The Impact of Digital–Real Integration on Firm Green Technology Innovation: Evidence from Chinese A-Share Listed Companies

Abstract

Based on Chinese A-share listed companies’ panel data from 2008 to 2023, we examine the impact of digital–real integration on firm green technology innovation. The empirical results show that digital–real integration can significantly improve firm green technology innovation. Mechanism analysis reveals that digital–real integration improves firm green technology innovation by reducing financing constraints, improving productivity and reducing information asymmetry. Heterogeneity analysis results show that the positive effects are more pronounced in private and other non-state-owned firms, large-scale firms and firms with stronger internal control. The study also confirms that government green subsidy, governance quality, intellectual property protection, digital infrastructure and marketization level can enhance the positive effect of digital–real integration on firm green technology innovation.

1. Introduction

Since the reform and opening up, China’s economic and social development has progressed at an extraordinary pace, with its GDP rising to the second largest in the world. China has thus entered the ranks of “middle powers,” with the manufacturing industry playing a pivotal role in this transformation. However, behind the rapid advancement of industrialization and urbanization lies a development model characterized by high energy consumption and high pollution, which imposes substantial environmental costs and leads to resource depletion and environmental pollution (Feng et al., 2024) [1].This highlights the urgency of green governance. Green technological innovation has “double externalities” (Rennings, 2000) [2] and has a significant demonstration effect on green development (Wang et al., 2025) [3]. Advancing green technological innovation is not only an optimization and upgrade of traditional innovation models but also an essential path for driving global economic transformation and sustainable development (Ge et al., 2024; Liu et al., 2024) [4,5]. Green technology innovation not only constitutes a key measure to implement green transformation but also prompts firms to strike a balance between economic development and ecological benefits. For firms, green innovation entails higher capital investment, involves greater risks, and features longer profit cycles compared to general technological innovation. Due to these inherent traits, firms often lack the necessary drive to voluntarily engage in green innovation activities. Consequently, a central question for both scholars and regulatory bodies is the design of effective incentives to spur green innovation within the private sector.

With the breakthroughs in digital technologies, the emerging business models and innovative forms driven by the widespread adoption of digital technologies are shaping the trajectory of technological and industrial revolutions and have become a pivotal driver for China’s economic transformation (Janowski, 2015) [6]. Notably, China’s transition into a new phase of economic development hinges, to a critical extent, on fostering high-quality growth within the real economy. This process is fundamentally underpinned by the profound integration of the digital and real economies. Whether the digital economy and the real economy (abbreviated as digital–real integration) can be deeply integrated has become an intrinsic requirement for constructing the modern industrial system. In other words, traditional research separately investigating the digital economy and the real economy no longer aligns with the reality of China’s current economic development. The extent of digital–real integration determines the capacity of the digital economy to serve the real economy. At present, despite China’s significant advancements in digital technology domains such as artificial intelligence, big data and the Internet of Things, there remain issues of “inability to integrate,” “incomplete integration,” “shallow integration,” and “reluctance to integrate” in digital–real integration (Hong and Ren, 2023) [7].

Promoting digital–real integration not only helps release digital dividends and advance green technology innovation in the manufacturing industry but also drives high-quality economic development. Digital–real integration is a systematic process of deep integration at multiple levels, driven by both digital technologies and data elements in enterprise organization, industrial ecology and social operation. Currently, some studies have found positive effects of digital–real integration on firms’ low-carbon transformation (Xu et al., 2025; Yang et al., 2024) [8,9], but these studies have not yet revealed whether the low-carbon transformation is driven by fundamental green technology innovation or by short-term coping behaviors adopted by firms. This determines whether the integration of the digital and real economies can fundamentally and sustainably promote China’s green transformation.

Therefore, this study contributes to analysing the influence mechanism and impact of digital–real integration on firm green technology innovation, which can provide new ideas for promoting the intelligent and green transformation of firms. The findings contribute to the theoretical foundation for policy formulation aimed at enhancing firms’ green innovation capabilities in the digital economy era, thereby accelerating the realization of the country’s “dual carbon” strategic objectives. Based on a dataset of Chinese A-share listed companies from 2008 to 2023, we empirically investigate both the overall and heterogeneous effects of digital–real integration on firm green technology innovation. Building upon this foundation, this study conducts robustness tests, heterogeneity analyses and mechanism analyses to derive further valuable conclusions. Furthermore, we explore the moderating roles of the “well-functional government” and “effective market”, which can provide theoretical reference for local governments in formulating policies for intelligent and green transformation.

2. Literature Review

2.1. Research on Digital–Real Integration

With the rapid growth of the digital economy, scholars have increasingly recognized the importance of the digital economy in serving the development of the real economy. Consequently, research on “the digital–real integration” has gradually emerged and evolved. Regarding its conceptual connotation, scholars have expounded on the connotation of digital–real integration through various theoretical frameworks, including the systematic “technology–industry–enterprise–ecology” framework (Hong and Ren, 2023) [7], the “technology–economic paradigm” theory and the “social reproduction theory” (Ding et al., 2024) [10]. Overall, most studies consistently argue that digital–real integration refers to the deep digital–real integration (Liu et al., 2022) [11]. In terms of indicator measurement, scholars have employed input–output analysis (Zhu et al., 2025) [12], the entropy weight-coupling synergy model (Sun et al., 2024; Zhao et al., 2025) and the patent citation method (Huang and Gao, 2023) to depict the integration of the digital economy and the real economy at the regional [12,13,14], industrial and enterprise levels. In terms of impact effects, some scholars have found that the integration of digital and real economies exerts a positive driving effect on carbon emission intensity (Xu et al., 2025) [8], green total factor productivity (Guo et al., 2024) [15], the green transformation of resource-based cities (Meng et al., 2023) [16], and green emission reduction efficiency (Pang et al., 2025) [17]. However, researchers have primarily directed their efforts toward meso- and macro-level analyses of the environmental benefits associated with digital–real integration, but they neglect the micro-level mechanisms. Consequently, existing research remains inconclusive on whether digital–real integration can generate a sustained and fundamental impact on green transformation.

2.2. Research on the Impact of the Digital Economy on Firm Green Technology Innovation

The literature closely related to this article is the research on the relationship between the digital economy and enterprise green technology innovation. Some studies have confirmed that the digital and intelligent transformation of firms can promote enterprise green technology innovation (Zhang and Song, 2025; Zhou et al., 2025) [18,19], with reducing financing costs and enhancing collaborative research as the main intermediary channels. Furthermore, a number of studies have investigated the paths to promote green technology innovation regarding the application of digital technology. Some scholars believe that the application of digital technology can enhance the efficiency of green technology innovation through environmental regulations (Wang et al., 2022) and green human resource allocation (Liu et al., 2024) [20,21]. Furthermore, some scholars have introduced the spatial spillover perspective to confirm that the digital economy has a significant positive impact on green technology innovation (Li et al., 2023) [22]. Notably, some scholars have put forward the opposite argument, believing that the digital economy may have resource extraction and technological constraint effects that are detrimental to green technology innovation in the early stage (Wu et al., 2024) [23]. From the above research, we find that the vast majority of studies focus on the impact of the digital economy on corporate green technology innovation, lacking research that explores the mechanisms and effects of enhancing corporate green technology innovation from the perspective of digital–real integration.

In summary, there are still several critical deficiencies in existing research. First, most studies primarily investigate the environmental effects of digital–real integration from meso- and macro-level perspectives, with limited in-depth exploration of the micro-level mechanisms and specific effects within firms. Second, regarding the pathways for enhancing enterprise green technology innovation, a vast literature examines only the isolated effects of data elements, digital technologies, or the digital economy, neglecting the practical requirement for the digital economy to serve the real economy. As digital–real integration accelerates, treating them as separate and independent systems has become increasingly unrealistic and lacks empirical grounding, thereby overlooking the multiple synergistic effects generated through their integration. Therefore, it is imperative to develop a theoretical framework that examines enterprise green technology innovation from the integrated perspective of digital–real integration.

This study makes three key marginal contributions: (1) This paper systematically investigates the mechanisms and effects of digital–real integration on enterprise green technology innovation, which can indicate whether digital–real integration can fundamentally drive the sustainable transformation of firms toward green development. (2) This study finds that digital–real integration significantly improves firm green technology innovation by alleviating financing constraints, improving production efficiency and reducing information asymmetry. This finding provides a meaningful extension to the existing literature. (3) We suppose that market mechanisms and government regulation serve as critical external environments that shape the development of the digital economy and foster enterprise innovation. These factors inevitably influence the effectiveness of digital technologies and the progress of green R&D. Accordingly, this paper incorporates the concepts of “active government” and “efficient market” into its analytical framework.

3. Theoretical Framework

3.1. The Direct Effect of Digital–Real Integration on Firm Green Technology Innovation

Deep digital–real integration refers to the integration of digital technology into the real economy across a broader scope and wider range. This process can enhance the promotion of firm green technology innovation. Firstly, firms can precisely analyze the current bottlenecks in their green technology through real-time monitoring of production energy consumption data, which can guide firm green technology innovation through the application of digital technology. At the same time, firms can efficiently obtain a vast amount of information and resources related to green technology research and development through digital technology. This can reduce the cost of obtaining external information and help firms accumulate experience in green technology innovation (Zhao and Qian, 2024) [24]. Secondly, promoting digital–real integration can enhance the efficiency of collaborative innovation in green technology among firms, universities and research institutions. The deep integration of digital technology and real firms can break down information barriers among R&D entities, expand the speed and scope of the diffusion of information, and thereby enhance the efficiency of cross-departmental collaboration in clean-tech R&D. Thirdly, firms can achieve dynamic monitoring and analysis of customers, competitors, and product competitiveness through big data technology, which can provide reference information for managers’ production decisions and lay the foundation for enhancing green technology innovation capabilities (Gao et al., 2024) [25]. In conclusion, digital–real integration helps improve the level of firm green technology innovation. Based on the above analysis, this paper proposes Hypothesis 1:

H1:

The integration of the real economy and the digital economy can enhance firm green technology innovation.

3.2. The Indirect Effect of Digital–Real Integration on Firm Green Technology Innovation

3.2.1. The Mechanism of Alleviating Financing Constraints

Comprehensive information disclosure combined with a high degree of digitalization enhances firms’ credibility among financial institutions and investors, thereby expanding their access to specialized financing channels such as green credit and sustainable investment (Feng et al., 2022) [26]. Meanwhile, the application of digital technologies enables firms to more effectively identify and utilize financing opportunities, thus strengthening their financing capacity and generating a self-reinforcing cycle of resource allocation and innovation. Firstly, firms can improve the transparency of their environmental performance information by leveraging digital technologies. These technologies enable financial institutions and investors to more accurately assess the value and risks associated with firm green R&D projects, as well as to conduct a comprehensive evaluation of their financial status, operational conditions and environmental performance (Du and Wang, 2024; Tang et al., 2023) [27,28]. Comprehensive information disclosure and a high degree of digitalization contribute to building trust between firms and financial institutions or investors, thereby expanding access to specialized financing channels such as green credit and sustainable investment. Meanwhile, the adoption of digital technologies enables firms to more effectively identify and utilize financing opportunities, thus strengthening their financing capacity and fostering a self-reinforcing virtuous cycle. Secondly, digital–real integration contributes to the optimization of firm resource management processes, thereby minimizing capital waste (Xue et al., 2022) [29]. For example, firms can utilize real-time monitoring of energy consumption data during production to dynamically adjust the direction and investment in green technology R&D, which can effectively mitigate the capital waste arising from deviations in R&D strategies and consequently enhance the input–output efficiency of green technology innovation. Thirdly, for firms constructing carbon asset trading platforms based on blockchain technology, the adoption of blockchain enhances the transparency of carbon asset transactions and reduces transaction costs, thereby strengthening market confidence in the quality of firm carbon assets and improving their capacity for multi-party financing through these assets. This demonstrates that digital–real integration can effectively alleviate financing constraints faced by firms during the process of green technology innovation, thereby providing more substantial financial support. Based on the foregoing analysis, the following hypothesis is proposed:

H2:

Digital–real integration exerts a positive effect on firm green technology innovation through the mediating role of financing constraints.

3.2.2. The Mechanism of Enhancing Firm Productivity

Digital–real integration can promote firm green technology innovation by enhancing their total factor productivity. Firstly, digital technologies significantly improve the efficiency of resource allocation and information transmission in real firms, which enables them to acquire and integrate diverse resources more rapidly and thereby boost overall production efficiency (Wang, 2023; Luo et al., 2025) [30,31]. Secondly, the infusion of digital technologies into traditional manufacturing facilitates the acceleration of firm transformation toward automation and intelligence, which can enhance operational efficiency across key stages, including research and development, design, production assembly and distribution (Li et al., 2020) [32]. This integration reduces total production costs and consequently releases greater financial resources for green technology R&D. Thirdly, digital–real integration can drive the transformation and upgrading of labor skills, generating more high-skilled positions that require interdisciplinary knowledge and core technological R&D capabilities (Acemoglu, 2020) [33]. This integration necessitates increased investment in talent training by firms to enhance employees’ digital literacy and technical proficiency. However, digital talents must continuously update their knowledge systems to keep pace with the evolution of new processes and advanced equipment. These changes contribute to further improvements in enterprise production efficiency. Overall, by enhancing enterprise total factor productivity, digital–real integration facilitates the release of additional resources for green technology innovation, establishing a virtuous closed-loop mechanism of “efficiency improvement–resource release–innovation investment”, thereby effectively alleviating the resource constraints firms face in green technology R&D. Based on the above analysis, this paper proposes Research Hypothesis 3:

H3:

Digital–real integration exerts a positive effect on firm green technology innovation through the mediating role of total factor productivity.

3.2.3. The Mechanism of Information Asymmetry in Firms

Digital–real integration helps enhance the level of firm green technology innovation by reducing information asymmetry. On the one hand, integrating and absorbing the cutting-edge knowledge and resources in the field of green technology is the key factor for firms to promote green technology research and development (Jiang et al., 2025) [34]. By applying digital technologies, real firms quickly mine and process vast amounts of knowledge resources related to green technology, promptly grasp the latest green innovation achievements and environmental protection policy dynamics. This lowers the costs associated with information search and collaboration, while enhancing their capacity to monitor the trajectory and advancement of green technology R&D (Xin et al., 2024) [35]. These improvements enable firms to signal their commitment to digital and sustainable transformation to investors, thereby increasing attention from investors and financial analysts. On the other hand, digital–real integration enhances the efficiency of information exchange between firms and diverse innovation entities, thereby strengthening technological spillover effects and further promoting firm green innovation (Yang et al., 2014) [36]. Meanwhile, green technology R&D activities are characterized by prolonged development cycles and significant externalities. By leveraging digital technologies, firms can strengthen collaborative innovation with upstream and downstream partners within the supply chain, while also enabling rapid cross-industry sharing of green technology R&D information. This capability is crucial for advancing green collaborative innovation across firms. Based on the foregoing analysis, the following hypothesis is proposed:

H4:

Digital–real integration exerts a positive effect on firm green technology innovation through the mediating role of reducing information asymmetry.

4. Materials and Methods

4.1. Variable Definition

4.1.1. Core Explanatory Variable

We define digital–real integration as the core explanatory variable (TechConv). Building upon the theories of innovation economics and general-purpose technologies and drawing upon the research of Kwon (2020) [37], this study adopts a technology-driven industrial integration perspective. Utilizing Chinese patent application data, the study extracts classification codes from each patent to obtain technology-related data. By leveraging the correspondence between technologies and industries, this data is mapped into industrial integration metrics to measure the level of digital–real integration. This concept highlights the proactive role of the real economy, whereby real-sector industries actively incorporate digital factors or technologies to drive technological transformation. To measure this integration, we use the citation of digital technology patents by non-digital patents in the real sector as a proxy for technological integration between the digital and real industries. Following Huang Xianhai et al. (2023) [14], we employ the reference table that maps the correspondence between the classification of the digital economy’s core industries and the International Patent Classification (IPC) (2023) to identify the primary IPC codes associated with firms’ patents. Based on this classification, we distinguish between non-digital and digital technology invention patents. If a non-digital patent cites a digital technology patent, we regard this as an instance of digital–real integration undertaken by the firm. Finally, we take the natural logarithm of the number of such citations plus one as the proxy measure for the core explanatory variable.

4.1.2. Explained Variable

We define firm green technology innovation as the explained variable (GreT). In this paper, green technology innovation refers to innovative activities that contribute to environmental protection, emission reduction, resource recycling, and the green transformation of the economy throughout the entire product lifecycle. Following Tao Feng et al. (2021) [38], we measure firm green technology innovation as the sum of green utility model patent applications and green invention patent applications.

4.1.3. Control Variables

The following control variables are selected in this paper: (1) Firm asset profitability (Roa). The profitability of enterprise assets provides the necessary resource foundation for green technological innovation. However, the ultimate impact of this depends on whether the enterprise can overcome the short-term profit pressure and invest funds in long-term green innovation under the guidance of external regulations and internal strategies. (2) Financial situation (Lev). An excessively high debt-to-asset ratio can lead to an increase in financial risks, a reduction in resources for research and development, and heightened short-term debt repayment pressure, thereby discouraging enterprises from investing in long-term and high-risk green technological innovations (Xu, 2021) [39]. (3) Revenue growth rate (Dbincome). To enhance short-term performance, excessively rapid growth in revenue may restrain long-term and high-risk investment in green technological innovation. (4) Firm size (Size). A larger enterprise scale helps to reduce the risks associated with green technology innovation (Lin and Chen, 2022) [40]. (5) Firm growth (Dtasset). An enterprise’s annual operating revenue can provide more abundant financial resources for green technological innovation. However, it may also weaken investment in long-term green innovation projects due to the management’s focus on short-term performance. Therefore, the ultimate impact depends on whether the enterprise regards green transformation as the core growth driver. (6) Nature of ownership (SOE). Enterprises of different natures usually have different motivations and capabilities for green technological innovation. (7) Firm age (Age). A larger enterprise scale usually facilitates green technological innovation, as it can provide more abundant resources and greater risk tolerance for such innovation.

4.1.4. Mediating Variables

The following mediating variables are selected in this paper: (1) Financing constraint (FC). Digital–real integration alleviates enterprise financing constraints by reducing information asymmetry and expanding financing channels. Digital technologies enhance the transparency of enterprises’ credit, lower financing costs and thresholds, and provide stable financial support for long-cycle and high-investment green research and development activities, thereby promoting green technological innovation. (2) Total factor productivity (TFP). Digital–real integration enables the deep application of digital technologies to restructure the production processes and management models of enterprises, significantly enhancing total factor productivity. Productivity helps enterprises unlock more innovative resources and strengthens their ability to invest in green research and development. (3) Investor attention (Investorf). Digital–real integration promotes green technological innovation by enhancing investor attention. On the one hand, digital transformation increases the transparency of enterprises’ information, attracting investor attention, alleviating financing constraints, and providing financial support for green innovation; on the other hand, high attention from investors creates external supervision pressure, compelling management to increase green R&D investment and ultimately enhance the green innovation level of enterprises. Drawing on the practice of Wang et al. (2021) [41], the annual median of the Baidu Search Index is used as the measurement indicator for investor attention. (4) Analyst attention (Analystf). Digital–real integration can strengthen enterprise information transparency, thereby attracting heightened attention from financial analysts. As external stakeholders with monitoring functions, analysts mitigate corporate financing constraints by facilitating timely and reliable information dissemination. This, in turn, incentivizes management to prioritize long-term green development strategies, ultimately fostering green technological innovation within enterprises.

4.1.5. Moderating Variables

The following moderating variables are selected in this paper: (1) Government green subsidy (GGS). The government’s green subsidies provide financial support and policy incentives for enterprises to carry out green technological innovations, which can enhance the enabling effect of digital–real integration on green technological innovation. (2) Government environmental regulation (GER). Government environmental regulations, as an important external institutional pressure, can enhance the enabling effect of digital–real integration on green technological innovation. This is because strict environmental policies prompt enterprises to more effectively utilize digital technologies to optimize production processes and reduce emissions, thereby accelerating the green transformation process. (3) Government quality (GQ). High-quality governments usually have stronger policy implementation capabilities, resource allocation abilities and institutional guarantees, which can optimize the external environment for the integration of the real economy and the digital economy, thereby enhancing their role in promoting green technological innovation. (4) Digital infrastructure development (DACL). Digital infrastructure can enhance the technological empowerment effect of the integration of the digital and physical worlds. A well-developed digital infrastructure can reduce the cost of enterprises’ digital transformation, accelerate the circulation of data elements, and thereby strengthen the promoting effect of the integration of the digital and physical worlds on green technological innovation. (5) Marketization level (Market). The level of marketization reflects the efficiency of regional resource allocation and the degree of improvement of the institutional environment. In regions with high marketization, there is high transparency of information and fairer competition, which can enhance the promoting effect of digital and physical integration on green technological innovation; in regions with low marketization, there may be institutional frictions, thereby weakening the integration effect.

4.2. Data

This study uses panel data from A-share listed firms in Shanghai and Shenzhen spanning from 2008 to 2023. The variable definitions and data sources are as shown in

Table 1. Variable definitions and data sources.

Variable	Variable Name and Variable Symbol	Variable Description	Data Source and Website Address
Core Explanatory Variable	Digital–real integration (TechConv)	Take the logarithm of the number of times a non-numeric patent references a numeric patent plus one.	China National Intellectual Property Administration Database https://pss-system.cponline.cnipa.gov.cn
Explained Variable	Green technology innovation (GreT)	The total number of green utility model patent applications and green invention patent applications filed by the firm.	Chinese Research Data Services Database https://www.cnrds.com
Control Variables	Firm asset profitability (Roa)	The ratio of net profit to total assets.	China Stock Market & Accounting Research Database https://data.csmar.com/
	Financial situation (Lev)	Total liabilities divided by total assets.
	Revenue growth rate (Dbincome)	The change in revenue divided by lagged revenue.
	Firm size (Size)	The natural logarithm of total assets.
	Firm growth (Dtasset)	The change rate of total assets.
	Nature of ownership (SOE)	A dummy variable equal to 1 if the firm is a state-owned enterprise (SOE), and 0 otherwise.
	Firm age (Age)	The natural logarithm of one plus the number of years since listing.
Mediating variable	Financing constraint (FC)	The cost of equity financing measured using the price–earnings ratio (PEG) and dividend yield model.	China Stock Market & Accounting Research Database https://data.csmar.com/
	Total factor productivity (TFP)	Firm total factor productivity estimated using the Olley–Pakes (OP) method.
	Analyst attention (Analystf)	The number of analyst teams covering the firm each year.
	Investor attention (Investorf)	The annual median value of the Baidu Index.	Baidu Index database https://index.baidu.com/v2/index.html#/
Moderating variable	Government green subsidy (GGS)	green-related government subsidies/total assets.	China Stock Market &Accounting Research Database https://data.csmar.com/
	Government environmental regulation (GER)	The proportion of environmental regulation–related keywords in municipal government work reports.
	Digital infrastructure development (DACL)	The proportion of keywords related to “new digital infrastructure” in government reports.
	Government quality (GQ)	The intensity of intellectual property protection, measured by the number of concluded intellectual property trial cases.	Peking University Law Database https://www.pkulaw.com/
	Marketization level (Market)	The regional marketization index.	China Market Index Database https://cmi.ssap.com.cn/

.

4.3. Data Analysis

4.3.1. Descriptive Statistical Analysis

The descriptive statistical analysis of the variables in this study is shown in

Table 2. Statistical description of all variables.

Variable	Obs	Mean	Std. Dev.	Min	Max
GreT	13,071	1.1295	1.3055	0.0000	7.1647
TechConv	13,071	0.4241	0.7475	0.0000	6.3919
Size	13,071	22.2880	1.3557	19.1595	28.6365
Roa	13,071	0.0496	0.0623	−1.0570	0.5415
Lev	13,071	0.3979	0.1885	0.0140	0.9793
Dbincome	13,071	0.2447	5.4069	−0.9805	526.0425
Dtasset	13,071	0.1723	0.9329	−0.9830	77.6999
SOE	13,071	0.2978	0.4573	0.0000	1.0000
Age	13,071	1.8046	0.9314	0.0000	3.4340

. According to the descriptive statistics results, the sample contains 13,071 observations. The mean value of the explained variable GreT is 1.1295, with a standard deviation of 1.3055, and the range of values is from 0 to 7.1647. This indicates that there is a significant variation in the level of green technological innovation among enterprises, with zero values present and the leading enterprises performing particularly well. The mean value of the core explanatory variable TechConv is 0.4241, with a standard deviation of 0.7475. The overall level is relatively low, but the distribution is dispersed. Among the control variables, Size (22.2880), Roa (0.0496), and Lev (0.3979) have a reasonable distribution; the mean value of SOE is 0.2978, and the proportion of state-owned enterprises is approximately 29.78%. The overall heterogeneity of the sample is good, making it suitable for conducting empirical analysis.

4.3.2. Correlation and Collinearity Tests

The results of the correlation analysis for the variables are shown in

Table 3. Results of variable correlation analysis.

Variable	GreT	TechConv	lnAge	Size	Roa	Lev	Dbincome	Dtasset	SOE
GreT	1
TechConv	0.4406 *	1
lnAge	0.2383 *	0.1499 *	1
Size	0.5330 *	0.3954 *	0.5032 *	1
Roa	−0.0625 *	−0.00580	−0.1663 *	−0.0347 *	1
Lev	0.3549 *	0.1796 *	0.3898 *	0.5476 *	−0.3587 *	1
Dbincome	−0.0111	−0.0100	−0.0216 *	−0.0120	−0.0228 *	−0.0220 *	1
Dtasset	−0.00920	−0.0140	−0.0304 *	−0.0458 *	0.0423 *	−0.0198 *	0.3280 *	1
SOE	0.2337 *	0.1559 *	0.4497 *	0.4281 *	−0.1074 *	0.3238 *	−0.0119	−0.0249 *	1

Note: * indicate statistical significance at the 10% levels, respectively

. It can be seen that the correlation coefficients and their significance levels align with the research hypotheses, providing preliminary validation for the hypotheses proposed in this paper.

To ensure the reliability and interpretability of the regression estimates, we conducted a formal multicollinearity diagnostic using variance inflation factors (VIFs). As reported in

Table 4. Results of the multicollinearity test.

	Size	Lev	lnAge	SOE	Roa	X	Dtasset	Dbincome	Mean
VIF	2.080	1.740	1.520	1.360	1.203	1.190	1.130	1.130	VIF
1/VIF	0.482	0.574	0.659	0.738	0.812	0.844	0.887	0.889	1.420

, all explanatory variables exhibit VIF values well below the conventional thresholds—ranging from 1.130 (Dtasset and Dbincome) to 2.080 (Size), with a mean VIF of 1.420. Given that VIFs < 5 are widely regarded as indicative of negligible multicollinearity and all observed values fall substantially below this stricter benchmark, we conclude that linear dependencies among predictors are minimal. Consequently, the selected variables are statistically appropriate, and the regression model is robust against collinearity-induced estimation bias, supporting valid inference and further analysis.

4.4. Model

4.4.1. Basic Model

The empirical model of the impact of digital–real integration on firm green technology innovation is constructed as follows:

GreT_it = α₀ + α₁TechConv_it + α_nControls_it + μ_i + γ_t + ε_it

(1)

where i is the firm, and t is the year. GreT_it is the green technology innovation of firm i in year t. TechConv_it is the digital–real integration of firm i in year t. Controls represents the control variables. μ_i is the firm fixed effects, γ_t is the year fixed effects, and ε_it is the random error term.

4.4.2. Mechanism Verification Model

Referring to the practice of Jiang Ting (2022) [42], this paper uses the two-stage regression method for the mechanism analysis and constructs the econometric model as follows:

GreT_it = α₀ + α₁TechConv_it + α_nControls_it + μ_i + γ_t + ε_it

(2)

FC_it = α₀ + α₁TechConv_it + α_nControls_it + μ_i + γ_t + ε_it

(3)

Investorf_it = α₀ + α₁TechConv_it + α_nControls_it + μ_i + γ_t + ε_it

(4)

TFP_it = α₀ + α₁TechConv_it + α_nControls_it + μ_i + γ_t + ε_it

(5)

Analystf_it = α₀ + α₁TechConv_it + α_nControls_it + μ_i + γ_t + ε_it

(6)

where i is the firm, and t is the year. Details of other variables are presented in Table 1.

4.4.3. Expansive Analytical Model

From the perspective of a “well-functional government” and an “efficient market”, this paper explores whether government behavior and market mechanisms moderate the impact of digital–real integration on firm green technology innovation. We construct the following regression models:

$${GreT}_{it} = {\alpha }_0 + {\alpha }_{1}Tec\mathrm{h}Con{v}_{it} + \sum {\alpha }_{n}Control{s}_{it} + {\mu }_i + {\gamma }_t + {\epsilon }_{it}$$

(7)

$${GreT}_{it}={ \alpha }_0+{ \alpha }_{1}Tec\mathrm{h}Con{v}_{it}+{ \alpha }_{2}Tec\mathrm{h}Con{v}_{it} \times {GGS}_{it}+{\alpha }_3{GGS}_{it}+\sum {\alpha }_{n}Control{s}_{it}+{ \mu }_i+{\gamma }_t+{\epsilon }_{it}$$

(8)

$${GreT}_{it}={ \alpha }_0+{\alpha }_{1}Tec\mathrm{h}Con{v}_{it}+{ \alpha }_{2}Tec\mathrm{h}Con{v}_{it} \times {\mathrm{GER}}_{it}+{\alpha }_3{\mathrm{GER}}_{it}+\sum {\alpha }_{n}Control{s}_{it}+{\mu }_i+{\gamma }_t+{\epsilon }_{it}$$

(9)

$${GreT}_{it}={ \alpha }_0+{\alpha }_{1}Tec\mathrm{h}Con{v}_{it}+{\alpha }_{2}Tec\mathrm{h}Con{v}_{it} \times {GQ}_{it}+{\alpha }_3{GQ}_{it}+\sum {\alpha }_{n}Control{s}_{it}+{ \mu }_i+{\gamma }_t+{\epsilon }_{it}$$

(10)

$${GreT}_{it}={ \alpha }_0+{ \alpha }_{1}Tec\mathrm{h}Con{v}_{it}+{\alpha }_{2}Tec\mathrm{h}Con{v}_{it} \times { DACL}_{it}+{\alpha }_3{DACL}_{it}+\sum {\alpha }_{n}Control{s}_{it}+{ \mu }_i+{\gamma }_t+{\epsilon }_{it}$$

(11)

$${GreT}_{it}={\alpha }_0+{\alpha }_{1}TechCon{v}_{it}+{\alpha }_{2}TechCon{v}_{it}\times {Market}_{it}+{\alpha }_3{Market}_{it}+\sum {\alpha }_{n}Control{s}_{it}+{\mu }_i+{\gamma }_t+{\epsilon }_{it}$$

(12)

where i is the firm, and t is the year. Details of other variables are presented in Table 1.

5. Results of Empirical Analysis and Their Discussion

5.1. Baseline Regression

Table 5. Baseline regression.

Variables	(1)	(2)	(3)	(4)
	GreT	GreT	GreT	GreT
TechConv	0.769 ***	0.472 ***	0.265 ***	0.201 ***
	(56.10)	(34.66)	(12.60)	(8.85)
Size		0.370 ***	0.405 ***	0.397 ***
		(37.58)	(19.87)	(9.90)
Roa		−0.403 **	−0.159	−0.054
		(−2.42)	(−1.04)	(−0.33)
Lev		0.651 ***	0.155	−0.181
		(10.06)	(1.60)	(−1.37)
Dbincome		−0.002	−0.001 **	0.002
		(−1.12)	(−2.37)	(1.32)
Dtasset		0.023 **	0.026 ***	0.018 **
		(2.26)	(3.81)	(2.47)
SOE		0.039 *	0.166 ***	0.195 ***
		(1.66)	(3.80)	(2.63)
Age		−0.063 ***	−0.047 ***	−0.082 ***
		(−5.19)	(−2.83)	(−2.65)
Constant	0.803 ***	−7.468 ***	−8.232 ***	−7.946 ***
	(68.15)	(−37.72)	(−19.31)	(−9.36)
Time FE	No	No	Yes	Yes
Firm FE	No	No	No	Yes
Observations	13,071	12,764	12,764	12,764
R-squared	0.194	0.347	0.237	0.241

Note: ***, ** and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively.

presents the basic regression results. The panel autocorrelation test results indicate the presence of first-order autocorrelation (F = 113.923, p = 0.000). We address this issue by using robust firm-level clustering standard errors in parentheses. In columns (1) to (4), control variables, year fixed effects, and firm fixed effects are sequentially introduced into the regression analysis. Overall, the regression coefficients in columns (1) to (4) are significantly positive at the 1% significance level, which suggests that digital–real integration has a significant positive effect on enhancing firm green technology innovation. From columns (1) to (4), the coefficients of the core explanatory variables progressively decrease. This is because by controlling for other influencing factors, firm-level fixed effects, and year-level fixed effects, the impact of these variables on firm green technology innovation is gradually eliminated. Ultimately, this yields the “net effect” of digital–real integration on firm green technology innovation. The result shows that by adopting digital technologies, real industries can more accurately identify bottlenecks in current green technologies and obtain extensive R&D information and resources. This facilitates the breakdown of information barriers among R&D entities, thereby improving the efficiency of green technology R&D and collaborative innovation within firms. These empirical findings provide strong support for Hypothesis 1 proposed in this paper. Furthermore, the coefficients of certain control variables (such as Roa and Lev) become statistically insignificant upon the inclusion of firm and time fixed effects. This suggests that firm green technology innovation is primarily driven by enduring firm-specific attributes, including long-term strategic orientation, managerial capability and organizational culture, rather than transient financial performance fluctuations.

5.2. Endogeneity

Instrumental Variables

Considering that the model may have endogeneity problems due to omitted variables and bidirectional causality, and drawing on Nunn et al. (2014) [43] to ensure the validity of the fixed-effects model, this study uses the average number of fixed telephones per million people at the city level in 1984 as the instrumental variable for digital–real integration. The sample data in this study are panel data, whereas this instrumental variable is cross-sectional data. To satisfy the requirements of the panel regression model, the average number of fixed telephones per million people at the city level in 1984 was interacted with the time trend term to construct the instrumental variable for digital–real integration. This instrumental variable constructed from historical fixed-line telephone density is exogenous at the firm level because the legacy telecommunications infrastructure has no direct causal effect on contemporary green innovation activities except indirectly via its role in shaping subsequent digitalization. The number of fixed-line telephones in a city historically does not have a direct impact on current levels of firm green technology innovation, satisfying the exogeneity condition. Furthermore, a higher number of fixed-line telephones in a city may indicate faster development of local information technology infrastructure, which can provide essential infrastructure and platform support for digital–real integration. Therefore, this variable also satisfies the relevance criterion required for selecting an instrumental variable.

We use the 2SLS method to estimate the econometric model. The first-stage regression results are presented in column (1) of

Table 6. Endogeneity.

Variables	(1)	(2)
	TechConv	GreT
FixTel	0.0235 ***
	(0.0033)
TechConv	0.332 ***	2.727 ***
	(7.33)	(7.02)
F-statistic	98.5263
Kleibergen–Paap rk LM		49.069
		[0.0000]
Cragg–Donald Wald F		61.933
		{16.38}
Kleibergen–Paap rk Wald F		49.251
		{16.38}
Constants	−4.6370 ***	3.047 *
	(0.2152)	(1.68)
Control Variables	Yes	Yes
Time FE	Yes	Yes
Firm FE	Yes	Yes
Observations	12,111	12,111
R-squared	0.332	0.353

Note: The p-values are within [ ], and the critical values for the Stock–Yogo weak identification test at the 10% level are within { }. *** and * indicate statistical significance at the 1% and 10% levels, respectively.

. Additionally, the regression coefficients of the instrumental variable on digital–real integration are positive, satisfying the relevance principle of the instrumental variable. The results of the second-stage regression are shown in column (2) of

Table 7. Robustness tests.

Variable	(1)	(2)	(3)	(4)	(5)
	Adjusting the Patent Publication Period	Deleting Some Industries	Setting a Three-Year Window Period	Considering the Time Lag	Replacing the Measurement Method
TechConv	0.199 ***	0.201 ***	0.203 ***
	(8.26)	(8.69)	(6.52)
L. TechConv				0.176 ***
				(5.84)
TechConv_new					0.167 ***
					(9.61)
TechConv_w
Constant	−7.82 ***	−7.79 ***	−7.67 ***	−7.61 ***	0.20
	(−8.38)	(−8.82)	(−7.15)	(−5.66)	(0.91)
Controls	Yes	Yes	Yes	Yes	Yes
Firm FE	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes
Observations	10,608	11,869	8287	6518	14,761
R-squared	0.794	0.793	0.809	0.820	0.197

Note: *** indicate statistical significance at the 1% levels, respectively.

. The coefficient of digital–real integration remains significantly positive, indicating that digital–real integration can still significantly promote firm green technology innovation after controlling for potential endogenous interference. As shown, the p-value of the LM statistic is 0.000, and the Wald F statistic is 61.933, which exceeds the 10% critical value of 16.38. These results indicate that the instrumental variables do not have under-identification and weak-instrument issues.

5.3. Robustness Tests

5.3.1. Considering the Cycle of Patent Disclosure

Generally speaking, the longest time span from patent application to publication is 1.5 years. Therefore, we exclude the sample data from the last two years, which can avoid measurement errors caused by unpublished patents at the end of the sample period. This approach aims to more accurately reflect the true impact of digital–real integration on firm green technology innovation. Thus, we delete the data for 2022 and 2023 in the sample. The robustness check presented in Table 7 (column 1) addresses the concern regarding the patent publication cycle. The findings remain consistent with the main analysis, indicating that this temporal factor does not substantially alter the research conclusions.

5.3.2. Excluding Samples from the Communication and Information Technology-Related Industries

Digital–real integration refers to the capability of digital industries to serve and empower real industries, emphasizing the level of mutual integration between them. However, industries such as telecommunications and information technology inherently already have a certain level of digital infrastructure and digital talent reserves, which gives them a comparative advantage in integrating digital industry technologies. To avoid potential biases in the measurement of digital–real integration caused by special values, this study excludes data samples from the communication and information technology-related industries. We find that digital–real integration still significantly promotes firm green technology innovation.

5.3.3. Setting a Three-Year Window Period for Patent Citations

Some scholars have found that citations of outdated patents may be counted by patent examiners or lawyers rather than truly cited by firms, leading to overestimation of the digital–real integration measure (Jaffe, 2000) [44]. Therefore, referring to Yi et al. (2021) [45], we adjust the window period for patent citation information to three years, i.e., only counting the number of publicly available digital patents cited by real industry patents in the past three years. Then we recalculate digital–real integration. The results show that the regression coefficient of the core explanatory variable remains robust.

5.3.4. Considering the Time Lag of the Digital–Real Integration Effect

Acknowledging the potential temporal dimension of the effect, we re-estimate the model using the one-period lag of digital–real integration. As presented in column (4) of Table 7, its coefficient remains positive and statistically significant, confirming a persistent impact on green technology innovation.

5.3.5. Replacing the Measurement Method of Digital–Real Integration

In the previous measurement, a patent is defined as a firm’s digital–real integration behavior when two conditions are met. On the one hand, its IPC main classification belongs to non-digital industry technology. On the other hand, at least one of the cited patents is classified as digital industry technology. Accordingly, we aggregate this indicator to the firm-year level to recalculate the number of digital–real integration events of the firm. Then we add one to this value and take the natural logarithm, using the result as the firm’s digital–real integration index (TechConv_new). The regression results in column (5) of Table 7 show that the impact of the re-measured and aggregated digital–real integration on firm green technology innovation remains significantly positive.

5.4. Heterogeneity Analysis

5.4.1. Heterogeneity Based on Firm Green Technology Innovation Types

To explore the impact of digital–real integration on the different types of firm green technology innovation, this paper classifies firm green patents into three categories: green patents for new energy, green patents for conventional energy efficiency and green patents for end-of-pipe pollution control. This paper classifies green technology patents based on the classification system released by the China National Intellectual Property Administration (CNIPA) in August 2023. By identifying the IPC of the firms’ green patents in the sample, we categorize them into the following three types: green patents for new energy (NPP), green patents for conventional energy efficiency (GEP), and green patents for end-of-pipe pollution control (EPP). As presented in

Table 8. Heterogeneity analysis: types of firm green technology innovation.

Variables	(1)	(2)	(3)
	Green Patents for New Energy (NPP)	Green Patents for Conventional Energy Efficiency (GEP)	End-of-Pipe Pollution Control Green Patents (EPP)
TechConv	1.816	0.823 **	3.210 ***
	(1.27)	(2.19)	(2.71)
Constant	−19.577	−11.821	−23.295
	(−1.26)	(−0.91)	(−0.85)
Controls	Yes	Yes	Yes
Firm FE	Yes	Yes	Yes
Year FE	Yes	Yes	Yes
Observations	2694	2694	2694
R-squared	0.130	0.212	0.220

Note: *** and ** indicate statistical significance at the 1% and 5% levels, respectively.

, digital–real integration exerts a significantly positive impact on both GEP and EPP. However, its effect on NPP is statistically insignificant. This indicates that the current effect of digital–real integration on promoting the formation of green patents for new energy is limited. First, at the technical level, the development of new energy technologies is characterized by long cycles and high trial-and-error costs (Zhang et al., 2020) [46]. Breakthroughs in new energy technologies rely more on sustained R&D investment than on short-term digital technology deployment and application. Second, in terms of integration, the integration of China’s real economy and digital economy currently remains largely focused on optimizing production processes, resource monitoring, and efficiency improvements. The penetration into foundational technological innovation remains insufficient. Third, regarding corporate motivation, enterprises may prioritize digital technology investments in lower-risk, quicker-return end-stage management areas, thereby crowding out digital technology investments needed for breakthrough new energy technologies. However, it is relatively less difficult to develop green technologies that enhance the energy efficiency of conventional energy and end-of-pipe treatment-type technologies, especially end-of-pipe governance green patents. By applying digital technologies such as big data and the Internet of Things, firms can monitor pollution emission data from pollution sources in real time. Through analyzing the operational status of end-of-pipe pollution control and energy equipment, they can identify and analyze existing problems. This process can stimulate R&D demand for more efficient governance technologies and energy utilization technologies, which can promote the invention of EPP and GEP. In conclusion, digital–real integration has a greater promoting effect on EPP, followed by GEP, while its impact on NPP is not statistically significant.

5.4.2. Heterogeneity Based on the Level of Firm Internal Control

Superior internal control within a firm indicates stronger internal governance capacity, which can directly affect the efficiency of the firm in obtaining and utilizing internal and external resources such as human capital, financial assets, and physical resources. This can influence the promoting effect of digital–real integration on firm green technology innovation. Therefore, referring to Chen (2018) [47], this paper uses the internal control index provided by the Dibo database to measure the internal control level of firms. We then divide the sample into subgroups based on the median of this index to examine the differential impacts of digital–real integration on firm green technology innovation under varying internal control levels. The regression results indicates that digital–real integration has a more significant promotional effect on firm green technology innovation with higher internal control levels. Firms with high internal control levels tend to exhibit higher management efficiency, which can effectively reduce information asymmetry among departments and ensure financial information is more open and transparent, which can create a stable and favorable internal environment for digital–real integration. Conversely, firms with weak internal control may face issues such as resource waste and inefficiency, which may constrain the efficiency of digital technology adoption in real firms. These problems can prevent digital–real integration from realizing its potential advantages.

5.4.3. Heterogeneity Based on Firm Property Rights

Firms with different property rights structures exhibit significant disparities in resource acquisition and innovation incentives. Based on institutional theory, state-owned firms may have greater access to policy resources for green transformation, while non-state-owned firms rely more on market incentives and efficiency-driven approaches. This determines varying degrees of enabling effects of digital–real integration on green innovation. Therefore, in this section, we investigate the heterogeneous impacts of digital–real integration on firm green technology innovation under different firm property rights. We classify the research samples into two types: state-owned firms and private firms. As shown in the regression results of columns (3) and (4) in

Table 9. Heterogeneity analysis: firm internal control, property rights and firm size.

Variables	(1)	(2)	(3)	(4)	(5)	(6)
	Low Internal Control Level	High Internal Control Level	State-Owned Firms	Private Firms	Small and Medium-Sized Firms	Large-Scale Firms
TechConv	0.107 *	0.223 ***	0.176 ***	0.226 ***	0.113 ***	0.205 ***
	(1.96)	(8.14)	(5.83)	(6.52)	(4.44)	(7.36)
Constant	−9.108 ***	−6.637 ***	−8.767 ***	−6.039 ***	0.368 ***	1.401 ***
	(−5.29)	(−6.12)	(−8.45)	(−3.49)	(6.73)	(6.86)
Inter-group Coefficients Difference p-values	0.039		0.038		0.005
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Firm FE	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
Observations	2226	7872	8434	3670	5787	6051
R-squared	0.808	0.814	0.766	0.819	0.738	0.806

Note: The p-value for the difference in inter-group coefficients is obtained based on Fisher’s Permutation Test with bootstrap, calculated by repeating sampling 1000 times. The same applies below. *** and * indicate statistical significance at the 1% and 10% levels, respectively.

, the regression coefficients of digital–real integration for both state-owned and private firms are significantly positive. After Fisher’s Permutation Test, the p-value of the inter-group coefficient between state-owned and private firms is 0.038, indicating that there is a significant inter-group difference. That is to say, digital–real integration has a greater promotional effect on the green technology innovation of private firms. Digital–real integration requires sound and perfect governance systems, standard systems, regulatory systems and ecosystems as support. Compared to state-owned firms, private firms have stronger competitive awareness and higher flexibility in digital transformation. Private firms can quickly adjust institutional mechanisms and organizational structures to establish support systems for digital–real integration. Therefore, they can more efficiently leverage the positive promoting effect of digital–real integration on firm green technology innovation. However, state-owned firms are characterized by their large size, complicated management levels and slow decision-making processes, which makes it difficult to quickly adjust and establish internal and external conditions to support firm digital transformation. The existence of these problems has led state-owned firms to be more sluggish in the introduction and application of digital technology, thus leading to a relatively low degree of digital–real integration. Therefore, compared with state-owned firms, digital–real integration in private firms has a more obvious promoting effect on firm green technology innovation.

5.4.4. Heterogeneity Based on Firm Size

From the perspective of innovation economics, firm size is the key factor influencing technological integration capabilities and resilience to innovation risks. Compared to small firms, large firms possess redundant resources and greater risk-bearing capacity, which facilitates enhanced technological integration. Differences in firm size affect the mechanisms and outcomes through which technological integration enables green innovation. Therefore, in this section, the paper aims to investigate how digital–real integration differently influences firm green technology innovation across varying firm sizes. By using the median of total firm assets as the baseline, we classify firm size into two categories: large-scale firms and small and medium-sized firms (SMEs). The regression results in columns (5) and (6) of Table 9 show that digital–real integration has a positive impact on firm green technology innovation of both large-scale firms and SMEs. Specifically, compared with SMEs, the digital–real integration of large-scale firms has a greater promotion effect on firm green technology innovation. However, SMEs have relatively limited resources in terms of human, financial and material resources. On the one hand, this makes it difficult to realize the positive effects of digital–real integration on firm green technology innovation; on the other hand, this makes firms more inclined to invest in the purchase of environmental protection equipment rather than carry out high-input and high-risk green technological research and development when dealing with the pollutants.

5.4.5. Technological Characteristics of Core Industries in Digital Economy

To explore the heterogeneous impacts of integrating different types of digital economy core industries with the real economy on firm green technology innovation, this paper first classifies and identifies the IPC codes of patents held by firms operating in the digital economy core industries. The classification is conducted based on the reference table for mapping digital economy core industry classifications to international patent classifications (2023), issued by the China National Intellectual Property Administration (CNIPA) in March 2023. Through this process, four categories of digital economy core industries are identified: digital product manufacturing, digital product services, digital technology application, and digital factor-driven industries. As presented in

Table 10. Heterogeneity analysis: technological characteristics of core industries in digital economy.

Variables	(1)	(2)	(3)	(4)
	Digital Product Manufacturing Industry	Digital Product Service	Digital Technology Application	Digital Factor Driven Industry
TechConv01	0.197 ***
	(8.53)
TechConv02		0.333 ***
		(4.09)
TechConv03			0.261 ***
			(9.14)
TechConv04				0.300 ***
				(8.70)
Constant	−7.666 ***	−8.064 ***	−7.745 ***	−7.861 ***
	(−8.74)	(−8.89)	(−8.80)	(−8.90)
Control variables	Yes	Yes	Yes	Yes
Firm FE	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
Observations	12,141	12,141	12,141	12,141
R-squared	0.795	0.792	0.795	0.795

Note: *** indicate statistical significance at the 1% levels.

, the integration of each type of digital core industry with the real economy exerts a significantly positive influence on firm green technology innovation. The regression coefficient for digital product services is the highest among the four categories, indicating the strongest positive effect on green technology innovation. In contrast, the regression coefficient for digital product manufacturing is the smallest, suggesting a relatively weaker impact on firm-level green technology innovation.

5.5. Mechanism Analysis

We examine the potential mechanisms in

Table 11. Results of mechanism analysis.

Variables	(1)	(2)	(3)	(4)	(5)
	GreT	FC	TFP	Investorf	Analystf
TechConv	0.201 ***	−0.010 **	0.013 *	0.520 ***	0.047 ***
	(8.85)	(−2.19)	(1.90)	(3.17)	(2.79)
Constant	−7.946 ***	0.037	−4.976 ***	−76.574 ***	−12.24 ***
	(−9.36)	(1.02)	(−12.41)	(−7.93)	(−14.64)
Controls	Yes	Yes	Yes	Yes	Yes
Firm FE	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes
Observations	12,764	8130	11,457	10,550	12,764
R-squared	0.241	0.182	0.794	0.214	0.219

Note: ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively.

. Column (1) of Table 11 presents the baseline result of the impact of digital–real integration on firm green technology innovation. The results in column (2) indicate that digital–real integration significantly alleviates firm financing constraints. Combined with the results in column (1), this suggests that digital–real integration promotes firm green technology innovation by mitigating financing constraints, thereby validating Hypothesis H2 of the theoretical analysis. Column (3) shows that digital–real integration enhances firm green technology innovation by significantly improving productivity, thereby validating Hypothesis H3 of the theoretical analysis. The regression results in columns (4) and (5) reveal that digital–real integration significantly increases both investor and analyst attention, indicating that digital–real integration reduces external information asymmetry, thus promoting firm green technology innovation, thereby validating Hypothesis H4 of the theoretical analysis. In conclusion, digital–real integration can promote firm green technology innovation through three channels: alleviating financing constraints, improving productivity and reducing information asymmetry.

5.6. Extension Analysis

We have verified that digital–real integration has a significant positive impact on firm green technology innovation. Next, from the perspective of “well-functional government” and “efficient market”, this paper explores whether government behavior and market mechanisms moderate the impact of digital–real integration on firm green technology innovation. This paper selects government green subsidy, government environmental regulation, governance quality, digital infrastructure and marketization level to characterize government behavior. Specifically, on the basis of the baseline regression equation, we introduce the interaction terms to examine the moderating effect of government behavior.

5.6.1. The Moderating Effect of Government Green Subsidy

Drawing on Li et al. (2020) [48], this paper categorizes and identifies the detailed data of government subsidies of listed companies. We identify government subsidies with related fields of “green”, “environmental protection subsidy”, “environment”, “sustainable development”, “clean” and “energy-saving”. We sum these government subsidies and compare them to total assets to obtain the government green subsidy proxy variable (GGS). The regression results indicate that the coefficient of the interaction term between digital–real integration and government green subsidy (TechConv × GGS) is significantly positive. The reason may be that government green subsidy can provide financial support for firms, which can make it easier for them to overcome financial barriers in introducing new digital technology. Sufficient funds can incentivize firms to apply digital technologies to the real economy, which can promote digital–real integration. At the same time, government green subsidies can steer firms to align their digital–real integration efforts with green and sustainable development objectives, thereby incentivizing increased R&D in green patenting. In this way, firms can continuously transmit to the government the signals of green and low-carbon transformation, which can make firms continue to obtain environmental protection subsidies provided by the government.

5.6.2. The Moderating Effect of Government Environmental Regulation

Following the methodology of Tang et al. (2019) [49], we manually collected and compiled keywords related to environmental regulation from municipal government work reports, including “environmental protection, pollution, energy consumption, emission reduction, ecology, green, PM10, PM2.5, sulfur dioxide”, among others. We constructed a proxy for government environmental regulation (GER) by calculating the proportion of these keywords relative to the total word count in each report. The coefficient of the interaction term between digital–real integration and government environmental regulation (TechConv × GER) is significantly positive. Environmental regulation refers to the use of stringent environmental laws and standards by the government to encourage and guide firms toward low-carbon transformation. In practice, environmental regulation can facilitate the integration of the real and digital industries. This is because firms, in order to avoid the high penalties and taxes associated with non-compliance, tend to increase their adoption of digital technologies, optimize production processes, and improve resource utilization efficiency to meet government environmental standards. Furthermore, strengthened environmental regulation can raise public awareness of environmental protection, fostering broader social consensus and a foundation for collective action. At the same time, public oversight can effectively constrain the behavior of both government and firms. Overall, government environmental regulation serves as a complementary mechanism that reinforces the role of digital–real integration in promoting firm green technology innovation.

5.6.3. The Moderating Effect of Government Quality

Referring to Nam et al. (2019) [50], this paper uses the intensity of intellectual property protection as the proxy variable of government quality (GQ). Specifically, we obtain the number of concluded cases of intellectual property trials at the city level, the number of concluded domestic intellectual property trials, and the annual “Report on the Judicial Protection of Intellectual Property Rights in Chinese Courts”, respectively. As shown in

Table 12. Extension analysis.

Variables	(1)	(2)	(3)	(4)	(5)
	GGS	GER	GQ	DACL	Market
TechConv	0.198 ***	0.147 ***	0.192 ***	0.143 ***	0.027
	(8.75)	(3.46)	(8.71)	(3.25)	(0.24)
GGS	−0.143 **
	(−2.43)
TechConv × GGS	0.076 **
	(2.33)
GER		−0.001
		(−0.99)
TechConv × GER		0.001 **
		(2.11)
GQ			0.011
			(0.39)
TechConv × GQ			0.043 **
			(1.99)
DACL				−15.135
				(−1.30)
TechConv × DACL				28.023 **
				(2.37)
Market					−0.022
					(−0.66)
TechConv × Market					0.027 **
					(2.20)
Constant	−8.004 ***	−8.133 ***	−8.027 ***	−8.987 ***	−7.319 ***
	(−9.31)	(−9.02)	(−13.32)	(−9.15)	(−6.94)
Control Variables	Yes	Yes	Yes	Yes	Yes
Firm FE	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes
Observations	12,669	11,285	8586	9332	8172
R-squared	0.242	0.250	0.263	0.250	0.225

Note: *** and ** indicate statistical significance at the 1% and 5% levels, respectively.

, the regression coefficient of the interaction term between digital–real integration and government quality (TechConv × GQ) is significantly positive. The reason may lie in the fact that higher government quality can prevent the illegal use of corporate innovation outcomes by other firms and ensure that firms exclusively enjoy the benefits brought by innovation. Higher government quality also provides a sound external institutional environment to support green technology innovation in the context of digital–real integration. Higher government quality can not only safeguard the rights and interests of innovators, but also help encourage firms to disclose and license their technology, thereby promoting the dissemination, sharing and application of green technology. Overall, improving government quality can enhance the promotional effect of digital–real integration on firm green technology innovation.

5.6.4. The Moderating Effect of New Digital Infrastructure Construction

We collect the work reports of prefecture-level city governments from 2008 to 2023 and identify the number of words related to “new digital infrastructure” by using Python 3. We then calculate the proportion of these words to the total number of words as a proxy variable for the level of digital infrastructure construction (DACL). As shown in column (4) of Table 12, well-developed digital infrastructure can enhance the positive impact of digital–real integration on firm green technology innovation. The underlying reason may be that robust digital infrastructure improves firms’ capabilities in data collection, processing, and transmission, enabling faster access to digital information resources and overcoming information barriers or delays caused by underdeveloped infrastructure during the digital innovation process. This provides a favorable external infrastructural environment that supports digital–real integration in promoting green technology innovation.

5.6.5. The Moderating Effect of Marketization Level

The marketization index serves as a composite measure that captures the interplay between governmental and market forces within a given region. We use the marketization index as a proxy variable for “effective market”. Referring to Fan et al. (2011) [51], to examine how regional marketization moderates the impact of digital–real integration on corporate green technology innovation, we introduce a provincial marketization index and interact it with our core explanatory variable. As reported in column (5) of Table 12, the coefficient on the interaction term is positive and statistically significant. This result indicates that a higher level of marketization strengthens the positive effect of digital–real integration on firms’ green innovation outcomes.

6. Conclusions and Policy Implications

6.1. Conclusions

This study primarily includes the following conclusions. Digital–real integration helps improve firm green technology innovation. The results of the mechanism analysis find that digital–real integration can promote firm green technology innovation through three channels: alleviating financing constraints, improving production efficiency and reducing information asymmetry. The results of the heterogeneity analysis show that digital–real integration has a greater promoting effect on the green technology innovation of non-state-owned firms, large-scale firms and firms with high internal control levels. At the same time, digital–real integration has a greater promoting effect on the R&D of end-of-pipe pollution control green patents, followed by the impact on the formation of conventional energy efficiency patents, while its impact on new energy-related green patents is not statistically significant. In the extension analysis, the study finds that government green subsidy, government environmental regulation, government quality, digital infrastructure and marketization can strengthen the promoting effect of digital–real integration on firm green technology innovation.

6.2. Policy Implications

Based on the above findings, we propose the following policy recommendations.

(1)

We need to deepen the technological-level integration between the digital economy and real economy sectors. This study indicates the possibility that such digital–real industrial integration at the technological level exerts a significant positive effect on advancing green technology innovation. Therefore, it is necessary to pay attention to the application of digital industry technology in the innovation process. Firms can leverage digital technologies, including artificial intelligence and simulation tools, to enhance cost efficiency during the technological renewal and upgrading of physical industries. This approach facilitates the digital and intelligent transformation of key processes such as production, manufacturing, and supply chain management within traditional real-economy sectors.

(2)

Firms may enhance the supply of key and general digital technology in the R&D process of firms, which can provide initial digital technology support for their digital transformation. Additionally, the government should provide government green subsidies and other policies to reduce the R&D costs of small and medium-sized firms, which can alleviate the uneven distribution of innovation opportunities among firms caused by the digital technology gap. In this way, firms can lower the threshold for technology integration and stimulate the willingness of firms to integrate and innovate technologies.

(3)

The government may improve the relevant laws, regulations and standards for digital–real integration and strengthen the protection of intellectual property rights. The government should provide corresponding protection measures based on the characteristics of the different links, such as R&D, flow and application of digital technology. The government may promptly define the classification governance standards for emerging industries generated by digital–real integration technology. The government may adopt an intellectual property protection strategy that combines moderate leniency and anti-monopoly governance.

(4)

It is imperative for the government to enhance the development of robust digital infrastructure, encompassing telecommunications networks, high-speed internet, and data center facilities. The government can promote the popularization and application of integrated infrastructure, such as the industrial Internet, and innovative infrastructure, such as industrial chain innovation platforms. The government may also facilitate the intelligent transformation of traditional infrastructure in sectors like energy and transportation, improve the service capacity and operational efficiency of traditional infrastructure and build an external support system for digital–real integration technology.

(5)

The government may accelerate the establishment of a unified national data element market. Efforts should be made to strengthen cooperation, promote data exchange across different industries, and encourage the government, firms and research institutions to jointly promote data circulation among industries. The government should establish a unified policy framework for data sharing to break down data barriers across industries and regions. The government ought to promote the construction of cross-departmental and cross-sectoral data sharing platforms, which can break down data silos to facilitate the flow and integration of data among different departments.

6.3. Limitations of the Study and Future Directions

Although this study has made some valuable discoveries, there are still several shortcomings that need to be further improved in future research. Specifically, these shortcomings are manifested in the following aspects:

(1)

Refining the measurement of digital–real integration. This paper employs the frequency of “non-digital patents citing digital patents” as a proxy variable for digital–real integration. While this approach effectively captures technological integration at the enterprise level, it fails to comprehensively reflect the breadth and depth of integration across other dimensions. For instance, digital integration in non-technical areas such as production processes, organizational management, and supply chain collaboration remains difficult to capture through patent data. Future research could integrate multi-source data, such as textual analysis of corporate digital investment, to develop multidimensional convergence indices that more comprehensively measure digital–real integration.

(2)

Expanding the research scope from China to the world. This study, grounded in data from Chinese listed companies, may be influenced by China’s specific institutional environment and other factors. Both the digital economy and green transition are global issues, with significant differences existing among countries in terms of digital economies, innovation systems and environmental policies. In the future, we can use multinational corporations as case studies to compare the differing impacts of digital–physical convergence on empowering green innovation under various institutional frameworks.