Digital Economy Development and Green Innovation Efficiency from a Two-Stage Innovation Value Chain Perspective

Abstract

Resolving the conflict between economic growth and ecological sustainability is crucial when transitioning from traditional economic development towards a sustainable green model. In this context, the super-NSBM model was applied to measure the efficiency of green innovation technology research and development. Additionally, it was applied to measure the efficiency of the green innovation achievement transformation among 30 Chinese provinces. This evaluation was conducted for the period from 2011 to 2021, and it utilized a two-stage innovation value chain perspective. The entropy method was used to comprehensively calculate the digital economy development index, and the impact of digital economy development on the two-stage green innovation effect was empirically explored using SDM. The findings revealed the following: Firstly, both two-stage green innovation efficiency and digital economy development exhibited significant spatial characteristics. Secondly, digital economy development had a positive effect on two-stage green innovation efficiency not only in the local region but also in neighboring regions. This conclusion remained valid throughout a series of robustness tests. Thirdly, in terms of regional heterogeneity, the eastern region and non-resource-based regions had greater digital economy development dividends than the central, western, and resource-based regions; therefore, the effect on two-stage green innovation efficiency was more significant. Lastly, both intellectual property protection and data factor endowment exerted positive moderating effects on the influence of the digital economy on two-stage green innovation efficiency. The conclusions of this research provide a theoretical backing for and empirical proof of the mechanism of digital economy development and its impact on two-stage green innovation efficiency. Policy implications are suggested based on these findings, such as reinforcing digitalization, establishing targeted policies, and fostering a favorable external environment conducive to technological innovation.

1. Introduction

Since the reform and opening up of the Chinese economy, an extensive growth mode, driven by the factors of production and capital investments, has led to long-term, high-speed economic growth. However, this great positive development has also been accompanied by increasingly serious environmental problems. Addressing the conflicting issues between economic growth and the ecological environment and achieving a comprehensive transition to a green development model have become pressing concerns. Green innovation is an inevitable requirement for achieving the construction of an ecological civilization and sustainable development, and it is also an important way to resolve conflicts and contradictions [1,2]. With the arrival of the wave of digitization, the digital economy spawned by “Internet+”, big data, blockchains, and other new business forms is characterized by high efficiency, a high level of coordination, a low cost, and a high level of knowledge intensity. It has emerged as the principal force driving changes within the new normal of the Chinese economy. Only when the development of the digital economy and green development are mutually promoted and deeply integrated can we truly establish a new pattern of high-quality development where “digital and green dance together, economy and society advance together”. Digital economy development and green innovative development have a profound relationship of mutual need and assistance, rather than being two relatively isolated developmental concepts [3]. Although information on the impact of digital economy development on the efficiency of green innovation is still limited, many scholars have proposed the driving role of digital economy development from various dimensions such as technological innovation, environmental performance, and economic efficiency, laying a solid theoretical foundation for the in-depth study of the relationships among them. In reality, green innovation encompasses two stages: green innovation technology research and development and the transformation of green innovation achievements. These stages constitute a complex and diversified process of passing down the innovation value chain [4,5]. What facilitating role does the development of the digital economy play in enhancing the efficiency of green innovation throughout these two stages? Furthermore, how does this role vary spatially? Delving into the aforementioned questions served as the primary focus of this study.

For the development of the digital economy, the existing research has mainly focused on sorting out its connotations, measurements, and influencing factors. The connotation of the digital economy can be divided into narrow and broad definitions. Narrowly defined, the digital economy is a new type of information industry closely related to information and communication technology. It is a new national economic industry that continuously utilizes digital technology for production and creation [6]. In a broad sense, the digital economy refers to a series of economic activities that promote economic efficiency and the optimization of the industrial structure by changing the original method of resource allocation, investing in the production factors of data resources, and taking the Internet platform and network clients as the main carriers to achieve a high degree of integration between the market and the government [7,8]. There are two general approaches to measuring the digital economy. One method is the national economic accounting approach, which evaluates the scale of the digital economy by calculating its total output or added value [9,10]. However, due to the limited availability and completeness of data, it often becomes necessary to rely on other means of estimation. The other method is the index method. First, the development level of the digital economy is evaluated from two perspectives: digital industrialization and industrial digitization [11,12]. Then, the measurement of the digital economy’s development level is extended to three major systems: digital infrastructure, digital industry, and digital integration [6,13]. Finally, based on the previous research, the measurement dimensions of the digital economy’s development level are further increased and comprehensively evaluated in four dimensions: the digital industry, digital users, digital platforms, and digital innovation [14,15]. This index method integrates the various forms of impact activities brought about via digital technology into a unified evaluation system, and it could obtain more systematic and comprehensive indicators. In addition, some scholars have established quasi-natural experiments based on low-carbon city pilot policies, using a double-difference model to quantify the development of the digital economy, and they have evaluated policy effectiveness [16]. The influencing factors of digital economy development primarily encompass external factors such as new infrastructure and associated institutional development alongside intrinsic factors such as industrial digitization, the digital industry, and the allocation of data factors [17].

Green innovation efficiency refers to the maximization of technological innovation and economic value achieved by enterprises in the innovation process, guided by environmental friendliness and resource conservation. It is a complement and an improvement to traditional innovation efficiency. Numerous studies have shown that, although the research background and perspectives of green innovation efficiency are different, they integrate ecological, economic, and innovative considerations [18,19]. Measurements of green innovation efficiency are mainly categorized into non-parametric and parametric methods. The parametric method involves a calculation using the setting function and efficiency term [20]. The non-parametric method employs an analysis (DEA) and its improved forms, including the specific models EBM-DEA and SBM-DEA [21,22]. Related studies have also been conducted on two-stage green innovation efficiency from the perspective of the innovation value chain [23]. Regarding the influencing factors of green innovation efficiency, starting from the macro environment, studies have mainly focused on the influence of regional economic development [24], the science and technology innovation environment [25], environmental regulation, and government support [26]. Starting from industrial transformation, the influencing factors are mainly concentrated on the scale of the industry and the upgrading of industrial structure [27]; starting from micro-enterprises, factors such as R&D intensity, the strengthening of scientific research investments, and improvements to the greenness of researchers all contribute to the greenness of green innovation [28,29]. The conclusions of studies investigating the relevant influencing factors differ due to the different research methods and research regions selected.

The current research examining the impact of the digital economy on green innovation efficiency primarily focuses on three perspectives: innovation, the environment, and the economy. From the perspective of innovation, the digital economy can enhance the quality of enterprises’ innovative activities by introducing digital technologies such as 5G, thereby promoting an improvement in green technology innovation efficiency [30]. Furthermore, the digital economy promotes better coordination among different innovation entities and elevates their technological innovation capabilities, leading to further improvements in green innovation efficiency [31]. From the perspective of the environment, the digital economy has played a positive role in promoting green innovation efficiency. On one hand, as the digital economy continues to grow, the development structure of green finance is becoming optimized, providing a broader development space for green industries. This has not only promoted the sustainable development of green industries but also laid a solid foundation for green innovation [32]. On the other hand, as enterprises increasingly adopt digital technologies and innovative advancements continue, enterprises can effectively mitigate environmental pollution and refine energy consumption patterns. This means that promoting economic development can also reduce negative impacts on the environment and achieve green and low-carbon development goals [33,34]. From the perspective of economics, green development plays a crucial role in driving high-quality economic development. Firstly, the rise of the digital economy has boosted innovation levels, making knowledge spillover and interactions within innovation networks faster and cheaper [31]. This has injected new momentum into technological advancement, thereby enhancing green innovation efficiency. Secondly, a digital transformation is crucial for establishing a green patent priority review system, which not only raises the economic returns of green patents for businesses but also sparks their enthusiasm for green innovation, further driving up green innovation efficiency [35]. Lastly, the further growth of the digital economy has enabled stronger cross-regional interactions and collaborations, leading to the creation of a circular economy system, which, in turn, has boosted regional green development and overall green innovation efficiency [36].

In summary, despite scholars having explored digital economy development and green innovation efficiency from diverse viewpoints, certain limitations persist. Firstly, few studies have studied or analyzed digital economy development and green innovation efficiency under a unified theoretical framework. The impact of digital economy development on green innovation efficiency and its mechanism of action still remains uncharted territory. The impact of digital economy development on two-stage green innovation efficiency is not yet known. Secondly, in the analysis of regional heterogeneity, scholars usually choose to analyze geographical location heterogeneity, ignoring the heterogeneity of the development mode, which may distort the role of the digital economy and produce a heterogeneity analysis that does not meet expectations. The distinctive contribution of this study lies in the following: Firstly, taking the perspective of a two-stage innovation value chain as its entry point, a more systematic theoretical analysis framework was constructed. Additionally, this study systematically analyzed and tested the effect of digital economy development on two-stage green innovation efficiency. Secondly, taking spatial factors and heterogeneity into consideration, this study constructed a spatial Durbin model (SDM) to explore the spatial spillover effect of digital economic development on the efficiency of two-stage green innovation, examining the variations in this effect across different geographic locations and development modes. Finally, this study further extended the moderating effects achieved via intellectual property protection and data factor endowment.

2. Theoretical Analysis and Research Hypotheses

2.1. Mechanism Analysis

Green innovation encompasses the following closely interconnected subprocesses: green research and development (R&D), green production, and green transformation. Based on the perspective of factor inputs and outputs, scholars have divided technological innovation into two stages: technology R&D and the transformation of results for an evaluation of efficiency [37]. From the perspective of an innovation value chain, this study constructed an innovation efficiency evaluation system that included the green initial inputs, green intermediate outputs, and green final outputs. The green innovation process is divided into two stages: the first stage focuses on green development, investing in green innovation elements and green patents as inputs and outputs for green innovation technology research and development, and the second stage is market-oriented, applying patented technologies to production and sales processes in order to achieve economic benefits and reduce pollution emissions and converting green innovation achievements.

(1)

Green innovation technology research and development phase.

The strong expansion of digital technology and new infrastructure underlies the advancements in the digital economy. The rise of digital technology has not only driven profound shifts in the technological paradigm but also effectively amplified the impact of innovation agglomeration. Enhancing the supporting capacity of green innovation resources improves resource allocation efficiency, ultimately fueling the output of green technology innovations. Meanwhile, the popularity of new infrastructure has accelerated the circulation and sharing of key elements, such as funds, digital technology, and R&D talents, among different systems and research themes. This has led to a positive multiplier effect on the efficiency of other production factors, further unleashing new driving forces for economic development. It has enabled the efficient matching of innovation resources, providing a feasible path for enhancing the efficiency of green innovation technology research and development. Meanwhile, the popularization of new infrastructure has accelerated the circulation and sharing of key elements such as funds, digital technology, and R&D talents among different systems and research themes. This has led to a positive multiplier effect on the efficiency of other production factors, further releasing new drivers for economic development. Compared with other economic development modes, advancements in the digital economy stand out for their efficiency, coordination, cost-effectiveness, and intensity of knowledge utilization. The growth of the digital economy can stimulate scientific and technological research and development by leveraging economies of scale and scope [38]. The development of the digital economy has had numerous positive impacts on the stage of green innovation technology research and development. Firstly, the development of the digital economy can stimulate collaborative innovation within the industry, allow a high degree of integration of capital, talent, and innovation projects to be realized, break circulation barriers, and ultimately boost green innovation efficiency [39]. Secondly, researchers and developers can leverage convenient and speedy information platforms to expedite the transformation and upgrading of industrial structures, facilitate the establishment of demonstration zones, and expand the digital economy across diverse sectors, ultimately enhancing the efficiency of green innovation [40]. The development of the digital economy effectively lowers the production and circulation costs among various sectors, shortens the R&D cycle and transaction time, and acts as a catalyst for enhancing green innovation efficiency [41]. Green innovation, in contrast to traditional methods, enhances organizational management techniques, thereby reducing the operating costs of enterprises and boosting investments in sustainable innovation. This leads to a harmonious balance between regional economic, environmental, and social benefits. The improvement of green technology research and development capabilities in this region can improve the speed at which regional enterprises identify new technologies and respond to market demand and reduce operating costs, research costs, and development trial and error costs, thereby broadening sales channels and forming a spillover trend for green technology research and development in neighboring regions.

(2)

Green innovation achievement transformation phase.

The digital economy can be distinguished due to its intense technological focus and high permeability. In response to the accelerated integration and infiltration of digital technology with the real economy, forms of businesses and new models are continuously emerging, reshaping the regional production form and the quality of economic development. This has laid a solid foundation for enhancing the industrial structure’s efficiency and expanding the benefits of network platforms [42]. The participation of innovative technology in production and operation determines the transformation efficiency of green innovation achievements. Firstly, the development of the digital economy enhances enterprises’ production, management, and sales efficiency, endows traditional enterprises with innovative capabilities, and boosts the economic and environmental product benefits, ultimately improving the transformation efficiency of green innovation outcomes. The development of the digital economy enhances enterprises’ production, management efficiency, and sales efficiency, endows traditional enterprises with an innovation ability, increases the economic and environmental benefits of products, and improves the transformation efficiency of green innovation. Secondly, the advancement of digital technology has facilitated the emergence of digital finance, offering a convenient platform for funding the supply and demand for green innovation projects. This effectively solves the problem of the shortage of innovation financing in provinces, and it contributes to the formation of more competitive core technologies that, in turn, improve the transformation efficiency of green innovation achievements [43]. Moreover, the speedy expansion of the digital economy has led to the continuous enhancement and rapid rise of the open innovation system and the new ecology of platform industrialization, which can effectively accelerate the free flow of green innovation results among regions. The growth of the digital industry offers a technical platform for cross-regional innovation cooperation for various innovation subjects, thus comprehensively broadening the green innovation boundary, realizing the migration of core technology and green innovation knowledge, and causing a spatial spillover effect [44]. Therefore, this study proposed the following postulate:

Hypothesis H1 (H1). 

Digital economy development can effectively enhance the efficiency of two-stage green innovation, and it can influence the efficiency of two-stage green innovation in neighboring regions via spillover effects.

2.2. Analysis of the Moderating Mechanism

Intellectual property protection serves as an important guarantee for a series of innovative activities, mitigating regional R&D risks, facilitating the transformation of innovation results, and elevating the efficiency of technology-transfer transactions [45]. Data factor endowment, by virtue of its own characteristics of high permeability, high value, and large-scale transformation, can enhance the quality of factor inputs by radiating to other factors of production, ultimately leading to improved efficiency in green innovation.

Regarding the protection of intellectual property rights, during the research and development stage of green innovation technology, an imperfect intellectual property rights system may arise. This can potentially result in the plagiarism of research and development activities, leading to financial losses for innovators. As a consequence, such losses may discourage further innovation and, subsequently, decrease the overall output of innovative results. The flow of data elements not only enhances the efficiency of green innovation technology R&D but also facilitates the above-mentioned undesirable consequences. Therefore, strengthening the intensity of intellectual property protection is beneficial for multiple reasons. It can promote improvements in the regional green innovation capacity and also reduce the risk of green technology and knowledge infringement, thus positively affecting the green innovation environment [46]. Innovative subjects can obtain additional income through authorized patents, which can help motivate them to research and develop green innovation technologies. Consequently, this promotes the steady enhancement of green innovation efficiency. In the transformation stage of green innovation, a perfect intellectual property protection system can help create a good financing environment. This environment can effectively boost investment in green innovation technologies and enhance the conversion rate of high-tech results. Technology transformation tends to be more market-oriented, forming a virtuous cycle and further improving the efficiency of green innovation. In the stage of green innovation technology R&D, companies can drive digital transformation through the endowment of data elements. By employing data technology, they can manage resources more effectively, change the traditional approach to resource utilization, and consequently enhance their resource utilization efficiency and generate efficient green technology innovations. In the process of green innovation achievement transformations, boosting the advantage of data elements supports data mining and information analyses. The “creative value” gained from mining digital information with digital technology has become a new kinetic energy to enhance the efficiency of regional green innovation. This, in turn, has further accelerated the advancement of green technology. Therefore, this study proposes the following postulate:

Hypothesis H2 (H2). 

Intellectual property protection and data factor endowment can positively moderate the impact of digital economy development on the efficiency of green innovation in both phases.

3. Methodology and Materials

3.1. Model Construction

Based on the aforementioned theoretical analysis, there may exist a spatial correlation between green innovation efficiency and digital economy development in two stages. Specifically, the digital economy development in a certain region may influence the green innovation efficiency of neighboring regions. From the perspective of the innovation value chain, which encompasses “green initial inputs—green intermediate outputs—green final outputs”, in this study, a spatial Durbin model (SDM) was constructed to investigate the spatial spillover impact of digital economy development on green innovation efficiency across its two stages. The SDM utilizes the distance and association among geospatial data to predict the mutual influence among variables, thereby enhancing the comprehension of the relationships and impacts within geospatial data. The specific model is set as follows:

$${gier}_{it}={\rho Wgier}_{it}+{\alpha }_1{dige}_{it}+{\alpha }_2{Wdige}_{it}+{\alpha }_3{x}_{it}+{\alpha }_4{Wx}_{it}+{\mu }_i+{\phi }_t+{\epsilon }_{it}$$

(1)

$${giet}_{it}={\rho Wgiet}_{it}+{\beta }_1{dige}_{it}+{\beta }_2{Wdige}_{it}+{\beta }_3{x}_{it}+{\beta }_4{Wx}_{it}+{\mu }_i+{\phi }_t+{\epsilon }_{it}$$

(2)

$i$ denotes the province; $t$ denotes the year; ${gier}_{it}$ denotes the R&D efficiency of the green innovation technology; ${giet}_{it}$ denotes the transformation efficiency of the green innovation achievements; ${dige}_{it}$ denotes the digital economy; $x_{it}$ represents the relevant control variables; $W$ represents the spatial weight matrix, with a neighbor matrix of 30 provinces being selected in this study; ρ denotes the spatial autoregressive coefficients; ${\mu }_i$ and ${\phi }_t$ denote the individual and time fixed effects, respectively; and ${\epsilon }_{it}$ represents the random disturbance term.

This study aimed to investigate the moderating effects of intellectual property protection (ipr) and data factor endowment (data). To achieve this, interaction terms between $dige\times ipr$ and $dige\times data$ were introduced for the study’s empirical research. The model was established as described in the following sections.

$$\begin{array}{ll}{gier}_{it}={\rho Wgier}_{it}& +{\alpha }_1{dige}_{it}+{\alpha }_2{Wdige}_{it}+{\alpha }_3{adj}_{it}+{\alpha }_4{Wadj}_{it}\\ & +{\alpha }_5{dige}_{it}\times {adj}_{it}+{\alpha }_6{Wdige}_{it}\times {adj}_{it}+{\alpha }_7{x}_{it}+{\alpha }_8{Wx}_{it}+{\mu }_i\\ & +{\phi }_t+{\epsilon }_{it}\end{array}$$

(3)

$$\begin{array}{ll}{giet}_{it}={\rho Wgiet}_{it}& +{\beta }_1{dige}_{it}+{\beta }_2{Wdige}_{it}+{\beta }_3{adj}_{it}+{\beta }_4{Wadj}_{it}\\ & +{\beta }_5{dige}_{it}\times {adj}_{it}+{\beta }_6{Wdige}_{it}\times {adj}_{it}+{\beta }_7{x}_{it}+{\beta }_8{Wx}_{it}+{\mu }_i\\ & +{\phi }_t+{\epsilon }_{it}\end{array}$$

(4)

where ${adj}_{it}$ represents the moderating variables, including ${ipr}_{it}$ and ${data}_{it}$, which denote intellectual property protection and data factor endowment.

3.2. Measurement and Description of Variables

(1) Explained variables. Green innovation activities, viewed from the perspective of a two-stage innovation value chain, can be categorized into two distinct stages: the research and development stage and the achievement transformation stage. The first stage involves using green innovation factors as the inputs and green innovation as the output, and the second stage involves selling the green innovation output through a series of production and sales so as to achieve revenue and reduce pollution emissions. Therefore, drawing inspiration from the research conducted by Huang et al., in this study, the super-NSBM model, which incorporates unexpected outputs, was chosen to assess the two-stage green innovation efficiency [47]. The super-NSBM model is an efficiency evaluation model based on data envelopment analysis (DEA), which is primarily used to evaluate the efficiency of DMUs. The formula for the calculation is as follows:

$${\rho }_0^{*}=min\frac{\sum _{k=1}^K{w}^k\left[1+\frac{1}{m_k}\left(\sum _{i=1}^{m_k}\frac{s_i^{k-}}{x_{i0}^k}\right)\right]}{\sum _{k=1}^K{w}^k\left[1-\frac{1}{v_{1k}+v_{2k}}\left(\sum _{r=1}^{v_{1k}}\frac{s_r^{gk}}{y_{r0}^{gk}}+\sum _{r=1}^{v_{2k}}\frac{s_r^{bk}}{y_{ro}^{bk}}\right)\right]}$$

(5)

$$s.t.\left\{\begin{array}{l}\sum _{j=1,\ne 0}^n{x}_{ij}^k{\lambda }_j^k+s_i^{k-}={\theta }^k{x}_0^k,i=1,..,m_k,k=1,..,k\\ \sum _{j=1,\ne 0}^n{y}_{ij}^{gk}{\lambda }_j^k+s_i^{gk}={\phi }^k{y}_0^{gk},r=1,..,s_k,k=1,..,k\\ \sum _{j=1,\ne 0}^n{x}_{ij}^{bk}{\lambda }_j^k+s_i^{bk}={\delta }^k{y}_0^{bk},r=1,..,s_k,k=1,..,k\\ \epsilon \le 1-\frac{1}{v_{1k}+v_{2k}}\left(\sum _{r=1}^{v_{1k}}\frac{s_r^{gk}}{y_{r0}^{gk}}+\sum _{r=1}^{v_{2k}}\frac{s_r^{bk}}{y_{ro}^{bk}}\right)\\ Z^{\left(k,h\right)}{\lambda }^h=Z^{\left(k,h\right)}{\lambda }^k,\sum _{j=1,\ne 0}^N{\lambda }_j^k=\sum _{k=1}^K{w}^k=1\\ {\lambda }^k\ge 0,s^{k-}\ge 0,s^{gk}\ge 0,s^{bk}\ge 0,w^k\ge 0\end{array}\right.$$

(6)

In Equations (5) and (6), $m_k$ represents the number of inputs in the first k stages. $v_k$ denotes the number of outputs in the same stages; ${\phi }_k$ indicates the number of intermediate indicators. $x$, $y$, and $z$ denote the inputs, outputs, and intermediate outputs, respectively, and $s^{k-}$, $s^{gk}$, and $s^{bk}$ are the slack variables of the input, expected output, and unexpected output, respectively. This study focused on a two-stage efficiency evaluation, so k = 2, and the weights of the two stages were set to the same value. In the first stage, drawing on Liu et al.’s study [27], the indicators of capital, labor, and resource inputs were selected as follows: science and technology expenditures and environmental governance expenditures; urban unit employment engaged in scientific research and technological services; and urban unit employment engaged in water conservancy, the environment, and public facility management. Additionally, the amount of electricity supplied was also included. The number of green innovation patent grants was chosen as the output indicator. In the second stage, the first stage’s input and output indicators were jointly utilized as the input indicators of the second stage. The GDP served as the desired output for the second stage. Industrial wastewater, industrial sulfur dioxide, and smoke (dust) emissions were undesired outputs, reflecting the economic and environmental outputs of the second stage, respectively.

(2) Explanatory variables. Digital economy development, which served as the core explanatory variable in this study, refers to economic activity centered around a modern information network and digital information as a key factor of production, and structural optimization and efficiency were achieved through information and communication technology. The application of new digital-type technologies such as the Internet, cloud computing, big data, the Internet of Things, and financial technology has revolutionized socialization by enhancing information collection, storage, analysis, and sharing. Drawing from existing research and the practice of Lu et al. [48], this study comprehensively constructed a digital economic development index by incorporating the digital transaction index system, ensuring the availability and scientific validity of relevant data, and utilizing Internet development and digital financial inclusion as measurement benchmarks, as shown in

Table 1. Composite index of digital economy development.

	First-Level Index	Second-Level Index	Specific Indicators
	Internet development	Internet penetration	Internet users per 100 population
		Cell phone penetration rate	Cell phone subscribers per 100 population
Composite index of digital economy development		Internet industry output value	Total value of telecommunications services per capita
		Internet professional	Information transmission, software, and information technology services as a percentage of
	Digital inclusive finance	Digital inclusive finance index	Peking University Digital Inclusive Finance Index

. The Internet, serving as the backbone of digital economy development, was assessed across four dimensions to measure the level of its inter-provincial development. The digital financial inclusion index can reflect the digital economy from the breadth, depth, and digitization degree, which is an important embodiment of its development. Finally, a comprehensive index of digital economy development was obtained by measuring the five indicators above while comprehensively using the entropy method.

(3) Moderating variables. The level of intellectual property protection varies according to the differences in innovation systems across regions, which is crucial to transforming innovation into economic and environmental benefits. The level of intellectual property protection is mainly realized by penalizing infringement, creating a good innovation atmosphere, reducing innovation risk, and promoting innovation transactions. Therefore, it is worthwhile to study the impact of intellectual property protection in the development of the digital economy on the efficiency of two-stage green innovation. In this study, we referred to the study by Chen, W, et al. [49] and measured the level of intellectual property protection to select the ratio of technology market turnover to the regional gross domestic product. Enhancing data factor endowment can coordinate the ratio of factor inputs, improve the quality of factor inputs, create a convenient channel for knowledge flow, and strengthen the level of regional technological innovation through effective imitation and learning so as to further enhance the efficiency of green innovation. This study started from the two aspects of a data-generating body and a flow carrier, used the total number of Internet broadband access users and mobile Internet users to measure the regional user scale, used the scale of annual mobile Internet traffic usage to measure the regional traffic scale, and obtained a composite index through the entropy method for the user scale and traffic scale as a measurement index of data factor endowment.

(4) Control variables. Based on the existing research [8], this study further integrated foreign direct investment (fdi), the degree of government intervention (gov), the marketization level (mar), and environmental regulation (env) as control variables. Among them, an increase in foreign direct investment can promote technological innovation, as evidenced by the proportion of the FDI to the GDP in each region. The degree of government intervention, represented by the ratio of budgetary expenditures to the GDP of each province, serves as a macro-control tool used by the national government to regulate and control the socio-economy. The level of marketization, captured in the marketization index, reflects the significance of market forces in resource allocation, with higher marketization leading to less distortion and a more rational allocation of resources. Environmental regulation is a short-term, rapid reduction in the pollution emissions of mandatory measures, which may have adverse effects on regional undesirable outputs, as indicated in the share of the local financial expenditure on environmental protection in the GDP of each region.

3.3. Data Sources and Descriptive Statistics

Considering the availability of data related to the digital economy, for this study, data from 30 provinces in China (excluding Tibet, Hong Kong, Macao, and Taiwan) from 2011 to 2021 were selected as the research sample. The data of relevant indicators were obtained from the China Statistical Yearbook, the China Science and Technology Statistical Yearbook, the Peking University Digital Financial Inclusion Index, and the statistical yearbooks of each province for previous years. Missing individual data were primarily supplemented through linear interpolation.

Table 2. Descriptive statistics of variables.

Variable Type	Variable Symbol	Obs	Min	Max	Mean	Std
Explained variable	gier	330	0.018	1.263	0.269	0.206
	giet	330	0.011	1.297	0.191	0.202
Explanatory variables	dige	330	0.008	0.954	0.215	0.151
Moderator variable	ipr	330	0.001	0.191	0.017	0.029
	data	330	0.001	1.000	0.136	0.144
Control variable	fdi	330	0.001	0.121	0.018	0.018
	gov	330	0.107	0.758	0.266	0.115
	mar	330	3.360	12.922	7.988	1.977
	env	330	0.002	0.043	0.008	0.006

presents the descriptive statistics for each variable.

4. Empirical Analysis

4.1. Spatial Correlation Test and Spatial Econometric Model Selection Test

As shown in

Table 3. Global spatial correlation analysis of the two-stage green innovation efficiency and digital economy development.

Year	Gier		Giet		Dige
	Moran’s I	Z-Value	Moran’s I	Z-Value	Moran’s I	Z-Value
2011	0.304 ***	2.826	0.297 ***	2.801	0.146 **	1.811
2012	0.342 ***	3.151	0.335 ***	3.135	0.149 **	1.869
2013	0.322 ***	2.980	0.309 ***	3.109	0.124 *	1.552
2014	0.300 ***	2.922	0.279 ***	2.935	0.106 *	1.417
2015	0.397 ***	3.691	0.349 ***	3.547	0.108 *	1.441
2016	0.342 ***	3.393	0.266 ***	3.334	0.120 *	1.575
2017	0.419 ***	3.750	0.419 ***	3.879	0.104 *	1.391
2018	0.435 ***	3.883	0.410 ***	3.830	0.095	1.238
2019	0.420 ***	3.755	0.413 ***	3.773	0.090	1.185
2020	0.196 **	1.917	0.048	0.709	0.112 *	1.389
2021	0.440 ***	3.925	0.238 **	2.230	0.165 **	2.054

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

, the Moran’s I indices for two-stage green innovation efficiency in all years except 2020 were positive, and all passed the significance level test of at least 5%. This result supports Zhang et al.’s finding that the Moran’s I indices for two-stage green innovation efficiency were significantly positive from 2009 to 2020, passing a 10% significance test [50]. This indicates that China’s regional green innovation has generated a “spatial diffusion” effect, showcasing a positive interaction among regions in terms of green innovation R&D efficiency and green achievement transformation efficiency. The Moran’s I indices of digital economy development were significantly positive except for 2018 and 2019. This result corroborates the finding that Moran’s I for the digital economy and industrial green innovation efficiency was significantly positive at a 5% statistical level from 2005 to 2019 [18], indicating a pronounced spatial correlation in digital economy development.

This study followed a testing approach from specific to general. Before estimating the spatial econometric model, an LM test, Wald test, LR test, and Hausman test were conducted for optimal model selection. The results are shown in

Table 4. Spatial econometric model selection tests.

Test Methods	Gier		Giet
	Statistical Value	p-Value	Statistical Value	p-Value
LM spatial lag	10.939	0.001	19.208	0.000
R-LM spatial lag	8.342	0.004	7.462	0.006
LM spatial error	99.826	0.000	77.896	0.000
R-LM spatial error	97.229	0.000	66.150	0.000
Wald spatial lag	15.73	0.003	13.14	0.010
LR spatial lag	28.21	0.000	26.17	0.000
Wald spatial error	23.30	0.000	24.54	0.000
LR spatial error	22.69	0.000	24.01	0.000

. The LM test, robust LM test, the LR test, and the Wald test all passed the 1% significance level test. Meanwhile, the Hausman value of 68.78 was significant at the 1% level, and the spatial autoregressive coefficient ρ was significantly non-zero, indicating that the spatial Durbin model with fixed effects was the optimal model. Therefore, SDM with fixed effects was selected for the empirical research in this study.

4.2. Analysis of Basic Estimation Results

Table 5. Basic estimation results of digital economy development affecting two-stage green innovation efficiency.

Variable	Gier			Giet
	SAR	SEM	SDM	SAR	SEM	SDM
dige	0.903 *** (11.47)	0.861 *** (10.92)	0.818 *** (9.87)	0.966 *** (4.18)	1.054 *** (4.58)	0.843 *** (8.53)
fdi	−0.371 (−1.03)	−0.185 (−0.51)	−0.437 (−1.14)	−0.501 *** (−3.21)	−0.528 *** (−3.39)	−0.596 (−1.29)
gov	−0.511 *** (−4.22)	−0.547 *** (−4.32)	−0.662 *** (−4.72)	−0.038 *** (−3.29)	−0.040 *** (−3.44)	−0.474 *** (−2.81)
mar	0.029 *** (5.58)	0.033 *** (6.66)	0.034 *** (6.59)	−0.774 *** (−3.97)	−0.774 *** (−3.93)	0.032 *** (5.18)
env	5.047 *** (2.37)	4.571 ** (2.19)	6.539 *** (2.79)	0.362 *** (4.28)	0.341 *** (4.09)	4.990 * (1.77)
W * dige			0.328 * (1.67)			0.588 ** (2.55)
ρ	0.216 *** (3.99)		0.312 *** (4.67)	0.258 *** (2.75)		0.190 *** (2.85)
λ		0.330 *** (4.82)			0.342 *** (3.33)
R-sq	0.842	0.849	0.880	0.326	0.323	0.882
N	330	330	330	330	330	330

Note: ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively; values in parentheses indicate z-values.

reports the benchmark regression results of the SAR model, SEM, and SDM under fixed effects, which shows that the estimated coefficients of digital economy development on two-stage green innovation efficiency in the local region were significantly positive at the 1% level. This finding is similar to that of Li et al. (2022), who discovered that a higher level of digital economy development is associated with improved green innovation efficiency [31]. Meanwhile, from the perspective of the goodness of fit, the SDM exhibited a higher degree of fit compared to the SAR model and SEM, thus confirming the findings of spatial econometric model selection discussed in the previous section. Therefore, in this study, further effect decomposition of the SDM under fixed effects was performed to examine its spatial spillover effect.

As shown in

Table 6. Regression results of the spatial effect of digital economy development on the efficiency of green innovation in two stages.

Variable	Gier			Giet
	Direct Effect	Indirect Effect	Total Effect	Direct Effect	Indirect Effect	Total Effect
dige	0.869 *** (9.86)	0.799 *** (3.12)	1.668 *** (5.54)	0.881 *** (8.64)	0.889 *** (3.39)	1.770 *** (5.79)
fdi	−0.365 (−0.93)	1.427 (1.21)	1.062 (0.78)	−0.561 (−1.24)	1.333 (1.11)	0.773 (0.56)
gov	−0.630 *** (−4.70)	0.287 (1.21)	−0.343 (−1.04)	−0.459 *** (−2.87)	−0.074 (−0.24)	−0.533 (−1.60)
mar	0.034 *** (6.28)	0.001 (0.08)	0.035 ** (2.23)	0.032 *** (4.88)	−0.027 ** (−1.98)	0.004 (0.27)
env	6.472 *** (2.84)	2.449 (0.43)	8.921 (1.28)	4.862 * (1.83)	3.400 (0.58)	8.262 (1.16)

Note: ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively; values in parentheses indicate z-values.

, the spatial effect regression results revealed a significant positive impact of digital economy development on both the direct and spillover effects of two-stage green innovation efficiency. This indicates that the development of the digital economy not only promotes two-stage green innovation efficiency in the local region but also generates positive spillover effects in surrounding areas. This is consistent with the findings of Li et al. (2022), Li et al. (2023), and Liu et al. (2024), who discovered that the digital economy has a spatial spillover effect on green innovation efficiency [18,31]. The possible reasons are as follows. Firstly, digital economy development can break down barriers to the circulation of production factors during the stage of green innovation technology R&D, facilitating the free flow of technological innovation factors and helping these factors combine with leading production sectors. This, in turn, further enhances the efficiency of green innovation research and development within the local region. Simultaneously, the digital economy can leverage its advantages in the Internet and information services to capture the long-tail effects of digital platforms. By coordinating and collaborating with neighboring regions, it can promote the efficiency of green innovation R&D in those areas. Digital economy development can accelerate the upgrading and transformation of traditional industries in the transformation stage of green innovation achievements, enhance the allocation efficiency of talents and capital and innovation projects, provide a large amount of sustainable funding for the transformation of technological achievements, and then promote an increase in the rate of green innovation achievement in the local area. In addition, coupled with the support of multiple information platforms and information media, the digital economy can rapidly penetrate all industries, facilitate the circulation and dissemination of green technologies and production factors among regions and industries, increase the economic benefits of green innovation achievements, and further promote the sustainable growth of the local economy. Accordingly, Hypothesis H1was validated.

In terms of the control variables, the direct and spillover effects of foreign direct investment (fdi) on two-stage green innovation efficiency were found to be insignificant, indicating that fdi does not significantly impact the two-stage green innovation efficiency in either the local area or the neighboring regions. The degree of government intervention had a significantly negative direct effect on both stages, while the spillover effect was positive but not statistically significant. This suggests that excessive intervention may hinder the progress in both areas within this region. The level of marketization had a positive direct effect on both stages that was statistically significant at the 1% level. However, its spillover effect on the transformation efficiency of green innovation was significantly negative, suggesting that the marketization level promotes both stages in this region but hinders the green transformation efficiency in the neighboring regions. Environmental regulation has both positive direct and positive indirect effects on the two stages, indicating that appropriate environmental regulation can contribute to the promotion of these stages in both the local region and neighboring regions.

4.3. Robustness Tests and Endogeneity Tests

First, this study replaced the two stages of green innovation efficiency and re-examined the impact of the digital economy on them. Potential time delays in the green innovation process, from initial inputs to intermediate outputs and then to the final outputs, were taken into account. This study set the green innovation initial inputs to period T, the green intermediate outputs to period T + 1, and the green final outputs to period T + 2, and it re-measured the two-stage green innovation efficiency through the super-NSBM model. Columns (1) and (4) of

Table 7. Robustness test and endogeneity tests.

Variable	Gier			Giet
	Substitution of Explained Variables (1)	Transformed Nested Matrix (2)	Explanatory Variables Lagged One Period (3)	Substitution of Explained Variables (4)	Transformed Nested Matrix (5)	Explanatory Variables Lagged One Period (6)
dige	0.937 *** (12.79)	1.023 *** (11.46)	0.846 *** (9.54)	1.011 *** (11.62)	1.043 *** (10.04)	0.895 *** (8.39)
W * dige	0.318 ** (2.29)	1.629 *** (4.42)	0.317 (1.48)	0.380 (0.70)	1.764 ** (4.25)	0.458 * (1.77)
Control	Yes	Yes	Yes	Yes	Yes	Yes
ρ	0.338 *** (3.56)	0.356 ** (2.33)	0.323 *** (4.70)	0.419 *** (4.74)	0.350 ** (2.28)	0.206 *** (2.94)
R-sq	0.879	0.850	0.886	0.871	0.857	0.892
N	330	330	330	330	330	330

Note: ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively; values in parentheses indicate z-values.

demonstrate a significantly positive impact of digital economy development on the two-stage green innovation efficiency. Second, the spatial weights were replaced and re-estimated. To simultaneously assess the influence of geographic and economic factors on the two-stage green innovation efficiency, this study selected the economic geography nested matrix for robustness testing. Columns (2) and (5) of Table 7 demonstrate that the digital economy can still positively impact the two-stage green innovation efficiency by changing the economic geography for the nested matrix. Thirdly, considering the issue of endogeneity, this study introduced a lagged version of the explanatory variable as an instrumental variable, mitigating the concern of reverse causality between the explained variable and the core explanatory variable. The data were then re-estimated accordingly. Columns (3) and (6) in Table 7 demonstrate a significant impact of the digital economy on both stages of green innovation efficiency at the 1% level, proving the robustness of the aforementioned research results.

4.4. Heterogeneity Analysis

(1)

Geographic location heterogeneity.

Differences in the regional economic development stages and uneven resource endowment may result in heterogeneous impacts of the digital economy on the two-stage green innovation efficiency across different regions. Hence, this study categorized the 30 provinces into three regions: eastern, central, and western—based on the three regional classification criteria established by the National Bureau of Statistics. The results presented in

Table 8. Results of heterogeneity analysis for different geographic locations.

Variable	Gier			Giet
	Eastern Region	Central Region	Western Region	Eastern Region	Central Region	Western Region
Direct effect	0.909 *** (6.66)	−0.187 (−0.70)	−0.269 (−0.72)	0.821 *** (5.50)	0.018 (0.009)	−0.301 (−0.81)
Indirect effect	0.821 *** (2.66)	0.696 ** (2.46)	−0.874 (−1.22)	0.997 *** (3.29)	0.320 (1.47)	−1.170 * (−1.73)
Aggregate effect	1.730 *** (4.22)	0.508 (1.30)	−1.143 (−1.22)	1.817 *** (4.73)	0.337 (1.06)	−1.471 * (−1.67)
Control	Yes	Yes	Yes	Yes	Yes	Yes
ρ	0.129 (1.35)	0.520 *** (5.16)	0.119 (0.90)	0.128 (1.34)	0.403 *** (3.92)	0.197 (1.48)
R-sq	0.886	0.941	0.790	0.901	0.941	0.692
Log L	96.831	198.859	135.981	71.281	226.016	134.496
N	121	88	121	121	88	121

Note: ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively; values in parentheses indicate z-values.

indicate that, in the eastern region, both the direct and spillover effects of the two-stage green innovation efficiency were positive, and all the effects were statistically significant at the 1% level. The probable explanations for this suggest that the eastern region’s higher level of economic development, more complete industrial chain, and stronger R&D capabilities contribute to its advantages. In addition, the eastern region boasts superior R&D capabilities in green innovation technologies and a higher rate of green innovation achievement transformation compared to the central and western regions with traditional industries. Concurrently, the rapid digitization in the eastern region, coupled with the widespread use of digital technologies, may bring about a radiation-driven effect, thereby enhancing the green innovation efficiency of neighboring regions. The significance level test rejected the direct effect of digital economy development on green innovation efficiency in both stages for the central and western regions, and there was no spatial spillover effect in the western region at the transformation stage of green innovation achievement. The possible reason for this suggestion that the digital infrastructure in the central and western regions is insufficient. The industries are still primarily led by traditional industries, with low levels of digitalization; the green production industry is lacking, and the green inputs cannot be well transformed into green outputs. Concurrently, the spillover effect of technological innovation is constrained due to the unequal allocation of innovation resources and a scarcity of innovation talents.

(2)

Development approach heterogeneity.

By recognizing significant variations in development patterns, this study classified the regions that favor technological innovation towards a resource bias and are dominated by resource-based industries as resource-based regions, as well as the regions that favor technological innovation towards advanced technology and are dominated by the tertiary industry and service-oriented manufacturing as non-resource-based regions. The aim was to further explore the heterogeneity issues resulting from different development patterns. By calculating the proportion of resource-based industries in the secondary industry and the employment rate of such industries within each region, the 30 provinces were finally classified into 14 resource-based provinces and 16 non-resource-based provinces. Tianjin, Hebei, Shanxi, Inner Mongolia, Heilongjiang, Jiangxi, Hainan, Guizhou, Yunnan, Shaanxi, Gansu, Qinghai, Ningxia, and Xinjiang were classified as resource-based provinces, while all the other provinces were classified as non-resource-based provinces.

Table 9. Results of heterogeneity analysis of different development approaches.

Variable	Gier		Giet
	Resource-Based Provinces	Non-Resource-Based Provinces	Resource-Based Provinces	Non-Resource-Based Provinces
Direct effect	0.844 *** (3.19)	0.620 *** (4.88)	1.007 *** (3.24)	0.633 ** (3.97)
Indirect effect	0.060 (0.27)	0.605 *** (2.85)	0.113 (0.41)	0.131 (0.64)
Aggregate effect	0.905 ** (2.11)	1.225 *** (4.61	1.119 ** (2.18)	0.764 *** (3.31)
Control	Yes	Yes	Yes	Yes
ρ	0.288 *** (2.96)	0.120 (1.23)	0.310 ** (3.56)	0.197 ** (2.18)
R-sq	0.513	0.948	0.401	0.938
N	154	176	154	176

Note: *** and ** indicate significance at the 1% and 5% levels, respectively; values in parentheses indicate z-values.

demonstrates that, for two-stage green innovation efficiency, the resource-based provinces exhibited a significantly positive direct effect, yet the spatial spillover effect on this efficiency was not significant. A plausible explanation lies in the fact that, in the process of resource-based economic development, material capital mostly has a strong asset specificity, resulting in a crowding-out effect on green innovation capital. This, in turn, hinders the full utilization of digital economic benefits and diminishes the radiation-driven impact on neighboring regions. Both the direct and spillover effects of non-resource-based provinces are significantly positive in the research and development stage of green innovation technology. Additionally, the direct effect is significantly positive in the stage of green innovation achievement transformation. A possible reason for this is that non-resource-based provinces possess good innovation technology and policy benefits, enabling them to gather green R&D talents and capital and fully leverage the advantages of the digital economy. While enhancing the two-stage green innovation efficiency in their own regions, these provinces can also play a role in promoting the efficiency of green innovation technology research and development in neighboring regions.

5. Expansion Analysis

5.1. Moderating Effects of Intellectual Property Protection

The interaction term between the development of the digital economy and the level of intellectual property protection (dige×ipr) had a notably beneficial direct effect on the green innovation efficiency of both stages, as demonstrated in

Table 10. Moderating effects of intellectual property protection.

Moderator Variable	Variable	Gier			Giet
		Direct Effect	Indirect Effect	Total Effect	Direct Effect	Indirect Effect	Total Effect
Intellectual property protection	dige	0.725 *** (6.01)	0.173 (0.50)	0. 898 ** (2.35)	0.669 *** (4.70)	0.419 (1.15)	1.084 ** (2.08)
	dige×ipr	1.494 * (1.81)	6.431 *** (3.33)	7.925 *** (3.68)	2.145 ** (2.19)	5.293 *** (2.72)	7.439 *** (3.53)
	ρ	0.293 *** (4.39)			0.169 ** (2.55)
	R-sq	0.905			0.915
	N	330	330	330	330	330	330

Note: ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively; values in parentheses indicate z-values.

. This suggests that the level of intellectual property protection positively mediates the relationship between the digital economy and green innovation efficiency. This is similar to the findings of Fen and Nie (2024), who reported that intellectual property protection plays a crucial role in moderating the effect of digitalization on technological innovation, enabling synchronized development between the two [51]. A possible reason for this is that continuously improving the level of intellectual property protection during the R&D stage of green innovation technology can effectively stimulate enterprise enthusiasm for independent innovation, enhance the expected returns of green innovation investments, and contribute to the formation of a new pattern of technology and knowledge sharing, thus radiating and driving the green development of neighboring regions. During the transformation of green innovation achievements, a strong level of intellectual property protection can bring additional patented benefits to enterprises, stimulating them to continuously conduct green innovation research and development. Consequently, this continuously enhances the efficiency of green innovation in the local area. However, due to the existence of a certain time lag between green investments and the transformation of green achievements, the spillover effect on neighboring regions is not significant.

5.2. Moderating Effects of Data Factor Endowment

The interaction term between digital economy development and data factor endowment (dige×data) had a notably beneficial direct effect on the green innovation efficiency of both stages, as demonstrated in

Table 11. Moderating effects of data factor endowment.

Moderator Variable	Variable	Gier			Giet
		Direct Effect	Indirect Effect	Aggregate Effect	Direct Effect	Direct Effect	Indirect Effect
Data factor endowment	dige	0.854 *** (9.55)	0.903 *** (3.29)	1.759 *** (5.57)	0.853 *** (8.26)	1.045 *** (3.73)	1.899 *** (5.95)
	dige×data	0.377 ** (2.10)	−0.107 (−0.28)	0.270 (0.88)	0.641 *** (3.00)	−0.187 (−0.46)	0.454 (1.05)
	ρ	0.303 *** (4.52)			0.180 *** (2.69)
	R-sq	0.882			0.890
	N	330	330	330	330	330	330

Note: *** and ** indicate significance at the 1% and 5% levels, respectively; values in parentheses indicate z-values.

. This suggests that data factor endowment positively mediates the relationship between digital economy development and green innovation efficiency. This finding aligns with the research of Evenett and Kelly (2002), who discovered that regulating the factor endowments can balance economic disparities and, subsequently, foster technological innovation [52]. A possible reason for this is that enhancing the endowment of data elements during the R&D stage of green innovation technology can allow for the coordination of the proportion of the input elements, improve the quality of input elements, create convenient paths for knowledge flow, and accelerate the progress of green innovation technology R&D. In the stage of green innovation achievement transformation, effective imitation and learning can strengthen the regional technological innovation levels, accelerate the circulation of regional data elements, and further improve green innovation efficiency. Thus, Hypothesis H2, proposed in this study, was validated. Effective imitation and learning during the transformation stage of green innovation achievements can bolster regional technological innovation levels, expedite the circulation of regional data elements, and further enhance green innovation efficiency.

6. Conclusions and Policy Implications

6.1. Conclusions

This study comprehensively explored the influence of digital economy development on the efficiency of two-stage green innovation, considering the moderating effects of intellectual property protection and data element endowment from the perspective of a two-stage innovation value chain. Based on this, the following conclusions were drawn.

First, there exists a notable spatial correlation between two-stage green innovation efficiency and digital economy development. The development of the digital economy not only enhances the two-stage green innovation efficiency within a region but also exerts a positive and significant spillover effect on the neighboring regions. After conducting a robustness test by substituting the explanatory variables, rearranging the economic geographical nested matrix, and delaying the explanatory variables by one period, this conclusion remained valid. Currently, the majority of the literature on green innovation efficiency focuses on environmental regulations, green finance, industrial agglomeration, etc., and it primarily examines single-stage green innovation efficiency [53,54,55]. This study complements these studies, enriching the literature base in the field of green innovation efficiency by providing a new interpretation from the perspective of the digital economy.

Second, in terms of different geographic locations, the advancement of the digital economy notably boosts two-stage green innovation efficiency, particularly in the eastern region, which is characterized by its elevated economic development and stronger R&D capabilities. However, in the central and western regions, where digital applications are less advanced and the primary industry remains traditional, digital economy development has not yet had a significant impact on the efficiency of two-stage green innovation. In terms of different development modes, non-resource-based provinces, which are technology-oriented and dominated by the tertiary industry and service-oriented manufacturing, experience a more significant impact of digital economy development on the efficiency of two-stage green innovation compared to resource-based provinces, thus enjoying greater benefits from development. Currently, the existing literature mainly focuses on studying the heterogeneity across different geographical regions [18,31], neglecting the potential negative impact of heterogeneity in development models on the robustness of the conclusions. This study comprehensively considers the development models and geographical locations of different regions, thus revealing a more comprehensive association between the digital economy and two-stage green innovation efficiency.

Third, the moderating effect showed that both intellectual property protection and data factor endowment positively moderate the influence of digital economy development on the efficiency of two-stage green innovation. Furthermore, intellectual property protection can exert a certain radiation effect on neighboring regions’ green development throughout both stages of green innovation. There are currently few studies that have examined the moderating role of the digital economy on green innovation efficiency. This study further extends the understanding of the moderating effects of intellectual property protection and data factor endowments.

6.2. Policy Implications

Based on the research and conclusions obtained in this study, the following implications were identified:

Firstly, strengthening digital construction is essential to making the digital economy a significant driver in enhancing two-stage green innovation efficiency. A top priority for all provinces is to enhance the construction of new infrastructure, including networks and data centers, and increase investments in new-generation information technologies such as cloud computing, big data, and artificial intelligence. It is also crucial to continuously expand information technology research and development platforms, gathering various talents and resources to provide solid technical support and guarantee the release of digital dividends. At the same time, we need to continuously broaden the application scope of digital technology, deeply integrate it into traditional industries, and promote technological innovation and transformation, as well as the upgrading of these industries. On this basis, we should focus on breakthroughs in key areas and technologies, continuously developing and promoting advanced green technologies to facilitate the transformation and application of green innovation achievements. In addition, promoting the establishment of green digital service platforms can provide convenient and efficient green services for various enterprises and individuals, effectively promoting the transformation of consumer demand concepts. Through such measures, we can not only promote the rapid development of the digital economy but also provide powerful support for green innovation, continuously improving the market share of green innovation products and achieving sustainable development.

Secondly, attention should be paid to the different development situations among regions, and targeted policies should be formulated based on regional and phased approaches. The eastern regions and non-resource-based provinces, which typically possess advanced technological bases and high economic development levels, should fully utilize their technological advantages, continuously increase the excavation, analysis, and application of data elements, and play a leading and exemplary role. On the other hand, due to differences in industrial structure, resource endowment, and other aspects, the central and western regions and resource-based provinces need to accelerate the upgrading of their industrial structure and promote the deep integration of traditional resource-based industries with the digital economy. Additionally, these regions can strengthen regional cooperation, achieve resource sharing and technological complementarity, and promote coordinated green innovation development within the region.

Thirdly, it is crucial to strengthen the protection of intellectual property rights, enhance the advantages of data elements, and foster a positive environment for technological innovation. To achieve this, we need to focus on the following: (1) continuously improving the conversion mechanism for patent inventions and scientific research achievements by refining relevant policies and measures to form a stable institutional support system that encourages innovation and promotes the research and application of green technologies; (2) strengthening the legal sanctions against infringing actions on innovative achievements by establishing and improving relevant laws and regulations, clarifying legal responsibilities for infringing actions, increasing the cost of infringement, and reducing the risk, which helps safeguard the legitimate rights and interests of innovators and stimulates businesses and all sectors of society to participate actively in green technology research and development; and (3) enhancing the collection, collation, and analysis of data resources to improve data quality and utilization efficiency and promoting information sharing and resource integration by building bridges for the free flow of green resources across different regions, thereby enhancing the quality of input elements and providing robust support for green technology research.

6.3. Limitations and Research Directions

This study faced limitations. Firstly, the sample was limited to the provincial level, but future research could expand to the prefecture level if the necessary data are available. Secondly, while this study focused on China over the past 11 years, it would be more instructive globally if expanded to a cross-national context. Lastly, due to data availability, the indicators used to measure the development of the digital economy and the efficiency of two-stage green innovation were imperfect; future research could further refine these indicators.