Digital Government Construction and Provincial Green Innovation Efficiency: Empirical Analysis Based on Institutional Environment in China

Abstract

Green innovation provides powerful incentives to achieve sustained social progress. However, the available research examines the financial drivers of green innovation, overlooking the impact of digital government development and the institutional environment. The integration of digital government construction with the institutional environment, and the coupling of the two with green innovation, will paint a picture of the future that promotes sustainable social progress and the modernization of governance. This research utilizes data from 31 provinces in China from 2018 to 2022 to study the impact of digital government construction and the institutional environment on the provincial green innovation efficiency. An empirical analysis is conducted on the basis of analyzing the spatiotemporal evolution and pattern of digital government construction, the institutional environment and the provincial green innovation efficiency. Firstly, digital government construction emphasizes data openness and sharing, and data become a key link between those inside and outside the government. The digital platform becomes an important carrier connecting the government and multiple subjects in collaborative innovation to continuously shape a new digital governance ecology. The netting of digital ecology is conducive to the institutional environment, serving to break the path dependence and create a more open, inclusive and synergistic institutional environment. Based on this, we consider that digital government construction positively affects the institutional environment, and this is verified. Secondly, a good government–market relationship, mature market development, a large market service scale, a complete property rights system and a fair legal system brought about by the improved institutional environment provide macro-external environmental support for enhanced innovation dynamics. Based on this, it is proposed that the institutional environment positively affects the provincial green innovation efficiency. Meanwhile, building on embeddedness theory, the industrial embeddedness of the institutional environment for green innovation highlights the scattered distribution of innovation components. Geographical embeddedness stresses indigenous resource distribution grounded in space vicinity and clustering. The better the institutional environment, the greater the forces of disempowerment at the industrial tier and the easier it is for resources to flow out. This may potentially have a detrimental role in improving the local green innovation efficiency. In view of this, it is proposed that the institutional environment negatively affects the provincial green innovation efficiency, and this is verified. Thirdly, digital government construction, as an important aspect of constructing a digital governance system and implementing the strategy of a strong network state, can effectively release the multiplier effect of digital technology in ecological environment governance and green innovation, continuously enhancing the provincial green innovation efficiency. In view of this, it is proposed that digital government construction positively affects the provincial green innovation efficiency, and this is verified. When the institutional environment is used as a mediating variable, digital government construction will have a certain non-linear impact in terms of provincial green innovation efficiency improvement. Building on the evidence-based analysis results, it is found that the institutional environment plays a competitive mediating role. This study integrates digital government construction, the institutional environment and the provincial green innovation efficiency under a unified analytical structure, offering theoretical inspiration and operational directions to enhance the provincial green innovation efficiency.

1. Introduction

Under the new wave of scientific and industrial revolution, digital technology, as a powerful emerging productive force, has already become a new driver of social progress and social governance. In 2017, the report of the 19th National Congress of the Communist Party of China put forward the construction of a digital China. This is the first time that the term ‘digital China’ has appeared in the report documents of China. This means that, five years after 2017, the building of a ‘digital China’ has become one of China’s main strategic tasks. Digital government construction has become one of the connotations of the construction of a digital China. Digital government construction is also an important initiative to adapt to the development trend of digital transformation and promote innovation in government governance. Meanwhile, in the context of the transformation of Chinese society towards digitalization and intelligence, the development mode is evolving towards intelligence, servitization, greening and low carbonization [1]. Green development has been an crucial impetus for the development of Chinese society. Green innovation is the first driving force for green development [2]. Green innovation has the positive externality of unifying technological, ecological and economic benefits, which can realize the triple enhancement of green innovation technology, the economic development quality and the ecological environment. In this critical period of social transformation in China, can digital government construction promote provincial green innovation efficiency? What is the intrinsic connection between the two? This is a key issue that deserves government attention and will have a bearing on the future direction of governance.

Green innovation is the innovative activity of achieving the sustainable development of the economy, society and ecological environment by improving the resource utilization efficiency, minimizing environmental pollution and promoting technological innovation. Green innovation is influenced by many factors, the most important of which is the institutional environment. The institutional environment is the main boundary condition for the effect of digital government construction on the provincial green innovation efficiency. Green innovation development faces many uncertainties. The quality of institutions is the primary determinant of whether these uncertainties can be effectively controlled. The institutional environment affects innovation development [3]. Therefore, it is necessary to take the institutional environment into account when studying the relationship between digital government construction and the provincial green innovation efficiency. Academics believe that the institutional environment plays a promoting role regarding innovation [4]. However, regarding China, there is no literature on the role of the institutional environment as a moderating variable for digital government construction and green innovation, as well as the roles and mechanisms of the institutional environment. Moreover, the theoretical and empirical studies on digital government construction, the institutional environment and provincial green innovation efficiency are mostly based on the West. A theoretical framework for the analysis of the intrinsic linkages among digital government construction, the institutional environment and provincial green innovation efficiency based on Chinese development practices has yet to be constructed. Obviously, this is an important and highly topical issue for Chinese academics to address and develop in the field of digital governance and green innovation.

The existing literature has explored the intrinsic link between digital government construction and green innovation and the intrinsic link between the institutional environment and green innovation. However, the theoretical construction still has the following gaps. Firstly, rooted in Chinese practice, Chinese provinces and regions are more heterogeneous. The level of digital government construction varies considerably. The heterogeneity of institutional environment connotation elements such as the market development degree and market scale is also strong. No study has yet examined the questions of ‘can digital government construction promote provincial green innovation efficiency and what is the intrinsic connection between the two?’ and ‘can the institutional environment promote provincial green innovation efficiency and what is the intrinsic connection between the two?’ from a full-sample perspective, considering the heterogeneity in the geographic locations and development degrees of provinces. Secondly, no study has yet examined the impact of the institutional environment on the relationship between digital government construction and provincial green innovation efficiency from the above perspectives and levels. Moreover, no study has yet constructed an analytical framework on the intrinsic connection between digital government construction, the institutional environment and provincial green innovation efficiency based on Chinese development practices. It is not yet sufficient to provide adequate theoretical support. Therefore, this study will focus on filling the research gaps mentioned above.

Given the above situation and study gaps, this work conducts empirical research on 31 provinces in China. The Chinese provincial government represents the highest local administrative level in China. It provides unified leadership for the work of all levels of the government within its jurisdiction and unifies the management of economic, social and other affairs within its jurisdiction. Chinese provincial governments play an irreplaceable role in local economic development and social construction in China. Chinese provincial governments are pioneering new models of local digital governance and green development. In the context of this paper, the significant status of Chinese provincial governments suggests that they are typical and representative research samples. Based on this, they are suitable for the exploration of a new analytical framework and empirical research. The research questions are as follows. Firstly, can digital government construction enhance the provincial green innovation efficiency? What is the intrinsic connection between the two? Are such connections heterogeneous geographically and with regard to the provincial development levels? Can institutional environment optimization drive the enhancement of the provincial green innovation efficiency? What is the intrinsic connection between the two? Are such connections are heterogeneous geographically and with regard to the provincial development levels? Secondly, what is the impact of the institutional environment? What is the mechanism of interaction? Thirdly, what are the inspirations for the identified new theoretical models and the influence mechanisms that will enable future provincial governments to precisely implement policies to promote the development of digital government construction, green innovation and the optimization of the institutional environment?

The contributions of this study include three aspects. Firstly, this study makes theoretical contributions by filling the research gaps. On the one hand, most of the research around digital government construction and provincial green innovation efficiency focuses on testing the linear effects among the two. There is no empirical study that explores the nonlinear relationship between the two. Based on 31 provinces in China, this study explores the influence mechanism of digital government construction on provincial green innovation efficiency. To some extent, it fills this research gap. On the other hand, the question of how the institutional environment affects these two variables has not been answered. This paper conducts a mediating effect and mechanism analysis of the institutional environment. Based on this, an analytical framework for the three is constructed. To some extent, it fills this research gap. Secondly, this study deepens the comprehension of the intrinsic connection between the three variables from a multi-regional and multi-type viewpoint. This comprehension is expanded from the multiple regions of Eastern, Central, and Western China, and it considers multiple types, such as leading, high-quality, distinctive, developmental, and catching up. Thirdly, this paper is the first to apply MATLAB research tools, measurement methods, and geographic disciplinary methods together to study the intrinsic connections between these three variables. This enriches the methodological toolbox for research on this issue. The conclusions are significant in enabling Chinese local governments to make precise policy decisions.

2. Literature Review and Hypothesis Development

The evolution of digital governments shows regular patterns. This evolution can be summarized in terms of the four patterns of digital government evolution: digitization, transformation, engagement, and contextualization [5]. Moreover, digital government research has experienced progressive expansion from technology to system to ecology. It can be roughly categorized into three phases: the early technology-centered theory, the mid-term management-centered theory, and the ecocentric theory that has emerged in recent years. The early studies of digital governments were heavily influenced by the new public management movement, which was strongly characterized by technological determinism. The introduction of information technology in government departments gave rise to a wave of managerial innovation marked by ‘e-enablement’ [6]. Early studies focused on the technology perspective and explored the application of emerging technologies in government management and services. Studies from the technology perspective have argued that digital governments can help to improve the decision-making accuracy, governmental transparency, and administrative efficiency. Although some of the scholars in these early studies paid attention to the topic of organizational transformation, the focus was still on the technology that preceded the transformation [7]. Research on digital governments in this period had not yet involved deep-level organizational transformation. Subsequent studies have begun to focus on the institutional dimensions of digital government development, proposing an ecosystem analysis framework for digital governments. Research in this period reveals the compound logic of data flows, institutional constraints, and pluralistic interactions shaping digital governance [8]. With the arrival of the Web 2.0 era, the center of digital government research has shifted from ‘technology’ to ‘management’. Scholars have studied the value orientation and realization route of digital governments from a management view. New public management concepts such as public orientation, process reengineering, and performance management have received attention. This has also prompted scholars to think about how to realize the transformation of the service-oriented government [9]. The research on digital governments in this period has begun to focus on internal organization. After entering the Web 3.0 era, rising technologies have been integrated with government governance, and digital government research has shifted to the ‘ecology’ center. On the one hand, data have become a key link between those inside and outside the government. Data governance has become a new topic in digital government construction [10]. On the other hand, multiple subjects continue to shape the new ecology of digital governance in collaborative innovation [11]. The studies in this period are more stereoscopic, dynamic, and holistic. The benefits of digital government construction include support for good governance [12], cost savings, and efficiency improvements [13], as well as improvements in the quality of decision-making. Meanwhile, the challenge lies not in the technology utilization capacity but in overcoming the differences between government departments and hierarchies [14].

The relevance of digital government research continues to rise. Based on the actual governance of the current digital government applied to China, the present-day connotations of digital governance are summarized in terms of the following three aspects. The first is leading with conceptual reshaping. Digital governance is essentially a series of innovations brought about by the interface of technology and elements such as cognition, behavior, and organization [15]. In terms of the essence of information technology, it is conducive to increasing government transparency and creating a more harmonious digital governance environment. The digital government reshapes the concept of development with the help of technological means [16]. Secondly, technology empowerment is the driving force. Technology provides many possibilities for digital governments to realize public value. Through the positive interaction between the public and the government, the digital government and its governance are an effective way to improve service provision and respond to citizens’ demands [14]. Thirdly, process reengineering is the object. The essence of digital governments is to realize the goal of ‘process reengineering’. Its fundamental goal is to improve the organizational efficiency and governance effectiveness [17]. The organizational transformation and organizational reengineering brought about by digital governments promote the transformation of organizational structures towards modernization.

In the last few years, green innovation has flourished and taken hold. The cross-integration of digital technology and green technology has become a development impetus. The digital government uses digital technology to provide digital and informationized governance services [18]. Currently, the digital government is the pivotal core of governance. With technological development and the deepening of governments’ digital transformation, strengthening the interaction between technology and organization [19], as well as strengthening digital government construction, has become one of the core issues of government governance. Digital government construction stimulates the improvement of the provincial green innovation efficiency mainly through innovating data elements [20] and environmental systems [21] and through government R&D subsidies [22]. Digital government construction offers a rare opportunity for innovation [23]. Digital government construction provides the driving force for green innovation in terms of three aspects. From a technological perspective, information technology, digital technology, and green technology are the foundations of green innovation. Green innovation is a new form of innovation. Digital government construction creates an impetus for green innovation development. From the perspective of effectiveness, green innovation emphasizes the organic unity of material factor inputs, the institutional environment, and technological innovation [24]. The digital government is the integration of digital technology and digital governance. Digital government construction and its leadership can enable green innovation to advance in an integrated manner under an innovation-driven engine. From the perspective of quality orientation, green innovation is deeply integrated into the concepts of green, sharing, and innovation [25], which is in line with the concept of digital government construction. They represent a positive interaction between the two. The existing research presents a dynamic and systematic picture of the two. However, when implemented at the practical level, this concept still requires systematic integration in terms of system design and policy tools. At the system level, it requires cross-level and cross-sectoral synergistic mechanism construction. It is necessary to both guide and strengthen the incentives and constraints [15]. Meanwhile, governance oriented towards complex systems also places demands on policy tools. Behind the integrated development of digital government construction and green innovation and the optimization of its system design, what is implied is the reconstruction of the governance system from the traditional hierarchical system to the flattened, differentiated, and refined regulation of the digital era [7]. These aspects are crucial to improve the development environment of digital government construction and green innovation.

Under the digital and intellectual era, the question of how to carry out public sector transformation, promote digitalization, and promote green innovation development is an important issue regarding the enhancement of the governance effectiveness. From the perspective of the historical evolution of government reform, public sector transformation is always an important component of government reform. Following the evolution of the government from the role of ‘nightwatchman’ to the nationalization of administration, the public sector has experienced a transformation from the paradigm of traditional public administration to the paradigm of new public management [26]. Digital government construction is both the government’s self-adjustment to digital transformation and an inevitable requirement for the government to improve its digital governance levels. Currently, the paradigm of ‘new public management’, which is mainly characterized by holistic governance, puts forward new requirements for public governance in various countries. Holistic governance takes holism as its goal, coordinating the relationship between the government, market, and society [27]. Holistic governance provides a holistic analytical perspective to improve the government’s governance level.

Under the public sector’s digital transformation and holistic public value leadership, digital government construction applies technology to government transformation and social governance. The target is to provide a composite system for government reengineering with the integration of the government’s concept, organization, process, and method [27]. In turn, it empowers green innovation. Firstly, digital government construction is ‘technology-enabled’ for governance. It provides governance innovation paths for the development of governance practices, such as the governmental function transformation of decentralization and empowerment and the innovation of service provision methods with pluralistic participation. In turn, it creates a favorable environment for market subjects to carry out green innovation and stimulates the vitality of green innovation. Moreover, digital governance approaches such as digital platform construction and technology infrastructure construction provide technical support for green innovation. In turn, this will smooth the channel of green innovation technology and improve the provincial green innovation efficiency. Secondly, digital government construction can drive the realization of a governance landscape that is multifaceted and co-governed by the government, market, and society. Big data platform construction, cloud computing, and other digital governance tools build bridges to realize open communication channels between the government and the public. Thereby, they empower various social forces to monitor government affairs and participate in public decision-making [28]. The collaborative and shared governance landscape of multiple innovation subjects provides the impetus for green innovation. Furthermore, digital government construction, as an indispensable aspect in structuring a digital governance system and implementing the strategy of a strong network country, can effectively release the multiplier effect of digital technology at the level of eco-environmental governance and green innovation and continuously improve the provincial green innovation efficiency. On the one hand, digital government construction can effectively help cities to create the green mode of ‘Intelligence +’, carrying out whole-chain intelligent, green, and digital transformation and continuously improving the green innovation level. On the other hand, digital government construction can comprehensively update the digital control platform and help multiple subjects to obtain real-time information data about green innovation, which provides abundant data for the orderly promotion of green innovation. Therefore, we formulate Hypothesis 1.

Hypothesis 1 (H1): 

Digital government construction positively affects the provincial green innovation efficiency.

Innovation is marked by a heightened level of unpredictability and places high demands on the institutional environment [29]. The concept of the institutional environment originates from institutional theory and refers to the ecosystem of policies, laws, and regulations in which the system operates [30]. An excellent institutional environment is an important prerequisite for the innovation-driven effect to be exerted [31]. A good public service mechanism and business environment are conducive to incentivizing enterprise innovation and technological innovation [32]. Government regulations, market policies, and legal supervision can greatly affect innovation activities. There are differences in the institutional environment in different regions, as well as differences in innovation activities [33]. Some scholars put forward the concept of ‘appropriate innovation’, believing that the appropriate way to promote regional technological progress is to choose an innovation model that is suitable for the regional technological ecology [32]. Academic research on the relationship between the institutional environment and innovation generally agrees that regions with high institutional quality have stronger innovative capabilities. Quality institutions, property rights protection, financial marketization, and government–market relations positively affect technological innovation [34,35]. According to institution theory, the production practice of innovation subjects will be subject to the constraining effect of institutions, so that all innovation subjects in the region will comply with the existing institutional environment, reflecting the isomorphism effect of institutions [36]. When following established institutional norms, innovation subjects tend to gain higher legitimacy, more resources, and stronger viability. For the green industry, government policies, environmental regulations, and industry norms exert external institutional pressure. For example, governments may force innovation subjects to invest more in green innovation to meet new regulatory requirements by setting norms such as mandatory energy saving and emission reduction targets. Regulatory pressure indirectly promotes green innovation development [37]. Meanwhile, the institutional environment can establish an effective legal system and stable market to support the development of green innovation, protect innovation patents, property rights, etc., and promote the improvement of the green innovation capacity [38,39,40].

The institutional environment comprises the political, legal, social, and other basic rules that affect the behavior of innovation subjects. The institutional environment influences organizational behavior. There are abundant research achievements on the institutional environment and innovation. The view that the better the institutional environment, the greater the capacity of organizations to innovate is widely agreed upon. Good government–market relations, mature market development, a large scale of market services, a well-developed property rights system, and a fair legal system provide macro-external environmental support for enhanced innovation dynamics [34,35]. Academics mostly use the marketization index to measure the institutional environment. According to the marketization index, proposed by Fan Gang’s team in the ‘China marketization index report’, the marketization level of China is reflected comprehensively by the relationship between the government and market, the development of the non-state-owned economy, the development degree of the product market, the scale of market services, and the legal system environment [41]. The impact of the above five aspects on the provincial green innovation efficiency is specifically analyzed below. Firstly, straightening out the relationship between the government and market helps to reduce the crowding-out effect of rent-seeking behaviors on green innovation. In the process of government intervention in the market in China, the government will intervene in the scale and direction of market investment to achieve certain political or social purposes. This may lead to rent-seeking behavior and speculation by market subjects in order to seize innovation resources [42], while neglecting green technology research and development. Straightening out the relationship between the government and market can effectively avoid such crowding-out effects on the basis of ensuring sufficient innovation resources, and it can improve the provincial green innovation efficiency. Secondly, the equal development of the non-state-owned economy promotes the motivation of innovation subjects to carry out green innovation activities. A favorable institutional environment means a stable business environment. The government rationally allocates innovation resources according to the market scale and innovation intensity. The government boosts the enthusiasm of innovation subjects to carry out green innovation activities by optimizing the distribution of innovation resources. Thirdly, well-developed product markets stimulate enterprises to develop green products. A well-developed product market contributes to the role of the market mechanism in moderating product flows and reducing the green innovation costs. Fourthly, the expansion of the market scale contributes to enhancing the green innovation level by optimizing the distribution of elements. A favorable institutional environment contributes to the efficient flow of material, information, and value [43]. This will promote the optimization and upgrading of green innovation services and improve the provincial green innovation efficiency. Fifthly, the improvement of the legal system helps to stimulate the enthusiasm of innovation subjects to carry out green innovation. In regions with poor legal systems, speculation prevails, leading to increased risks and costs of green innovation [44]. This is because of inadequate intellectual property protection systems, contractual systems, etc. Intellectual property protection systems contribute to an increase in the number of patents in different countries [45,46]. Legal security for innovation subjects can compensate for the above systematic loopholes and stimulate green innovation initiative. Therefore, we formulate Hypothesis 2.

Hypothesis 2 (H2): 

The institutional environment positively affects the provincial green innovation efficiency.

In contrast, it has also been argued that China’s basic social, economic, and other institutional environments may affect innovation negatively [4]. Therefore, the impact of different institutional environments on green innovation is unclear. Green innovation development is largely driven by government subsidies. The crowding-out effect of government-subsidized inputs is not conducive to promoting innovation inputs by innovative subjects and will reduce productivity [47]. There are inter-industry, intra-industry, and spatial spillover effects in green innovation R&D by innovation subjects. When faced with different institutional environments and when green technology is not effectively utilized, green innovation cannot be promoted [35]. Green innovation is viewed from a mixed embeddedness perspective, where green innovation activities are embedded in multiple institutions [48]. An essential expression of hybrid embeddedness is that industry embeddedness and geographical embeddedness are not independent but mutually affected. Industry embeddedness highlights the dispersed distribution of innovation elements [49]. Regional embeddedness stresses localized resource assignment on the basis of space vicinity and clustering [50]. The relationship between the two is a ‘trade-off’ when the total amount of resources remains unchanged [51]. Specifically, the better the institutional environment, the greater the forces of disempowerment at the industrial tier and the easier it is for resources to flow out. This may potentially have a detrimental role in improving the local green innovation efficiency. Therefore, we formulate Hypothesis 3.

Hypothesis 3 (H3): 

The institutional environment negatively affects the provincial green innovation efficiency.

The aim of digital government construction is to embed digital technology into government governance in order to improve the governance level and capacity. The application of digital governance technologies by the government includes areas such as the rational allocation of market factors [52], effective public monitoring, and the suppression of corruption [35]. Thereby, it promotes a better institutional environment. Moreover, digital government construction emphasizes data sharing and openness, aiming to enhance the efficiency of data resource utilization by creating an efficient and complete data resource system. Digital platforms are increasingly becoming an important carrier connecting the government and multiple subjects in collaborative innovation to continuously shape a new ecology of digital governance. The netting of the digital ecology based on this collaborative governance of multiple subjects is conducive to breaking the path dependence in the institutional environment and creating a more open, inclusive, and synergistic institutional environment. Therefore, we formulate Hypothesis 4.

Hypothesis 4 (H4): 

Digital government construction positively affects the institutional environment.

If Hypotheses 2 and 4 both hold, it indicates that the institutional environment moderates the relationship between digital government construction and the provincial green innovation efficiency. Moreover, it is a complementary mediating process. If Hypotheses 3 and 4 hold simultaneously, it indicates that it is a competitive mediating process. Thus, we propose Hypothesis 5.

Hypothesis 5 (H5): 

The institutional environment mediates the relationship between digital government construction and the provincial green innovation efficiency.

These hypotheses provide a theoretical analytical framework to explain the relationship between digital government construction, the institutional environment, and the provincial green innovation efficiency (Figure 1). This multi-level (innovation subject, government, and institution level) theoretical framework has not received sufficient attention or widespread application.

3. Data, Variables, and Models

3.1. Data

This study is conducted on a research sample of 31 provinces in China, using data from 2018 to 2022. Hong Kong, Macao, and Taiwan are excluded from the study sample due to data availability problems. The data are obtained from the China Statistical Yearbook, China Environmental Statistical Yearbook, China Science and Technology Statistical Yearbook, and provincial official statistical yearbooks, etc. To ensure the validity, accuracy, and reliability of the data analysis, this study performs the following processing of the data before the data analysis. Firstly, to minimize the impact of extreme values on the analysis results, all continuous variables are winsorized at the 1% level. Secondly, a small number of samples with missing values are supplemented using the linear interpolation method.

3.2. Variables

3.2.1. Dependent Variable

The dependent variable is the provincial green innovation efficiency (GIE). Referring to the study by Wang Q et al. [53], the provincial green innovation efficiency is measured by the super-efficiency SBM model. The super-efficiency SBM model is one of the super-efficiency DEA models. Firstly, the efficiency assessment of production and management activities is more common in economic management research. Compared with regression models, data envelopment analysis (DEA), as a nonparametric estimation method, does not require us to set the specific form of the production function between the inputs and outputs in advance; rather, it directly utilizes the method of linear programming to set the weights of the input and output data, avoiding the influence of subjective factors on the measurement process. In addition, the DEA method can realize the integration analysis of multiple input and output elements and maximize the input and output indicators in the model, making the input and output results more in line with the reality. This demonstrates the effectiveness and applicability of DEA methods in efficiency assessment. Therefore, this study uses DEA. Secondly, green innovation is a product of the combination of innovation theory and environmental protection theory, emphasizing the maximization of resource saving and ecological protection. The super-efficiency SBM model takes into account the impact of slack variables on efficiency measurement. The super-efficiency SBM model solves the problem of non-desired outputs in the input–output model. If other DEA models, such as the BCC model, are used for measurement, the problem of not being able to rank the provincial green innovation efficiency when the efficiency value is equal to 1 may occur. However, the super-efficiency SBM model can solve such problems, thus enabling the ranking of the efficiency levels of the decision-making units. Therefore, this study adopts the super-efficiency SBM model for analysis.

Taking into account the existing literature, there are mainly two types of indicators to be considered when measuring the provincial green innovation efficiency. One type is innovation input indicators. After reviewing the existing literature, it is found that scholars in China and abroad select input indicators of provincial green innovation efficiency mostly from human, material, and financial inputs. Therefore, this study refers to this indicator system to further refine the specific indicators. (1) Labor input. Green innovation development and provincial green innovation efficiency enhancement are closely related to talent, such as highly skilled researchers and scientific and technical service personnel. Talent, as the most central subject in promoting green innovation, plays an irreplaceable role in green innovation development. Referring to the study by Xu YJ et al. [54], the number of employees in the research and comprehensive technical services industry is used to measure the labor input. (2) Financial input. Green innovation is a product of the combination of innovation theory and environmental protection theory, emphasizing the maximization of resource saving and ecological protection. The enhancement of the provincial green innovation efficiency requires the combination of green technology and innovative technology and its positive operation. Therefore, for green development, the financial input mainly refers to the government expenditure on energy saving and environmental protection. For innovative development, the cultivation of scientific research talent and the development of scientific research technology are the most crucial. Therefore, referring to the study by Wang Q et al. [53], financial input is measured as the sum of government expenditures on energy conservation and environmental protection, education, and science and technology. (3) Material input. The process of green innovation development requires the consumption of resources. These resources include a combination of resources, with water, electricity, and gas as the main consumables. Therefore, referring to the study by Wang H et al. [55], the entropy method is used to measure the material input by using the composite index of resource consumption consisting of the total electricity consumption, total water supply, and total liquefied petroleum gas supply of the entirety of society. Another type is innovation output indicators. Specifically, they include the following. (1) Expected output. Green innovation technology can help to realize value-added green products and generate economic benefits. The enhancement of the provincial green innovation efficiency interacts with the progress of green technology to produce technological benefits. Provincial green innovation efficiency enhancement positively affects energy savings, environmental protection, and ecological construction, generating ecological benefits. The economic, technological, and environmental benefits arising from efficiency improvements in green innovation will benefit society as a whole. Therefore, referring to the study by Wang H et al. [55], the GDP per capita, the number of green patent applications granted, and the green coverage rate are used as indicators to measure the economic, technological, and environmental benefits of provincial green innovation efficiency, respectively. (2) Unexpected output. Waste is undoubtedly generated during the green innovation process, i.e., the unexpected outputs of provincial green innovation efficiency improvement. Waste water, waste gas, and dust are the main types of waste. Therefore, referring to the study by Chen B et al. [56], the entropy method is used to measure the unexpected output by using the pollution composite index consisting of wastewater emissions, exhaust gas emissions, and dust emissions.

The formula for the super-efficiency SBM model used to measure provincial green innovation efficiency is as follows:

$$\min {\theta }^{*}={\displaystyle \frac{1+\frac{1}{m}{\sum }_{i=1}^m\frac{S_i^{-}}{x_{i0}^t}}{1-\frac{1}{q+h}\left({\sum }_{r=1}^q\frac{s_r^{+}}{y_{r0}^t}+{\sum }_{k=1}^h\frac{s_k^{-}}{b_{k0}^t}\right)}}$$

s.t.

$$x_{i0}^t\ge \sum _{t=1}^T \sum _{j=1,j\ne 0}^n{\lambda }_j^t{x}_{ij}^t-s_i^{-}i=1, 2,\cdots ,m$$

$$y_{r0}^t\ge \sum _{t=1}^T \sum _{j=1,j\ne 0}^n{\lambda }_j^t{y}_{rj}^t+s_r^{+}r=1, 2,\cdots ,q$$

$${ b}_{k0}^t\ge \sum _{t=1}^T \sum _{j=1,j\ne 0}^n{\lambda }_j^t{b}_{kj}^t-s_k^{-}k=1, 2,\cdots ,h$$

$${\mathsf{\lambda }}_{\mathrm{j}}^{\mathrm{t}}\ge 0\left(\forall \mathrm{j}\right), s_i^{-}\ge 0\left(\forall \mathrm{i}\right),{ s}_r^{+}\ge 0\left(\forall \mathrm{r}\right), s_k^{-}\ge 0\left(\forall \mathrm{k}\right)$$

${\mathsf{\theta }}^{*}$ is the provincial green innovation efficiency. $\mathsf{\lambda }$ is the model weight vector. ${\mathrm{s}}_{\mathrm{i}}^{-}$, ${\mathrm{s}}_{\mathrm{r}}^{+}$, and ${\mathrm{s}}_{\mathrm{k}}^{-}$ represent slack vectors of the green innovation inputs, expected outputs, and unexpected outputs, respectively. ${\mathrm{x}}_{\mathrm{i}}^{\mathrm{t}}$, ${\mathrm{y}}_{\mathrm{r}}^{\mathrm{t}}$, and ${\mathrm{b}}_{\mathrm{k}}^{\mathrm{t}}$ represent the green innovation inputs, expected outputs, and unexpected outputs of the decision-making unit in period t, respectively. n represents the overall number of decision-making units (individuals). m, q, and h represent the overall factor inputs, the number of expected outputs, and the number of unexpected outputs (individuals), respectively. T refers to the overall study period (years).

3.2.2. Core Independent Variable

The core independent variable is digital government construction (DG). It is used to determine the role that digital government construction plays in provincial green innovation efficiency improvement. We select the index of the online government service capacity of provincial governments as a proxy variable for the digital government construction level. The ‘Survey and Assessment Report on the Online Government Service Capacity of Provincial Governments and Key Cities’ contains an index for the online government service capacity of provincial governments from 2018 to 2022. This index consists of five first-level indicators: the effectiveness of online services, the maturity of online processing, the completeness of service modes, the coverage of service matters, and the accuracy of office guides. This index comprehensively reflects the public’s sense of experience and satisfaction with online government services. The index is currently only updated to the year 2021. Referring to the measurement method of Yu HH et al. [57], the average annual growth of each indicator over the years is used to fill in the 2022 data.

3.2.3. Moderator Variable

The moderator variable is the institutional environment (IE). It is used to measure the impact of the institutional environment on green innovation. Referring to the study by Yue MY et al. [58], this study measures the institutional environment from five dimensions. They are government and market relations, the development level of the non-state-owned economy, the development degree of product markets, the development degree of factor markets, and the development of market intermediary organizations and law-and-order environments. These five dimensions are the secondary indicators of the China marketization index constructed by Fan Gang’s team. The China marketization index is currently updated only to the year 2019. Therefore, referring to the treatment of Yu HH et al. [57], the average annual growth of the secondary indicators of the China marketization index over the years is used to calculate the data for 2020 to 2022.

The specific measurements are as follows. The first step is to make the data dimensionless. The indicators for all five dimensions are positive. As indicators of different natures have a significant impact on the results, it is essential to remove the impact of the dimensions in the calculation process and normalize the data. The processing is as follows:

$${{\mathrm{X}}_{\mathrm{i}\mathrm{j}\mathrm{t}}}^{\prime }={\displaystyle \frac{{\mathrm{x}}_{\mathrm{i}\mathrm{j}\mathrm{t}}-\mathrm{m}\mathrm{i}\mathrm{n}\left({\mathrm{x}}_{\mathrm{i}}\right)}{\mathrm{m}\mathrm{a}\mathrm{x}\left({\mathrm{x}}_{\mathrm{i}}\right)-\mathrm{m}\mathrm{i}\mathrm{n}\left({\mathrm{x}}_{\mathrm{i}}\right)}}$$

(1)

$\mathrm{i}$ indicates the selected indicator. $\mathrm{j}$ indicates the province. $\mathrm{t}$ indicates the year. ${\mathrm{X}}_{\mathrm{i}\mathrm{j}\mathrm{t}}$ indicates the raw data for indicator $\mathrm{i}$ in year $\mathrm{t}$ in province $\mathrm{j}$. ${{\mathrm{X}}_{\mathrm{i}\mathrm{j}\mathrm{t}}}^{\prime }$ indicates the value after it is dimensionless. $\mathrm{m}\mathrm{a}\mathrm{x}\left\{{\mathrm{x}}_{\mathrm{i}}\right\}$ and $\mathrm{m}\mathrm{i}\mathrm{n}\left\{{\mathrm{x}}_{\mathrm{i}}\right\}$ indicate the maximum and minimum values of indicator $\mathrm{i}$ for all provinces for all years, respectively. The second step is to use the entropy weight method to measure the weight of each indicator and the weighted average of the above dimensionless results, and we finally obtain the institutional environment index for each province in each year.

3.2.4. Control Variables

This study selects the following variables as control variables to overcome the deviations of missing variables (

Table 1. Main variables and measurements.

Variable Type	Variable Name	Symbol	Measurement
Dependent variable	Provincial green innovation efficiency	GIE	Measured based on super-efficiency SBM model
Core independent variable	Digital government construction	DG	Index of online government service capacity of provincial governments
Moderator variable	Institutional environment	IE	Measured based on five dimensions
Control variables	Economic development level	ED	Provincial GDP per capita
	Technology investment effort	TI	Proportion of provincial expenditure on science and technology to regional GDP
	Tax burden level	TB	Proportion of tax revenue to regional GDP
	Social consumption level	SC	Proportion of total retail sales of consumer goods of society to regional GDP
	Foreign direct investment level	FDI	Proportion of FDI to regional GDP
	Unemployment rate	UE	Registered unemployment rate

). (1) Economic development level (ED). Considering the differences in the scales of provinces, referring to the studies of Liang RB et al. [59] and Huang DQ et al. [60], the provincial GDP per capita is used to indicate the economic development level. (2) Technology investment effort (TI). The proportion of provincial expenditure on science and technology to the regional GDP is used to represent the technology investment effort [61]. (3) Tax burden level (TB). The proportion of tax revenue to the regional GDP is used to represent the tax burden level [62]. (4) Social consumption level (SC). The proportion of the total retail sales of consumer goods of society to the regional GDP is used to represent the social consumption level [63]. (5) Foreign direct investment level (FDI). The proportion of FDI to the regional GDP in each province is used to represent the FDI level [64]. (6) Unemployment rate (UE). Expressed using the registered unemployment rate for each province [63].

3.3. Models

This study uses a multiple linear regression model to analyze the impact of digital government construction and the institutional environment on the provincial green innovation efficiency. The model is built as follows:

$${\mathrm{G}\mathrm{I}\mathrm{E}}_{\mathrm{i}\mathrm{t}}=\mathsf{\alpha }+\mathsf{\beta }{\mathrm{X}}_{1\mathrm{i}\mathrm{t}}+\mathsf{\gamma }{\mathrm{X}}_{2\mathrm{i}\mathrm{t}}+\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(2)

${\mathrm{G}\mathrm{I}\mathrm{E}}_{\mathrm{i}\mathrm{t}}$ represents the provincial green innovation efficiency, which refers to the provincial green innovation efficiency in year $\mathrm{t}$ for province $\mathrm{i}$th. ${\mathrm{X}}_{1\mathrm{i}\mathrm{t}}$ and ${\mathrm{X}}_{2\mathrm{i}\mathrm{t}}$ represent the digital government construction level and institutional environment index, respectively. Controls is the group of control variables. ${\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$ is the random error term. $\mathsf{\alpha }$, $\mathsf{\beta }$, and $\mathsf{\gamma }$ represent the constant term, the regression coefficient of digital government construction, and the regression coefficient of the institutional environment, respectively.

Moreover, drawing on Gu JF’s research method [65], this study analyzes the mediating effect of the institutional environment through the following model:

$${\mathrm{G}\mathrm{I}\mathrm{E}}_{\mathrm{i}\mathrm{t}}=\mathrm{C}+\mathsf{\rho }{\mathrm{W}\times \mathrm{G}\mathrm{I}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_0{\mathrm{I}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_1{\mathrm{E}\mathrm{D}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_2{\mathrm{T}\mathrm{I}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_3{\mathrm{T}\mathrm{B}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_4{\mathrm{S}\mathrm{C}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_5{\mathrm{F}\mathrm{D}\mathrm{I}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_6{\mathrm{U}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}},{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}~\mathrm{N}\left(0,{\mathsf{\sigma }}^2{\mathrm{I}}_{\mathrm{n}}\right)$$

(3)

$\mathrm{i}$ represents the $\mathrm{i}$th province. $\mathrm{i}$ = 1, 2,…, 31. $\mathrm{t}$ represents the year $\mathrm{t}$. $\mathrm{t}$ = 2018, 2019,…, 2022. W is the spatial contiguity weight matrix, constructed depending on whether the two provinces are in the same region or not. $\mathrm{W}\times {\mathrm{G}\mathrm{I}\mathrm{E}}_{\mathrm{i}\mathrm{t}}$ is the product of the dependent variable and the spatial weight matrix, i.e., the spatial lag term of the dependent variable [66]. Thus, there is a spatial lag panel model constructed [67].

The test for the mediating variables is conducted via a three-stage regression approach [68,69].

$$\mathrm{Y}=\mathrm{c}\mathrm{X}+{\mathrm{e}}_1$$

(4)

$$\mathrm{M}=\mathrm{a}\mathrm{X}+{\mathrm{e}}_2$$

(5)

$$\mathrm{Y}={\mathrm{c}}^{\prime }\mathrm{X}+\mathrm{b}\mathrm{M}+{\mathrm{e}}_1$$

(6)

According to the study of Gu JF [70], the mediating effect test can be accomplished by adding the following equation to Equation (3).

$${\mathrm{G}\mathrm{I}\mathrm{E}}_{\mathrm{i}\mathrm{t}}=\mathrm{C}+{\mathsf{\rho }}_1\mathrm{W}\times {\mathrm{G}\mathrm{I}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_1{\mathrm{E}\mathrm{D}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_2{\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_3{\mathrm{T}\mathrm{B}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_4{\mathrm{S}\mathrm{C}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_5{\mathrm{F}\mathrm{D}\mathrm{I}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_6{\mathrm{U}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}},{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}~\mathrm{N}\left(0,{\mathsf{\sigma }}^2{\mathrm{I}}_{\mathrm{n}}\right)$$

(7)

$${\mathrm{I}\mathrm{E}}_{\mathrm{i}\mathrm{t}}=\mathrm{C}+{\mathsf{\rho }}_2{\mathrm{W}\times \mathrm{G}\mathrm{I}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_1{\mathrm{E}\mathrm{D}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_2{\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_3{\mathrm{T}\mathrm{B}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_4{\mathrm{S}\mathrm{C}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_5{\mathrm{F}\mathrm{D}\mathrm{I}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_6{\mathrm{U}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}},{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}~\mathrm{N}\left(0,{\mathsf{\sigma }}^2{\mathrm{I}}_{\mathrm{n}}\right)$$

(8)

Equations (7) and (8) correspond to Equations (4) and (5), respectively. Equation (3) corresponds to Equation (6). Equation (7) is the spatial lag panel model. Equation (8) is the spatial panel model with a spatial lag term for the target variable. Equations (7) and (8) are dynamic spatial panel models [67]. If ${\mathsf{\rho }}_1$ in Equation (7), ${\mathsf{\rho }}_2$ in Equation (8), and $\mathsf{\rho }$ and ${\mathsf{\beta }}_0$ in Equation (3) are all significant, this indicates that the institutional environment is the mediating variable. If ${\mathsf{\rho }}_2\times {\mathsf{\beta }}_0$ and $\mathsf{\rho }$ have reversed symbols, the mediation is competitive; if the opposite is true, the mediation is complementary.

4. Results

4.1. Analysis of Spatiotemporal Pattern Evolution

4.1.1. Temporal Evolution Analysis

Firstly, we consider the temporal evolution of the provincial green innovation efficiency. In order to clarify the distribution pattern and dynamic evolution trend of the provincial green innovation efficiency over a period of time, two methods, namely traditional static kernel density estimation and non-spatial dynamic kernel density estimation, are applied using the MATLAB 2021 software. Figure 2 shows the results of the traditional static kernel density estimation, showing the kernel density distribution of the provincial green innovation efficiency over the five years from 2018 to 2022. During the sample period, the main body of the density curve is concentrated towards the middle year by year. The highest point of each year’s peak occurs in 2019. The peak of the density curve continues to decrease in 2020, 2021, and 2022. However, the peak of the density curve in 2022 is still higher than that in 2018. The width of the curve continues to narrow. This indicates that the overall development level of the provincial green innovation efficiency in China is undergoing an upward trend, and the average difference in the provincial green innovation efficiency between provinces is reduced.

We continue to examine the trend of the provincial green innovation efficiency in Chinese provinces from year t to year t + 2 based on non-spatial dynamic kernel density estimation. Figure 3 shows the dynamic kernel density figure. Figure 4 shows the density contour plot. In the dynamic kernel density figure, the X-axis represents the result in year t. The Y-axis represents the result in year t + 2. The Z-axis represents the probability of each point in the plane of the X–Y-axis. In the density contour plot, the greater the distance from the contour, the lower the probability, increasing sequentially inward. The density of the contour represents the convergence rate of the provincial green innovation efficiency; the more dense the contour, the faster the convergence rate. If the main body of the contour is distributed around the positive 45-degree line, the efficiency level of a province in year t + 2 is not much different from that in year t. If the main body of the contour is parallel with the Y-axis and above the 45-degree line, the green innovation efficiency of a province in year t + 2 is substantially improved. If the main body of the contour is parallel to the X-axis and lies to the right of the positive 45-degree line, the green innovation efficiency of a province in year t + 2 converges to a certain efficiency level.

In Figure 3 and Figure 4, the main body of the density contour is distributed around the positive 45-degree line, showing that the provincial green innovation efficiency does not vary much between years t and t + 2. The main crest is formed at x = 0.4. There is a secondary crest in the density contour at approximately x = 1; the graphical trend occurs approximately around the positive 45-degree line, and the x-values are approximately in the range of 0.75 to 1.25. This indicates that provinces with green innovation efficiency around 0.75 tend to approach values of 0.75 to 1.25 after 2 years. When the green innovation efficiency of a province is less than 0.75, the green innovation efficiency rises more after 2 years. When the green innovation efficiency of a province is more than 0.75, the overall green innovation efficiency after 2 years shows a rising trend, but the increase is relatively small. During the sample period, no trend indicating the probability body being parallel to the X-axis was found. This indicates that the development trend continues to be positive, with more room for improvement, and has not shown signs of convergence to a certain level.

Secondly, we consider the temporal evolution of digital government construction. Figure 5 shows the results of the traditional static kernel density estimation, showing the kernel density distribution of digital government construction over the five years from 2018 to 2022. In Figure 5, the main body of the density curve shifts rightward from year to year, with the highest point of the peaks for each year occurring in 2018. The width of the curve tends to narrow. This indicates that the overall development level of digital government construction in China is on an upward trend, and the average difference between the provinces is narrowing in general. In Figure 6 and Figure 7, the main body of the density contour is mainly distributed around the positive 45-degree line. This indicates that the level of digital government construction in China does not differ much from year t to t + 2. The main body of the density contour peaks at approximately x = 82. When the level of digital government construction in a province is less than 82, the increase in the level of digital government construction after 2 years is greater. When the level of digital government construction in a province is more than 82, the level after 2 years rises relatively little, although the overall trend is upward. During the sample period, no trend indicating the probability body being parallel to the X-axis was found. This indicates that the level of provincial digital government construction in China has not shown signs of convergence to a certain level. The level of provincial digital government construction in China is on an upward trend in general.

Thirdly, we consider the temporal evolution of the institutional environment. Figure 8 shows the traditional static kernel density estimation figure for the institutional environment. Figure 9 shows the dynamic kernel density figure for the institutional environment. Figure 10 shows the density contour plot for the institutional environment. In Figure 8, the highest point of the peaks for each year occurs in 2018. The main body of the density curve shifts rightward year by year from 2019 to 2022, with the curve peak leveling off and the curve width remaining roughly the same. This indicates that the overall development level of the provincial institutional environment in China is on an upward trend. The average difference in the institutional environment index between provinces has not changed significantly. In Figure 9 and Figure 10, the main body of the density contour is mainly distributed around the positive 45-degree line. This suggests that the provincial institutional environment index in China does not vary much from year t to t + 2. The main body of the density contour peaks at approximately x = 9. During 2018 to 2022, the provincial institutional environment index in China shows an overall upward trend, but there is no significant or obvious improvement.

4.1.2. Spatial Evolution Analysis

The natural break point classification method is used to calculate the provincial green innovation efficiency, digital government construction level, and institutional environment index by intervals. The provincial green innovation efficiency, digital government construction level, and institutional environment index are categorized into five levels: the low level, medium–low level, medium level, medium–high level, and high level.

Firstly, we consider the spatial evolution of the provincial green innovation efficiency. The sub-interval statistics are shown in

Table 2. Statistical results regarding provincial green innovation efficiency by interval.

Year	0–0.2925	0.2926–0.5850	0.5851–0.7925	0.7926–1.0000	>1.0001
2018	10	9	1	4	7
2020	5	15	2	6	3
2022	2	16	2	0	11

Unit: pcs.

. The overall average value is 0.585 from the year 2018 to 2022. In 2018, only 12 provinces, including Guangdong and Beijing, had a medium level or above for green innovation efficiency. Ten provinces, including Heilongjiang and Jilin, had a low level of green innovation efficiency. Nine provinces, including Sichuan and Qinghai, had a low–medium level of green innovation efficiency. This is because, at that time, China was in the initial exploration stage of green innovation development, with limited innovative technologies and imperfect policy systems. Therefore, at that time, the general level was low. With the continuous development of green innovation technology and the improvement of the policy system, the number of provinces with a low level of green innovation efficiency was reduced to five in 2020. The overall level of development was improved. At the sub-regional level, clusters with medium to high levels of green innovation efficiency have been formed in Northern, Eastern, and Southern China, represented by parts of Beijing–Tianjin–Hebei, the Yangtze River Delta, and the Pearl River Delta, respectively. The green innovation efficiency of the central and western provinces has improved significantly. In 2022, the number of low-level provinces was reduced to two, namely Henan and Tibet. The number of high-level provinces increased to 11. Specifically, high-level zones consisting of Shanghai and Jiangsu, Beijing and Tianjin, and Guangdong and Fujian were initially formed. In general, the green innovation efficiency gradually improved during the study period, the provincial gap gradually narrowed, and the trend of spatial agglomeration was gradually more obvious. The green innovation efficiency generally followed a rising gradient ‘from west to east and from north to south’.

Secondly, we consider the spatial evolution of digital government construction. The sub-interval statistics are shown in

Table 3. Statistical results regarding digital government construction level by interval.

Year	0–73.4720	73.4721–85.7140	85.7141–89.8270	89.8271–93.9400	>93.9401
2018	3	19	3	4	2
2020	1	16	6	5	3
2022	0	8	8	6	9

Unit: pcs.

. The overall average value is 85.714 from the year 2018 to 2022. In 2018, nine provinces, including Beijing and Jiangsu, had a medium and above level of provincial green innovation efficiency, while other provinces were classed as medium–low- and low-level zones. This is because, in 2018, China was still in the early stage of digital government construction. The level of digital technology and digital governance was not high. It was developed only in some developed provinces along the eastern coast, as well as in very few western provinces and Beijing. The overall level was low. As the level of digital technology and digital governance continued to develop, the number of provinces with a medium level and above increased to 14 in 2020. The number of low-level provinces was reduced to one. The overall level of digital government construction improved. At the sub-regional level, the levels in the central and western provinces improved significantly. In 2020, the number of provinces with medium and above levels increased to 23. Provinces with levels below the medium level were mainly located in North China, Northeast China, and Northwest China. Specifically, a high-level zone consisting of Shanghai, Jiangsu, and Zhejiang was initially formed. Eight provinces, including Tianjin, Heilongjiang, and Gansu, remained stable, representing the main distribution areas for low to medium levels. In general, the level of digital government construction in China during the study period gradually improved, the provincial gap gradually narrowed, and the trend of spatial agglomeration was gradually more obvious.

Thirdly, we consider the spatial evolution of the institutional environment. The sub-interval statistics for the institutional environment are shown in

Table 4. Statistical results regarding institutional environment index by interval.

Year	0–3.9830	3.9831–7.9660	7.9661–9.0520	9.0521–10.1380	>10.1381
2018	1	13	10	5	2
2020	1	13	10	3	4
2022	1	10	2	10	8

Unit: pcs.

. The overall average value of the institutional environment in 31 provinces in China was 7.966 from the year 2018 to 2022. During the period from 2018 to 2020, the number of provinces with medium and lower levels in the institutional environment index remained essentially unchanged. The total number of provinces with medium and higher levels remained unchanged. The number of high-level provinces increased to four, with Jiangsu and Zhejiang being among these. The overall level of the institutional environment improved insignificantly. With economic development and policy improvement, the number of provinces with medium and above levels increased to 18 in 2022. The overall level of the institutional environment improved significantly. At the sub-regional level, the institutional environment index in the central and western provinces improved significantly. A medium–high-level zone comprising Beijing, Tianjin, and Hebei was initially formed. A high-level zone consisting of Guangdong and Fujian was initially formed. The high-level zone consisting of Shanghai, Jiangsu, and Zhejiang was essentially finalized. In general, the overall level of the institutional environment during the study period showed an upward trend, and the trend of spatial agglomeration was gradually more obvious.

4.2. Analysis of Empirical Results

4.2.1. Descriptive Statistics of Variables

Table 5. Descriptive statistics.

Variable Name	Symbol	Obs	Mean	Std. Dev.	Min	Max
Institutional environment	IE	155	7.967	2.115	1.574	12.672
Digital government construction	DG	155	85.715	6.838	61.230	99.590
Provincial green innovation efficiency	GIE	155	0.585	0.308	0.034	1.294
Economic development level	ED	155	73,871.077	33,262.628	31,336.000	190,313.000
Technology investment effort	TI	155	0.589	0.904	0.150	11.170
Tax burden level	TB	155	7.468	2.660	3.540	19.230
Social consumption level	SC	155	0.380	0.070	0.183	0.538
Foreign direct investment level	FDI	155	0.015	0.013	0.000	0.073
Unemployment rate	UE	155	2.635	1.109	0.200	4.600

shows the descriptive statistics for each variable.

4.2.2. Baseline Results

To test the impact of digital government construction and the institutional environment on the provincial green innovation efficiency in 31 provinces in China, this study uses the Stata 18 software to conduct a baseline regression analysis. Firstly, this study conducts an F-test and Hausman test. The F-value is 34.58 and the p-value is less than 0.01, passing the significance test. The Chi2 statistic is 11.2, corresponding to a p-value of 0.0824, and the Hausman test result is not significant. Therefore, the random effect model should be used. Secondly, the baseline regression analysis is performed. The baseline results are shown in

Table 6. Regression results of the baseline models.

	Explained Variable: GIE
	Model 1	Model 2	Model 3	Model 4	Model 5
DG	0.008 ***		0.007 ***		0.007 ***
	(3.13)		(2.60)		(2.64)
IE		−0.009		−0.027 *	−0.028 *
		(−0.61)		(−1.77)	(−1.89)
ED			0.000 ***	0.000 ***	0.000 ***
			(3.72)	(4.48)	(4.13)
TI			0.005	0.003	0.006
			(0.48)	(0.30)	(0.61)
TB			−0.010	−0.014	−0.013
			(−1.11)	(−1.42)	(−1.36)
SC			−0.125	0.072	0.081
			(−0.37)	(0.20)	(0.23)
FDI			−0.544	−0.324	−0.581
			(−0.41)	(−0.24)	(−0.44)
UE			0.017	0.020	0.019
			(0.90)	(1.04)	(1.00)
_cons	−0.054	0.643 ***	−0.123	0.469 **	−0.039
	(−0.26)	(5.17)	(−0.46)	(2.41)	(−0.14)
Control variables	NO	NO	YES	YES	YES
Observations	155	155	155	155	155

z-statistics in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1.

. Models 1 to 5 reflect the results of one-by-one regression and full-sample regression.

As can be seen from the baseline results, firstly, the regression coefficient of digital government construction (DG) is significantly positive at the 1% level in Model 1. This indicates that the improvement in the level of provincial digital government construction has a positive effect on the provincial green innovation efficiency. Hypothesis 1 is preliminarily verified. In Model 2, the impact of the institutional environment (IE) on the provincial green innovation efficiency is not significant. Secondly, after adding the control variables, the regression coefficient of digital government construction is significantly positive at the 1% level in Model 3. In Model 4, the regression coefficient of the institutional environment is significantly negative at the 10% level. In Model 5, the regression coefficient of digital government construction is significantly positive at the 1% level, and the regression coefficient of the institutional environment is significantly negative at the 10% level.

By comparing Models 1 and 3, it can be seen that the regression coefficient of digital government construction changes somewhat after adding the control variables, but it is still significantly positive at the 1% level. This indicates that the inclusion of the control variables mitigates the interference of omitted variable bias in the estimation of causal effects to some extent. By comparing Models 2 and 4, it can be seen that the information reflected by the institutional environment variable may be a subset of factors influencing the provincial green innovation efficiency. The use of the institutional environment variable may lead to the omission of some important variables, inevitably increasing the variance of the error term. The variance becomes smaller and the statistic becomes significant after adding the control variables, i.e., the inclusion of the control variables mitigates the interference of omitted variable bias in the causal effect estimates. This result indicates, to some extent, that the control variables may contain other variables that affect the provincial green innovation efficiency. Meanwhile, the regression coefficients of the institutional environment in Models 2, 4, and 5 are negative. Referring to the study of Han XF et al. [71], as the regression coefficients are negative, it indicates that, in this period, the institutional environment is still in the stage of hindering any improvement in the provincial green innovation efficiency. There is a non-linear relationship between institutional environmental improvement and green innovation. The development degree of the provincial institutional environment has not formed a scale effect to promote provincial green innovation efficiency. However, in the long term, the synergy between the institutional environment and green innovation will become more significant, with high-quality economic development, perfect market development, more sound legal and other institutional systems, more rationalized government and market relations, and a higher level of digital governance by the government [29].

In terms of the control variables, the regression coefficient for the economic development level (ED) is significantly positive at the 1% level. This indicates that the higher the economic development level of a province, the higher the green innovation efficiency. The technology investment effort (TI), social consumption level (SC), tax burden level (TB), foreign direct investment level (FDI), and unemployment rate (UE) are not significant in Models 3, 4, and 5.

4.2.3. Heterogeneity Test

Firstly, the geographic location heterogeneity test is performed. In order to verify the heterogeneous effects, in this study, based on the division of Eastern, Central, and Western China by the National Development and Reform Commission of China, group regressions are conducted for 31 provinces in China based on the different locations of each province.

Table 7. Group regression results for heterogeneity of geographic location.

	Explained Variable: GIE
	Model 1 Eastern China	Model 2 Central China	Model 3 Western China
DG	0.010	0.023	0.001
	(1.40)	(1.43)	(0.23)
IE	−0.058	0.046	0.040
	(−1.34)	(0.29)	(1.03)
ED	0.000 ***	0.000	0.000 ***
	(2.74)	(1.03)	(2.91)
TI	−0.065	0.051 **	0.304 *
	(−0.67)	(2.26)	(1.68)
TB	−0.014	0.090	−0.011
	(−0.88)	(1.06)	(−0.39)
SC	0.506	−0.689	−2.716 ***
	(0.53)	(−0.70)	(−3.24)
FDI	0.680	4.061	−11.610
	(0.22)	(1.03)	(−1.38)
UE	0.047 ***	0.125	0.022
	(4.47)	(1.21)	(0.34)
Constant	−0.313	−2.874 *	0.744 *
	(−0.48)	(−1.79)	(1.81)
Observations	60	45	50

z-statistics in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1.

shows the group regression results regarding the heterogeneity of the geographic location.

The spatial distribution of the provinces in China varies, and Eastern, Central, and Western China have different regional advantages. The regression results show that digital government construction does not significantly promote provincial green innovation efficiency in the provinces in Eastern, Central, and Western China. Combined with the spatial and temporal distribution pattern of digital government construction in each province, the level of digital government construction in the provinces in Central and Western China is lower than that in Eastern China. The provinces in Eastern China are geographically well located, with a strong export-oriented economy and economic development. The driving effect of the improvement in the digital government construction level on the enhancement of the provincial green innovation efficiency in Eastern China is not significant. Central and Western China do not have a significant geographic advantage compared with Eastern China. The extents of the export-oriented economy and economic development in Central and Western China are not high, and the level of digital government construction is relatively low. The low level of digital government construction inhibits the improvement in the green innovation efficiency in the provinces in Central and Western China to a certain extent.

Analyzing the regression results for the control variables, it can be concluded that the impact of the economic development level of the provinces in Eastern and Western China on the green innovation efficiency is significantly positive at the 1% level. The economic development level of provinces in Central China does not have a significant effect on the provincial green innovation efficiency. This indicates that the economic development of provinces in Eastern and Western China has formed a scale effect. The synergy between economic growth and green innovation is significant. The impact of technology investment effort on the provincial green innovation efficiency is significantly positive at the 5% and 10% levels for provinces in Central and Western China, respectively, while the impact is not significant for provinces in Eastern China. The provinces in Eastern China have relatively stronger technology investment effort, which can cause more capital and other green innovation factors to aggregate to the green technology industry. However, excessive factor agglomeration generates competitive effects, which intensifies market competition and inhibits the improvement in the provincial green innovation efficiency. The impact of the social consumption level on the provincial green innovation efficiency in Western China is significantly negative at the 1% level. This indicates that the social consumption level in Western China is still a blocking factor regarding the development of green innovation. The impact of the unemployment rate on the provincial green innovation efficiency in Eastern China is significantly positive at the 1% level. The impact of the unemployment rate on the provincial green innovation efficiency in provinces in Central and Western China is not significant. This is because the provinces in Eastern China have a higher level of economic development and labor absorption than those in Central and Western China. However, green innovation belongs to the service sector or green high-tech industry, and the demand for human resources is high, requiring highly skilled and knowledgeable labor forces. Thus, the demand for labor in the field of green innovation is relatively homogeneous, excluding the general labor force that does not have green high-tech expertise. Therefore, the impact of the unemployment rate in Eastern China is significantly positive.

Secondly, we conduct a development degree heterogeneity test. The 31 provincial governments in China are divided according to the provincial digital government development index described in the ‘Digital Government Development Index Report’ published by the Data Governance Research Center at Tsinghua University. In terms of the development gradient, the 31 provincial governments in China are categorized into five types with different levels of development. They are the leading type (more than 70 points), quality type (65–70 points), characteristic type (60–65 points), development type (50–60 points), and catching-up type (less than 50 points). Among them, Shanghai, Beijing, and Zhejiang belong to the leading type. Guangdong, Sichuan, Anhui, Shandong, Jiangsu, and Fujian belong to the quality type. Tianjin, Chongqing, Hainan, Guizhou, Hunan, and Hubei belong to the characteristic type. Shanxi, Hebei, Jiangxi, Henan, Shaanxi, Guangxi, and Inner Mongolia belong to the development type. Liaoning, Yunnan, Gansu, Heilongjiang, Jilin, Qinghai, Ningxia, Tibet, and Xinjiang belong to the catching-up type.

Table 8. Group regression results for heterogeneity in development degree.

	Explained variable: GIE
	Model 1 Leading Type	Model 2 Quality Type	Model 3 Characteristic Type	Model 4 Development Type	Model 5 Catch-Up Type
DG	0.004	0.013	−0.011 ***	0.006	0.017
	(1.56)	(1.42)	(−3.70)	(0.64)	(1.53)
IE	−0.172	−0.153	−0.006	−0.008	0.058 **
	(−1.39)	(−1.47)	(−0.29)	(−0.10)	(2.56)
ED	0.000 ***	0.000 ***	0.000 ***	0.000 **	0.000
	(3.25)	(4.22)	(7.63)	(2.40)	(0.27)
TI	0.005	0.006	0.019	1.989 ***	0.473 ***
	(0.71)	(0.85)	(0.61)	(5.26)	(2.77)
TB	0.091	0.086	−0.032 **	0.009	0.056
	(1.57)	(1.32)	(−2.30)	(0.49)	(0.97)
SC	4.129 *	3.029 *	−0.961 ***	−5.360 ***	−4.100 ***
	(1.92)	(1.88)	(−3.30)	(−9.35)	(−4.93)
FDI	10.136 **	12.722 **	−2.292 *	−4.952	−14.085 ***
	(2.32)	(2.18)	(−1.77)	(−0.68)	(−2.65)
UE	−0.043	−0.039	0.032	0.126	0.041
	(−0.39)	(−0.55)	(1.46)	(1.32)	(0.43)
Constant	−1.258	−2.156 ***	1.654 ***	0.573	−0.416
	(−0.45)	(−3.30)	(2.77)	(0.40)	(−0.42)
Observations	15	30	30	35	45

z-statistics in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1.

shows the group regression results regarding the heterogeneity in the development degree.

As shown in Table 8, there is a significant difference in the impact of digital government construction and the institutional environment on the provincial green innovation efficiency across provinces with different levels of development. In terms of the core independent variable, the impact of digital government construction on the provincial green innovation efficiency of provinces of the characteristic type is significantly negative at the 1% level. The impact of digital government construction on other types of provinces, besides the characteristic type, is insignificant and positive. The development of provinces of the characteristic type is related to their political statuses and leading industries. Among them, Tianjin and Chongqing are municipalities that are directly under the Chinese Central Government, which have obvious advantages in terms of their locations and political and economic statuses. Tianjin and Chongqing are more attractive in terms of factors such as labor and capital. This triggers factor agglomeration and intensifies the market competition. Tianjin and Chongqing do not lead the country in digital governance like the other two municipalities in China, Beijing and Shanghai. The level of digital governance in Tianjin and Chongqing is not sufficient to adequately cope with excessive factor concentration. Moreover, the synergy between digital government construction and green innovation has not been completely achieved. This in turn inhibits the improvement of the provincial green innovation efficiency. Hainan and Guizhou take tourism as the leading industry, mainly relying on the traditional tourism model, and the industry chain is relatively singular. The development of digital tourism is insufficient, which in turn inhibits the development of green innovation. The pillar industries in Hunan are non-metallic mineral products and other electronic equipment manufacturing industries. The industrial structure is unbalanced, and some cities have not eliminated the dependence on the development path of manufacturing, which hinders the development of green innovation. The pillar industries in Hubei are information technology, high-end equipment, advanced materials, and other emerging industries. The development degree of innovative industries is high. The scale effect of green innovation development is obvious. The market competition effect is also obvious. The level of digital governance is insufficient to adequately cope with excessive factor agglomeration, instead inhibiting the improvement of the provincial green innovation efficiency. Digital government construction in leading, quality, developing, and catching-up provinces can promote the provincial green innovation efficiency, but the driving effect on green innovation is not significant.

In terms of the mediating variables, the regression coefficient for provinces of the catching-up type is significantly positive at the 5% level. This shows that the improvement of the institutional environment in provinces of the catching-up type has a positive driving effect on the improvement of the provincial green innovation efficiency. In terms of the control variables, the level of economic development of provinces of the leading, quality, developing, and characteristic types has a significantly positive effect on the provincial green innovation efficiency. The impact of technology investment effort on the provincial green innovation efficiency in provinces of the developing and catching-up types is significantly positive at the 1% level. The impact of the tax burden level on the provincial green innovation efficiency in provinces of the characteristic type is significantly negative at the 5% level. This indicates that, in this phase, the impact of the tax burden level on the improvement of the provincial green innovation efficiency is still in the blocking stage. However, in the long term, as the digital government construction level and the institutional environment improve, the synergy between the tax burden level and green innovation will become more significant. Provinces of the leading and quality types have higher levels of social consumption and foreign direct investment. The impact of the social consumption level on the provincial green innovation efficiency is significantly negative at the 1% level in provinces of the characteristic, developing, and catching-up types. This indicates that the social consumption level is still a blocking factor regarding the development of green innovation in provinces of the characteristic, developing, and catching-up types. The impact of the FDI level on the provincial green innovation efficiency is significantly negative at the 10% and 1% levels for provinces of the characteristic and catching-up types, respectively. This shows that the FDI level is still a blocking factor regarding the development of green innovation in provinces of the characteristic and catching-up types. This negative blocking effect is more evident for provinces of the catching-up type.

4.2.4. Robustness Test

To further validate the robustness of the main research conclusions, this study performs a robustness test by changing the measurement of the explanatory variable and deleting the sample of municipalities directly under the Central Government of China. Firstly, we change the measurement of the provincial green innovation efficiency. The DEA model is used to re-measure the provincial green innovation efficiency of 31 provinces in China and conduct robustness tests. Secondly, we delete the sample of municipalities directly under the Central Government of China. The four municipalities directly under the Central Government in China have more advantages in terms of the government governance capacity, market demand, market scale, and human capital compared to other provinces. The municipality samples are removed and the robustness test is conducted again.

Models 1 and 2 in

Table 9. Robustness test results.

	Explained Variable: GIE
	Model 1 Change the Measurement of Provincial Green Innovation Efficiency	Model 2 Delete the Sample of Municipalities Directly Under the Central Government of China
DG	0.002 *	0.006 **
	(0.750)	(2.487)
IE	−0.034 ***	−0.014 *
	(−2.584)	(−1.026)
ED	0.000 ***	0.000 **
	(3.272)	(2.179)
TI	0.072 ***	0.006
	(5.942)	(0.599)
TB	0.048 ***	−0.003
	(5.765)	(−0.294)
SC	−0.906 ***	0.009
	(−3.180)	(0.026)
FDI	0.927	0.409
	(0.644)	(0.308)
UE	0.010	0.002
	(0.986)	(0.251)
_cons	0.331	−0.055
	(1.592)	(−0.197)
N	155	155

z-statistics in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1.

reflect the regression results after changing the measurement of the provincial green innovation efficiency and removing the sample of municipalities, respectively. The regression results in Table 9 are essentially consistent with the original baseline regression results. The new regression results still have strong explanatory power for the research findings. Therefore, the research findings are robust.

4.2.5. Mechanism Analysis

This paper empirically tests the mediating role of the institutional environment with reference to the traditional mediation effect analysis [68]. The mechanism analysis is conducted using the three-stage method of regression. The regression results are shown in

Table 10. Regression results for the mediating effect of the institutional environment.

	Explained Variable
	GIE	IE	GIE
	Model 1	Model 2	Model 3
DG	0.007 * (0.0038)	0.073 *** (0.021)	0.007 *** (2.64)
IE			−0.028 * (−1.89)
Control variables	Yes	Yes	Yes
N	155	155	155
Wald chi2	65.64	60.59	70.38

z-statistics in parentheses; *** p < 0.01, * p < 0.1.

. Models 1, 2, and 3 correspond to Equations (3), (7), and (8), respectively.

According to Table 10, Model 1 shows that the impact of digital government construction on the provincial green innovation efficiency is significantly positive at the 10% level. Therefore, Hypothesis 1 is verified. Model 2 shows that the impact of digital government construction on the institutional environment is significantly positive at the 1% level. Thus, Hypothesis 4 is confirmed. Model 3 shows that, after adding the mediating variable, the impact of digital government construction on the provincial green innovation efficiency is significantly positive at the 1% level. The impact of the institutional environment on the provincial green innovation efficiency is significantly negative at the 10% level. Therefore, Hypothesis 3 is confirmed. This study has completed the three-step causal stepwise regression put forward by Baron and Kenny [68]. The research results show that the institutional environment is the mediating variable between digital government construction and provincial green innovation efficiency. Thus, Hypothesis 5 is confirmed. Figure 11 shows this mediating mechanism.

Comparing the sign of ${\mathsf{\rho }}_2\times {\mathsf{\beta }}_0$ with that of $\mathsf{\rho }$ reveals that the sign of the former is the opposite of the latter. This suggests that there exists a competitive mediating process [72]. Competitive mediation can be understood as the combined effect of the costs and benefits of the antecedent variable [73]. Provincial green innovation efficiency improvements both create an inherent demand for the development of the institutional environment and potentially increase the cost of institutional environment improvement and development. When such costs outweigh the benefits, the institutional environment has a competitive mediating effect in this chain of relationships [74]. This study reveals that the moderating process is a competitive mediating process where positive and negative effects coexist. Specifically, due to the heterogeneity of the regions and provinces, the institutional environment does not only play a positive moderating role, as in provinces of the catching-up type, such as Jilin and Liaoning. The institutional environment may also lower the positive impact of digital government construction on the improvement of the provincial green innovation efficiency. For example, in some provinces, the market scale becomes larger and the market factors are aggregated, while the digital government construction level is not sufficient to cope with the excessive factor aggregation. In this case, the larger market scale may cause the institutional environment to negatively affect green innovation. Competitiveness in the mediating process stems from the fact that institutional environment improvement intensifies the competition for resources and inputs required for digital government construction and digital governance, as well as the competition for market scale and market resources.

5. Conclusions and Policies

5.1. Conclusions

This study uses data from 31 provinces in China from 2018 to 2022. This study is the first to apply MATLAB research tools, measuring methods, and a geographic disciplinary approach together to study the relationship between digital government construction, the institutional environment, and provincial green innovation efficiency. The research conclusions are as follows.

Firstly, we consider the spatiotemporal evolution characteristics of digital government construction, the institutional environment, and provincial green innovation efficiency. In terms of temporal evolution, the overall development level of the provincial green innovation efficiency in China is on an upward trend. There is more room for improvement. The average difference between the provinces is reduced. The level of provincial digital government construction in China is on an upward trend in general. The average inter-provincial differences generally show a narrowing trend. The provincial institutional environment index in China shows an overall upward trend, but there is no significant or obvious improvement. The average difference between provinces does not change significantly. In terms of spatial evolution, the trend of the spatial agglomeration of the provincial green innovation efficiency is gradually becoming more obvious. The provincial green innovation efficiency generally follows a rising gradient ‘from west to east and from north to south’. The trend of the spatial agglomeration of digital government construction and the institutional environment becomes gradually apparent.

Secondly, digital government construction positively affects the provincial green innovation efficiency. Digital government construction positively affects the institutional environment. The institutional environment negatively affects the provincial green innovation efficiency. The institutional environment mediates the relationship between digital government construction and the provincial green innovation efficiency. The empirical conclusions still have strong explanatory power after the robustness test. In the geographic location heterogeneity test, digital government construction does not significantly promote provincial green innovation efficiency in the provinces in Eastern, Central, and Western China. This is because the provinces in Eastern China are geographically well located, with a strong export-oriented economy and economic development. The driving effect of the improvement in the digital government construction level on the improvement in the provincial green innovation efficiency in Eastern China is not significant. Central and Western China do not have a significant geographic advantage compared with Eastern China. The extent of the export-oriented economy and economic development in Central and Western China is not high, and the level of digital government construction is relatively low. The low level of digital government construction inhibits the improvement of the provincial green innovation efficiency in provinces in Central and Western China to a certain extent. In the development degree heterogeneity test, digital government construction has an inhibitory effect on the improvement of the provincial green innovation efficiency in provinces of the characteristic type. An improvement in the institutional environment in provinces of the catching-up type positively affects the provincial green innovation efficiency.

Thirdly, the institutional environment plays a competitive mediating role in the relationship between digital government construction and provincial green innovation efficiency. Based on the theoretical model of mixed embeddedness, this study proposes the hypothesis that the institutional environment negatively affects the provincial green innovation efficiency. It also finds that the moderating process of the institutional environment between digital government construction and provincial green innovation efficiency is a competitive mediating process in which positive and negative effects coexist. The new analytical framework consisting of digital government construction, the institutional environment, and green innovative efficiency, constructed on the basis of these new research findings, is a theoretical breakthrough and has far-reaching practical significance.

5.2. Theoretical Contributions

Firstly, in the existing literature on the relationship between digital government construction and provincial green innovation efficiency, the main discussion considers the government’s static governance capacity. Specifically, it explores the impact of the integration of digital technology and government governance on green innovation and the organization’s ability to formulate and implement policies. This study finds that the impact of digital government construction on green innovation has achieved an organizational capability breakthrough in practice. The organizational use of technology is combined with elements of the institutional environment. The resulting dynamic circle of information and synergistic development enables more effective governance decisions and improves the governance capacity and innovation efficiency. Digital governance, characterized by organizational capacity breakthroughs, will resonate synergistically with the institutional environment in a way that has not been covered by the available literature.

Secondly, this study is the first to explore the institutional environment as a moderator of digital government construction and provincial green innovation efficiency, which fills this research gap. Moreover, this study is based on Chinese developmental realities and examines the relationship between the three factors. It is found that the institutional environment not only positively affects the provincial green innovation efficiency, as already found in the existing literature, but also negatively affects it. Moreover, the institutional environment plays a competitive mediating role in the relationship between digital government construction and provincial green innovation efficiency. This research finding constitutes a theoretical breakthrough. Meanwhile, it also provides a basis for theoretical analysis and empirical testing to construct a new theoretical framework based on the actual situation in China. This could contribute to a scientifically accurate basis for sustained improvements in the provincial green innovation efficiency.

Thirdly, based on the theoretical model of mixed embeddedness, this study proposes the hypothesis that the institutional environment negatively affects the provincial green innovation efficiency. It also finds that the moderating process of the institutional environment between digital government construction and the provincial green innovation efficiency is a competitive mediating process. The existing theories of embeddedness are mostly one-dimensional, single-level theoretical systems [75,76]. The new analytical framework consisting of digital government construction, the institutional environment, and green innovative efficiency, constructed on the basis of these new research findings, is a theoretical breakthrough and has far-reaching practical significance.

Fourthly, a multi-level analysis framework of provincial green innovation efficiency is constructed. In the field of green innovation, this multi-level analytical framework has not attracted sufficient attention or widespread application in the academic circle. This paper provides a multi-level theoretical framework that is conducive to subsequent research, deepening the study of the source of green innovation.

Finally, regarding the development of digital government research, all three of the Minnowbrook conferences have impacted the study of public administration in highly innovative ways. Among them, the themes of the second Minnowbrook conference are closely related to the research themes of this study, and retracing the path of Minnowbrook will expand the research significance of this study. The themes of the second Minnowbrook conference are summarized as follows: more skilled professionals; the growing importance of productivity and performance evaluation; and more relevance to the mainstream social sciences and to the positivist or Simon’s view, etc. An important outcome of the second Minnowbrook conference was the emergence of the governance pathway of new public management. The main aim of the second Minnowbrook conference was to emphasize new public services, efficiency, and entrepreneurial operations brought about by market mechanisms in achieving social goals while upholding social equity. While the first Minnowbrook conference shifted traditional public administration research from management to constitutionalism and from technical rationality to value rationality, the second Minnowbrook conference shifted from constitutionalism to management and technical rationality, emphasizing technical and process issues. The relationship between digital government construction, the institutional environment, and the provincial green innovation efficiency studied in this research essentially emphasizes the inter-construction of market mechanisms and digital government construction, as well as the importance of emerging concepts such as digital government construction and digital governance in promoting democratic administration and fairness and justice. This is in line with the theme of the second Minnowbrook conference. The dialectical unity and dynamic intermingling of digital government construction and institutional environment integration, and the coupling of the two with green innovation in a complex framework, is, to some extent, a development and extension of the themes of the Minnowbrook conference over the ages. To some extent, this study further develops the themes and concerns of the second Minnowbrook conference. In this sense, by retracing the Minnowbrook path, the theoretical and empirical research conducted in this study develops, to some extent, the themes of the second Minnowbrook conference and the spirit of the Minnowbrook conference, which is dedicated to academic prosperity and progress. Meanwhile, the construction of a localized discourse system supported by the Chinese public administration must be based on the Chinese reality. After the introduction of the market mechanism in China, the market has been playing an important role in social development and pursuing efficiency. Meanwhile, the rise of new concepts such as digital governance has led to the rethinking of the government–market relationship. The new research framework built in this study by combining the government and market offers some theoretical contributions to the impact of new forms of development, such as green development. This study develops on the above aspects.

5.3. Practical Implications

Based on the findings, this study suggests the following four policy recommendations to better exploit the important role of digital government construction and the institutional environment in enhancing the efficiency of green innovation.

Firstly, it is necessary to continue to promote digital government construction to enhance the governance level and capacity. On the one hand, the holistic governance concept should be used to build a scenario framework for the integration of digital thinking and modern government governance concepts. On the other hand, it is necessary to promote governance reform through digital government transformation and drive the realization of green development, as well as studying and formulating standards for digital government construction. A system of digital government construction standards and norms, containing multi-level concepts such as the overall architecture, government services, collaborative office work, and data sharing, needs to be established. Priority should be given to pilot projects in the areas of government services, collaborative office work, internal communications, decision-making support, and interactive interfaces. It is necessary to gradually incorporate digital government construction standardization into the content of the comprehensive reform pilot construction of standardization, thus offering support for green innovation development. Moreover, it is necessary to create a favorable environment for participation in digital governance and actively promote stakeholders’ participation in digital government construction. In this way, we can jointly realize green and innovative development. The specific practices are as follows. Firstly, they include promoting cooperation between government and enterprises; utilizing the technical advantages of technology research and development units, internet enterprises, and network service operators; adopting a variety of public–private cooperation methods; and attracting research institutions and enterprises to participate in the overall planning of digital government construction. Secondly, they include supporting government–citizen interaction; strengthening interactive functions such as online feedback and evaluation; exploring various ways to enable the public and enterprises to participate in public decision-making that affects their own interests; and optimizing the institutional environment to enhance the efficiency of green innovation. The first set seeks to continue to transform government functions and improve the ecological environment for innovation. The second set seeks to deepen the factor market reform and stimulate the green innovation vitality of innovation subjects. The third aim is to enhance the policy incentives for the transformation of green scientific development and to create a favorable atmosphere for green innovation. Moreover, it is necessary to give play to the synergy effect of digital government construction and the institutional environment and form a joint force to promote green innovation efficiency. The government should actively promote multi-subject green innovation to co-create, share, and fully utilize the synergy effect. On the one hand, this can be achieved by cultivating professional and technical talent. The development of digital government construction and green innovation require the support of a large number of professional and technical personnel. Chinese local governments can support high-level talent through housing, settlement, children’s schooling, and other policies that are conducive to the display of talent. On the other hand, it is necessary to create a number of new innovative subjects with strong growth, potential competitiveness, and leading roles, maximizing the innovation efficiency of innovative subjects. Fourthly, the government should formulate and implement differentiated policies to encourage green innovation. Considering the development heterogeneity of provinces in different regions and provinces with different levels of development, the local governments of China should adopt precise policies according to the local conditions. It is necessary to improve the level of digital government construction, optimize the institutional environment, improve the efficiency of green innovation, and make the three synergistic.

In addition, this study provides practical insights for policymaking in developing economies other than China. Firstly, this study shows the complexity and the systematic and dynamic nature of digital government construction, the institutional environment, and green innovation. Developing economies other than China also need to perform objective analyses and make scientific decisions based on their own technological bases, institutional endowments, innovation conditions, and other factors. Secondly, compared with developed economies, most developing economies do not have a high level of economic development. The level of digital government construction needs to be improved. The institutional environment needs to be optimized. However, along with the accelerated integration of emerging technologies and government governance, data governance and smart empowerment have become new topics. Digital platforms are playing a key role in connecting the government and society. While developing economies make full use of digital technology to develop digital governments and carry out digital governance, they need to consider the role of self-organization and other social forces in the digital government ecosystem. Policy decisions should be made with full input from the public, self-organizations, and other social forces. Thirdly, this study examines the relationship between digital government construction, the institutional environment, and green innovation from a more stereoscopic and dynamic perspective. The findings of this study offer several insights. For instance, the heterogeneity analysis reveals that other developing economies need to formulate different policies based on the specific situations in different regions of the country. Moreover, the development levels of some developing economies are insufficient to provide adequate support for digital government development, institutional environment optimization, and green innovation. At this point, it is vital to consider the potential risks of digital government construction, institutional environment optimization, and green innovation development for new forms of development, such as the digital economy. Fourthly, the application of the research findings to different aspects of digital governance needs to be analyzed under different countries in developing economies. In the course of social development, the breadth and depth of digital governance and innovative development continue to expand. The various subjects involved in digital governance and innovative development have both shared interests and conflicts. Clarifying the interaction mechanism, the power structure, and the interests of each subject will help to achieve a dynamic and refined understanding of digital governance and innovative development.

5.4. Deficiencies and Prospects

There are several limitations of this study that require further research in the future. Firstly, this paper examines the relationship between the digital economy, institutional environment, and provincial green innovation efficiency based on 31 provinces in China. It has not examined more detailed administrative units, such as prefecture-level cities and county-level cities in China. In the future, more detailed units in China can be studied to expand the theoretical framework constructed in this paper regarding the relationship between digital government construction, the institutional environment, and provincial green innovation efficiency.

Secondly, in a deeper sense, the current development of digital government construction in China emphasizes the transformation of government services, governance, and government operation modes. From a practical standpoint, digital government governance, institutional restructuring, and the evolution of innovation ecology are intertwined and interact in an open, dynamic, and multidimensional spatial and temporal context. The typical digital government construction practices, such as ‘receiving and handling complaints’, are more in line with the goal of promoting the reengineering of government governance processes and creating a collaborative and high-efficiency digital government service system. However, with economic and social development, the future of digital governments will see more complex and dynamic characteristics. Subsequent studies can further develop the digital government construction evaluation index system according to the actual development situation and the Chinese context and conduct more in-depth discussions.

Thirdly, the interactive mechanisms of digital government transformation, institutional environment adaptation, and green innovation development need to be further analyzed qualitatively. The institutional environment shapes the social context of digital governments and technological development. Digital government construction and technological advances in turn force institutions to break through path dependency. The intertwined impacts of the two lead to the dynamic evolution of green innovation. However, the above process is the result of a combination of political, economic, cultural, and social factors. Based on the data, methods, and models, this study empirically investigates digital government construction, the institutional environment, and provincial green innovation efficiency. The results of the empirical analysis are based on real and effective data. No qualitative analysis or case study on the relationship among the three has been conducted. Qualitative analysis will help to identify the interaction mechanism of the three from a more stereoscopic and dynamic perspective. In addition, the future picture of the synergistic development of digital government construction, the institutional environment, and green innovation needs to be systematically understood. The rapid development of digital technology has brought about the reconstruction of the government’s governance mode and governance system. The institutional environment has also been gradually improved. Green innovation is accelerating. The synergy of the three not only implies the joint drive of technology, organization, systems, and innovation, but also brings about the systematic reshaping of national governance. Based on this understanding, future research is needed on important topics such as digital government development and the cultivation of new, high-quality productivity. Theoretical research should be further translated into practical support for national governance.

Fourthly, this paper builds a theoretical framework based on the Chinese context and Chinese data, but no international comparative study has been conducted. In the future, studies can be based on different countries and regions, focusing on the differences between them in terms of the technological base, institutional endowment, social culture, etc., and their impacts. Comparative research could help to reveal the dialectical unity among universal rules and specific factors. In turn, it will provide a useful reference to promote the construction of digital governments and the development of green innovation according to the local conditions.