policies and measures not only accelerated the construction of digital infrastructure but also facilitated inter-enterprise connectivity and data flow, providing strong support for external connections and data sharing among enterprises. This concerted effort enabled China to overcome the “IT productivity paradox” and enter the “IT dividend” phase, thereby propelling the high-quality development of the manufacturing sector.

The analysis presented earlier indicates that the effectiveness of enterprise digital transformation is closely related to the development of digital infrastructure. Therefore, this paper examines the impact of external digital infrastructure maturity on corporate transformation outcomes at both provincial and municipal levels. At the provincial level, the length of regional optical cable lines (opticalcable) was selected as a proxy indicator for the digital infrastructure index, while at the municipal level, the penetration rate of citywide internet broadband (broadband, calculated as the proportion of broadband internet access users in each city’s year-end registered population) was chosen as a proxy indicator for the digital infrastructure index. We include opticalcable (or broadband) and the interaction term dt×opticalcable (or dt×broadband) in the regression model for testing.

Columns (1) and (2) in Table 12 indicate that the coefficients of the interaction terms dt×opticalcable and dt×broadband are both significantly positive at the 1% confidence level. In other words, under consistent conditions, the enhancing effect of digital transformation on enterprise operational performance increases as digital infrastructure gradually improves. Figure 3 illustrates the relationship between digital transformation and regional optical cable network length, as well as urban internet broadband penetration rate. As the length of regional optical cable lines and the penetration rate of citywide internet broadband increase, the impact coefficient of digital transformation on the operational performance of manufacturing enterprises gradually becomes larger. In other words, the level of digital infrastructure positively moderates the positive effect of digital transformation on enterprise operational performance, serving as an external condition for the effectiveness of digital transformation.

The digital transformation of enterprises is a long-term investment project that requires sustained financial support. Therefore, the level of financial services and funding support in the region where an enterprise is located is one of the factors that influence and constrain the effectiveness of digital transformation. In the practice of digital transformation, enterprises are often limited by insufficient capital investment, which may result in incomplete digital transformation and lead to management chaos. Therefore, this study posits that a higher level of financial development allows digital transformation to have a more significant impact on improving business performance.

This study measures the level of financial development at both the provincial and municipal levels. At the provincial level, the ratio of annual deposits and loans to regional Gross Domestic Product (GDP) (referred to as finance) is employed as a proxy indicator for the level of financial development. At the municipal level, drawing from the approach of Li, Yan, Song, and Yang (2020b), a financial technology index (fintech) is constructed as a proxy indicator for the level of financial development based on a Baidu News advanced search using the keyword “Financial Technology.” We include the level of financial development, finance (or fintech), as well as the interaction term dt×finance (or dt×fintech) in the regression model.

Columns (3) and (4) in Table 12 clearly indicate that the interaction terms, dt×finance and dt×fintech, are both significantly positive at a 1% confidence level. This suggests that the level of financial development positively moderates the enhancement of enterprise business performance by digital transformation, serving as an external condition for the effectiveness of digital transformation. As depicted in Figure 4, with the increase in the level of financial development, the marginal effect of digital transformation on enterprise performance gradually becomes more significant.

The competitive environment is a crucial factor influencing corporate digital transformation strategies (Zeng, Cai, & Ouyang, 2021). This study adopts the Structure-Conduct-Performance (SCP) research paradigm to examine how industry competition intensity impacts corporate digital transformation and its economic consequences. Specifically, we investigate whether the economic effects of digital transformation differ across industries with varying levels of competition intensity. To quantify industry competition intensity, we use the Herfindahl-Hirschman Index (HHI), where a smaller HHI indicates a more dispersed and competitive industry.

The results in column (5) of Table 12 show that the coefficient of the interaction term dt×HHI is significantly negative at a 1% confidence level. This suggests that, under consistent conditions, as industry competition intensity increases, the positive impact of digital transformation on business performance becomes more pronounced.

We analyze the mechanisms involved. First, competition drives companies to gain cost advantages and customer resource advantages through digital transformation, thereby improving their operational performance. The “learning effect” within industries (Wang, Zhang, & Yan, 2022), low-cost replication of data elements and the widespread adoption of online transaction models all play roles in industries with varying levels of competition intensity. Second, the impact of digital transformation exhibits significant differences across industries with different competition intensities. In dispersed industries, digital transformation fosters interconnectedness and data sharing, reducing costs and improving operational performance. Highly concentrated industries, companies typically possess significant market power and digital marketing does not necessarily open up new channels for value creation. Similarly, Figure 5 illustrates the relationship between the marginal effects of digital transformation on a company’s operational performance and the HHI index.

This study, based on the “Technology-Organization-External Network” framework, investigates how digital transformation affects the operational performance of manufacturing enterprises in China. It further examines the dynamic marginal effects and external conditions influencing this relationship. The empirical results reveal that digital transformation has a significant positive impact on the operational performance of Chinese manufacturing firms. This impact is primarily reflected in five key aspects: enhancing innovation efficiency, improving labor productivity, increasing organizational management efficiency, reducing supply chain concentration and strengthening market recognition. However, the effect of digital transformation on firm performance shows a transition from negative to positive over time. This indicates that digital transformation is a long-term investment that requires sustained efforts to endure the “growing pains” of transformation before reaping the “digital dividends.” Moreover, external factors such as digital infrastructure, financial support and industry competition are critical for the effectiveness of digital transformation in improving firm performance. The more developed the digital infrastructure, the higher the level of financial development and the more intense the industry competition, the stronger the positive impact of digital transformation on firm performance.

Drawing on the conclusions of the aforementioned research, we provide the following recommendations: First, it is imperative to formulate a long-term digital transformation strategy and ensure its implementation mechanism. Corporations must adopt a long-term perspective on digital transformation, crafting strategic plans accordingly and guaranteeing sustained financial and technological investment. To achieve this end, a comprehensive monitoring and evaluation framework should be established to periodically review the transformation process and outcomes, allowing for timely strategic adjustments. Concurrently, emphasis should be placed on the recruitment and cultivation of digital talent to ensure an adequate human resource pool that can provide intellectual support for the ongoing digital transformation efforts.

Second, efforts should be made to enhance the overall competitiveness of enterprises and the resilience of supply chains. To boost labor productivity, companies should adopt automation and intelligent tools, coupled with employee training, to adapt to the changes in work patterns brought about by new technologies. Building on this foundation, it is essential to strengthen innovation efficacy by increasing R&D investment and leveraging digital means to accelerate the innovation process, thereby maintaining a competitive edge in the market. Moreover, organizational management should be optimized to improve decision-making efficiency and digital approaches should be employed to construct a diversified supply chain system, reducing concentration and enhancing flexibility and efficiency, while also increasing the enterprise’s attractiveness to external investors.

Third, there is a need to reinforce the development of digital infrastructure and the establishment of a robust financial support system. Governments and businesses should collaborate to increase investment in digital infrastructure to meet the foundational needs of digital transformation. Financial institutions have a role to play in providing tailored financial services to assist enterprises in overcoming capital challenges. Additionally, companies should undertake digital transformation initiatives that are aligned with the characteristics of their respective industries, aiming to stand out in the fierce market competition and achieve sustainable development.

This study has several limitations. From the theoretical perspective, the five identified impact pathways were treated equally, although their relative importance may differ. Future research could further explore the relative significance of these pathways. Regarding the empirical strategy, the IV used in this study may not fully satisfy the exclusion restriction, as it might not completely isolate the direct effects of telecommunication infrastructure on firm performance beyond digital transformation. Spatial spillover effects could challenge the exogeneity assumption of the instrument, thereby affecting the accuracy of causal inference. Future studies may consider employing multiple instrumental variables or spatial econometric models to further examine the causal effects.

The article is sponsored by the National Natural Science Foundation of China (Grant No. 72573030) and the Scientific Research Platforms Foundation of Dongbei University of Finance and Economics in China (Grant No. PT-Y202225). All authors contribute to this paper equally. The authors acknowledge the useful comments from the Editor and anonymous reviewers. Certainly, all remaining errors are our own.

The supplementary material for this article can be found online.