Modeling Relations Between Digital Economy and Regional Innovation: A Multi-Transmission Effect from Environmental Performance

Abstract

Using the digital economy as a new carrier to drive innovation has become an indispensable strategic support for the progress of countries, while environmental protection is a necessary condition to ensure its sustainable development. A multi-transition framework is proposed in this article with the purpose of revealing the relations between the digital economy and regional innovation under the indirect effects of environmental performance; a mediating model and threshold effect model are used to examine empirical results considering the panel data at Chinese provincial levels. Moreover, the heterogeneity and the robustness also verified the multi-transition effect. The research findings are as follows: First, the digital economy can significantly enhance regional innovation capacity, which also shows obvious dimensional and geographical heterogeneities. Second, a mediating effect of the digital economy on regional innovation under environmental performance is considered, and environmental performance has an increasingly significant role in fostering regional innovation as the digital economy continues to grow, indicating that improving environmental performance is an essential strategy for achieving the steady release of innovation dividends. Furthermore, the above results are provided with heterogeneity and robustness under PM_2.5, COVID-19, and regional innovation quantiles. Therefore, focusing on regional innovation via environmental performance is necessary, and policymakers and professionals should pay more attention to further stimulating regional innovation potential.

1. Introduction

Nowadays, more and more countries are recognizing that innovation is a vital element for economic growth. Countries like Germany and Canada have placed great emphasis on innovation at a strategic level. In particular, regional innovation is becoming more and more significant in improving the nation’s overall competitiveness [1]. Meanwhile, the digital economy has irreplaceable advantages in changing economic structure and continuously stimulating regional innovation vitality. In particular, against the backdrop of international instability [2], it is imperative that nations prioritize the promotion of innovative growth via the digital economy to ensure survival and long-term progress. In contrast to the traditional economy, the digital economy, as a new economic form, has transcended regional boundaries and radically altered the course of economic growth [3]. It also creates a favorable technological, information, and financial environment for innovative activities [4]. Seizing the advantages of digital economic development to inject a steady stream of impetus into the improvement of innovation level will become the key path to breaking the innovation dilemma.

In addition, as several nations experience rapid economic expansion, the conflict between development and ecological health is also intensifying [5], including serious environmental pollution and a sharp decline in resource stocks [6], which are inevitable problems in promoting innovative development. At present, there is controversy about whether the progress of the digital economy is capable of improving environmental conditions. Numerous academics contended that the burgeoning digital economy could accelerate the application of environmental technologies [7], thereby protecting the natural environment, while others pointed out that it raises the demand for production resources [8], increasing the burden on the environment. So further elucidating the influence between technological progress and the environment is an important issue; it will help each country or region give priority to solving prominent environmental problems and accelerate the transformation of environmental quality from quantitative change to qualitative change to meet new development needs. Therefore, how does the digital economy stimulate regional innovation potential, and what kind of features may it present? Furthermore, will the inclusion of environmental performance change how the digital economy affects regional innovation, and what effect has environmental performance played on the aforementioned changes? After reading a large amount of the literature, it was found that existing studies cannot find sufficient evidence on the above issues.

Furthermore, China is the second-largest economy globally, and it has achieved significant progress in green development during the last decade, with air quality improvement being a landmark achievement. Specifically, since 2015, China’s average yearly concentration of PM_2.5 has dropped significantly, to 29 micrograms per cubic meter. It should be emphasized that China’s digital and green integration level continuously improved, and the digital control rate in key areas has increased from 24.6% in 2012 to 55.3% in 2021 [9], providing efficient green means for economic development. Meanwhile, there are notable disparities in the circumstances of economic advancement across China’s various areas. Studying China’s current development status and results can provide scientific experience, strategies, and development tools for more countries. Therefore, based on the provincial panel data of China from 2010 to 2022, this paper introduces environmental performance into the analytical framework of the digital economy and regional innovation, attempting to reveal the influence mechanism of environmental performance between the two, and ultimately provide policy references and evidence-based support for more countries to achieve innovative and sustainable development.

The contributions of this paper may be threefold. First, a multi-transmission path design for the digital economy to unleash innovation dividends is proposed to recognize the nonlinear and indirect impacts of environmental performance. Within the context of the air quality index, this research makes an effort to investigate the impact that environmental performance has on regional innovation from a fresh point of view. Meanwhile, the transmission effect is again verified by considering the environmental pollutant PM_2.5. Second, a theoretical framework is conducted to tap regional innovation potential through environmental performance. Differently from the existing studies, this paper independently adds environmental performance to the theoretical analysis framework of the digital economy and regional innovation, exploring complex impacts of environmental performance on regional innovation after significant achievements in green development. Third, considering that the development environment is an essential cornerstone for the construction of the economic system, it largely determines the development direction and growth potential of the digital economy. A more comprehensive measurement index of the digital economy is constructed, which incorporates the soft environment dimension, including governance, technology, and financial factors, based on digital infrastructure, industrialization, and digital integration. Considering the empirical measurements in China, we found that the dimensions that contribute the most to regional innovation are digital integration development and soft environment construction.

The paper includes six parts. The second part presents the developed mechanism considering a literature review of three key points. The model, variable, and data used to test the aforementioned mechanism are introduced in the third section. The empirical outcomes and discussions are elaborated on in the fourth part, followed by a multi-transmission effect tested from the progress of the digital economy enhancing regional innovation via environmental performance in the fifth part. The conclusion and discussion are presented in the sixth part.

2. Theoretical Background and Hypotheses

2.1. Main Theoretical Concepts

2.1.1. The Development of Digital Economy

Don Tapscott originally defined the digital economy as an economic model in which data represents the flow of information [10]. With the rapidly growing digital economy, new types of industries are continually emerging. International organizations and research institutes began discussing the digital economy from broad and specific viewpoints. Digital economy broadly refers to economic operations that carry out transactions via electronic means. However, since technology evolves quickly, other technologies often take the place of existing transaction techniques. The definition of digital economy is also changing. From a narrow perspective, the digital economy is defined by the BEA as industrial activity that relies on Internet technologies, which is more specific, but one-sided [11,12]. After that, the 2016 G20 Summit highlighted that the digital economy comprises an array of economic behaviors that utilize digital resources as a pivotal catalyst for enhancing production efficiency [13,14]. Later, based on this definition, pertinent research institutions and academics have launched discussions.

On the digital economy’s measurement criteria, there is currently no official classification standard and measurement system. Academics often describe the digital economy from its connotation. Gradually, among the many indicators, ‘digital industrialization’ and ‘industrial digitalization’ can be widely recognized as indicators reflecting its connotation, including both core activities in the digital field and businesses that depend on digital services. Additionally, digital infrastructure is also added to the analytical framework, which is both an essential foundation and a key carrier in spurring the improvement of regional development [3]. Nevertheless, the existing measurement neglects the consideration of the digital economic environment. Specifically, a good development environment is what determines the direction and quality of development; it acts as a catalyst for the digital economy in a manner that is both efficient and sustainable.

Subsequently, scholars began to empirically examine how the digital economy relates to different fields, including industrial structure [15], innovation [16], and other aspects. The continued penetration of digital technology enhances firms’ capacity to acquire information and efficiently integrate resources, resulting in reduced operational costs and improved competitiveness [17]. Moreover, it accelerates the upgrading of traditional means and achieves the optimal allocation of regional resources through data analysis and intelligent decisions [18]. In general, the digital economy has an undeniable impact on economic and social advancement, making it a crucial aspect of economic study. Therefore, the establishment of a comprehensive and systematic evaluation framework for the digital economy remains a challenge that necessitates ongoing exploration and refinement.

2.1.2. The Development of Regional Innovation

Innovation activities are essential for a region or country to enhance its overall competitiveness and ensure steady economic improvement. Peculiarly after experiencing the enormous impact that the pandemic had on the rate of economic growth, every national has taken innovation as the core of their national development strategies. Additionally, “Innovation” was proposed by Schumpeter in 1912, emphasizing that innovation reflects the capacity of a region to convert knowledge and existing factors of production into the primary driver of economic progress, and it is a significant means to occupy an economic leading position, which is essential for regional development [19].

As one of the important drivers of global innovation, China has placed significant emphasis on achieving breakthroughs in innovation potential, continuously increased investment in innovation, strengthened policy support, and actively built an innovation ecosystem to cope with increasingly fierce international competition. As for the measurement of regional innovation, scholars mainly use indicators such as knowledge breadth captured by the International Patent Classification (IPC) code, patent citations, and invention patents to evaluate innovation level [4,16]. Meanwhile, many scholars analyzed the drivers of regional innovation. The inflow of talents and technical factors brings new equipment and new technologies, which help to stimulate innovation vitality [20], promote the rapid dissemination and application of information, and thus accelerate the output of regional innovation results. The support of national strategies and policies is also crucial to improve overall innovation efficiency. Continuing to explore diverse paths to maximize regional innovation advantages is a long-term and critical task.

2.1.3. The Development of Environmental Performance

The long-term interests of human survival and development are affected by environmental conservation, making it a crucial global concern [21]. Environmental problems have become increasingly prominent in recent years, prompting scholars to conduct extensive research and discussion on how to solve environmental problems. However, it should be emphasized that there are a lack of studies to measure environmental performance at the provincial level. Most of the current studies on environmental performance focused on the corporate level. Montabon et al. [22] (2007) put forward a comprehensive index for evaluating enterprise environmental performance, which considered resource cost, project performance, and other aspects. Enterprises are required to conduct business within the constraints of environmental performance [23], and employees are also educated and cultivated with environmental awareness [24], which reflects the attitude of enterprise managers towards environmental protection and is also a commitment to society. This is crucial in attaining environmental performance goals.

Recently, the government has taken active measures to improve the ecological environment; the strict implementation of relevant green development policies has significantly improved ecological quality, which can greatly promote the environmental awareness of enterprises and social groups [25]. But it is still unknown at what stage the improvement of environmental performance is at and whether it has reached the initial expectation of the country.

2.1.4. Digital Economy, Environmental Performance, and Regional Innovation

The global innovation environment is undergoing major changes, and some countries’ innovation levels are gradually improving while there is still a large gap with others. Therefore, accelerating the release and radiation of the digital dividend to continuously tap the potential of regional innovation is still an important foothold. Academics agree that the digital economy positively impacts innovation, accelerating industrial revolutions [26] and fostering regional innovation systems via digital technology research and application [27]. It also reshapes regional innovation patterns and reduces development disparities through knowledge and technology spillovers [16], enhancing economic development quality. However, the pursuit of rapid economic growth sometimes neglects ecological protection, leading to environmental issues. Consequently, nations are exploring paths for harmonious economic and ecological development.

Existing research has not yet formed a system regarding digital economy and environmental quality, which is controversial. A few scholars argue that as the digital industry chain grows, natural resources are being used at an accelerated rate, which could further aggravate the imbalance between society and the ecosystem [8]. Other academics think that harmful emissions can be inhibited by the deep development of digital resources [28]. Specifically, the digital economy’s “green” aspect [29] accelerates the industries’ evolution to high-tech models [30], expanding the fields of resource reuse [31], leading to a reduction in high-polluting enterprises [32]. Meanwhile, strict environmental regulations force economic entities to transform, thus stimulating innovation. As a result, it is critical to further clarify how the current environment affects the course of unleashing regional innovation vitality with the digital economy.

2.2. Theoretical Model and Hypothesis

2.2.1. Relations Between the Digital Economy and Regional Innovation

The comprehensive penetration and advancement of digital technologies in social and economic production have made it a crucial tool to enhance regional innovation vitality. Firstly, in terms of accelerating the factor cycle [16], the digital economy alleviates inter-regional information asymmetry by boosting the cross-regional communication of talent elements and capital factors [15], forming a close and mutually reinforcing knowledge network, thereby narrowing the inter-regional innovation gap and promoting the collaborative development of innovation capabilities between regions [33]. In optimizing resource allocation, the space-time barrier to knowledge and information transfer can be broken by the digital economy, which expands the spatial scope of innovation resource allocation and increases the output of innovation transactions, thereby improving the overall innovation capability.

Secondly, the development and enhancement of the digital economic system is a protracted and intricate endeavor [34]. During the first phase, the input of various production factors is relatively large, the construction of optical cable, research and development of basic technology, and training of professional personnel all require a large amount of costs. Thus, the digital economy may not be effective in promoting regional innovation capacity. However, as an important medium for information exchange, the flow of digital technology is not restricted by space and time, which can break through the barriers of space and promote the exchange of information among economic entities in innovative activities [19]. In regions where the digital economic development system is more complete, the coverage of digital dividends is more extensive. The connection of information networks among innovation subjects in a region is greatly promoted through the digital economy [35], thus improving regional innovation.

Additionally, based on the above research, as a new economic form with strong regional differences, the digital economy presents obvious advantages in the construction of the regional innovation system. Thus, we propose the hypothesis 1:

Hypothesis 1. 

Digital economy can stimulate the improvement of regional innovation directly, which exhibits heterogeneity across different regions.

2.2.2. Relations of Digital Economy, Regional Innovation, and Environmental Performance

The dividends radiate to all areas of society from the digital economy. Specifically, it improves environmental performance through establishing a modern green development system with the help of digital environmental technologies. First, digital industrialization is the main engine to leverage the potential of regional innovation [36], which helps conventionally emissions-heavy industries to greatly lower the emission of harmful substances in the manufacturing process through intelligent production lines and scientific control technologies, thus improving the usage rate of various resources and ensuring regional innovation vitality. Among them, agricultural digitization enables traditional agriculture to integrate and analyze agricultural information with the help of information technology to complete agricultural large-scale production and intelligent management [5]. Secondly, enterprises can collect information and integrate resources more efficiently, implement scientific production decisions [37], and greatly hasten the dissemination of environmental knowledge and the establishment of a green production pattern, which is an important contribution to the orderly development of regional innovation activities.

With the comprehensive improvement of the digital economy, many fields have carried out all-around green upgrading, and its empowering effect on innovative activities has been fully released [38]. Subsequently, improvements in environmental quality can promote higher levels of regional innovation. The influence of environmental quality enhancement on the regional innovation level of the digital economy may be encapsulated in two key aspects. First, the ecological environment is rich in natural resources, which can meet the material and power supply needs, provide a material basis and important guarantee for economic activities, and thus promote regional innovation and sustainable development. Secondly, various research elements have certain tendencies and usually choose to flow to regions with higher marginal returns in order to maximize their own interests [39]. An optimal ecological environment fosters a habitable atmosphere and attracts an influx of innovative resources, including high-tech personnel and foreign capital. Accelerating the flow of factors and enhancing the awareness rate of resources [16,40] thereby achieves a higher level of innovation.

It is worth noting that since the development of the digital economy is a long-term and complex process involving continuous input, technological iteration, and institutional adaptation, its impact on environmental performance is likely to exhibit non-linear characteristics rather than a simple linear relationship [41]. Specifically, in the initial stage of digital economy development, regions may face increased resource demand due to large-scale infrastructure construction, which could temporarily exert pressure on the ecological environment and weaken the immediate environmental benefits. However, as the digital economy matures, the use of renewable energy in digital operations increases, and the environmental benefits tend to accelerate [16]. This non-linear dynamic further implies that the role of environmental performance in facilitating regional innovation may also vary depending on the stage of digital economy development. Thus, we propose hypothesis 2, and the research framework diagram of this paper includes the theoretical framework and the empirical framework, as shown in Figure 1:

Hypothesis 2. 

Under the influence of environmental performance, the digital economy can have an indirect impact on regional innovation levels, and this impact varies at different stages of development.

3. Model, Variable, and Data

3.1. Model Construction

Through the above theoretical analysis, it is found that the economic system and ecosystem are closely related and promote each other. Hence, this part scrutinizes the multiple and complex influence between the digital economy and regional innovation, and proposes a model framework from direct to indirect, including a baseline model, mediating effect model, and threshold effect model.

3.1.1. Baseline Model

Both time and individual fixed effects are considered in the bidirectional fixed effects model. By introducing a separate intercept term for each individual (i.e., the individual fixed effect) and a common intercept term for each time point (i.e., the time fixed effect), the individual characteristics and time characteristics in the panel data are controlled, thus obtaining an estimate closer to the true linkage between the explanatory and dependent variables. Referring to existing literature [16,21], this paper takes the bidirectional fixed-effect model as benchmark model to test the base effect of the digital economy on regional innovation:

$${{\displaystyle \mathrm{RI}}}_{\mathrm{it}}={{\displaystyle \alpha }}_1{{\displaystyle \mathrm{DE}}}_{\mathrm{it}}+{{\displaystyle X}}_{\mathrm{it}}^{\mathrm{T}}\beta +{{\displaystyle \lambda }}_{\mathrm{i}}+{{\displaystyle \nu }}_{\mathrm{t}}+{{\displaystyle \epsilon }}_{\mathrm{it}}$$

(1)

where i denotes the area; t denotes the year; RI_it represents the level of regional innovation; DE_it denotes the comprehensive development degree of the digital economy, and X_it is the control variables. λ_i is provincial fixed effects, and ν_t is year fixed effects. ε_it is the stochastic disturbance. If α₁ is significant and positive, it indicates that the digital economy contributes to enhancing the overall regional innovation capability.

3.1.2. Mechanism Model

Considering that environmental performance is an essential medium for the stable improvement of regional innovation through the digital economy [42], it is necessary to examine its transmission mechanism by introducing environmental performance, such as by testing its indirect effect and non-linear effect.

(1)

Mediating effect model

The digital economy is conducive to the improvement of environmental performance; meanwhile, high-quality environmental performance provides a key guarantee for the sustainable improvement of regional innovation. Hence, a mediating effect model with the environmental performance as the mediating variable is constructed [43]:

$${{\displaystyle \mathrm{EP}}}_{\mathrm{it}}={{\displaystyle \alpha }}_2{{\displaystyle \mathrm{DE}}}_{\mathrm{it}}+{{\displaystyle X}}_{\mathrm{it}}^{\mathrm{T}}\beta +{{\displaystyle \lambda }}_{\mathrm{i}}+{{\displaystyle {{\displaystyle \nu }}_{\mathrm{t}}+\epsilon }}_{\mathrm{it}}$$

(2)

$${{\displaystyle \mathrm{RI}}}_{\mathrm{it}}={{\displaystyle \alpha }}_3{{\displaystyle \mathrm{DE}}}_{\mathrm{it}}+{{\displaystyle \alpha }}_4{{\displaystyle \mathrm{EP}}}_{\mathrm{it}}+{{\displaystyle X}}_{\mathrm{it}}^{\mathrm{T}}\beta +{{\displaystyle \lambda }}_{\mathrm{i}}+{{\displaystyle \nu }}_{\mathrm{t}}+{{\displaystyle \epsilon }}_{\mathrm{it}}$$

(3)

where EP_it indicates environmental performance (EP). The mediating effect model includes Equations (1)–(3). Specifically, if α₁ is not significant, the test ends. If α₁, α_2, and α₄ are significant, and α₂, and α₄ are positive, it suggests that the DE can substantially improve RI through developing EP.

(2)

Threshold effect model

In the framework of the multi-transmission effect, apart from the mediating effect, the non-linear relationship of environmental performance is also the focus of consideration. The threshold model can effectively capture the differences in the effects of independent variables on dependent variables in different stages, so this paper constructs a threshold effect model to test the non-linear effect, in which the digital economy is the threshold variable:

$${{\displaystyle \mathrm{RI}}}_{\mathrm{it}}={{\displaystyle \theta }}_1{{\displaystyle \mathrm{EP}}}_{\mathrm{it}}\times \mathrm{I}\left({{\displaystyle \mathrm{DE}}}_{\mathrm{it}}\le {{\displaystyle \xi }}_1\right)+{{\displaystyle \theta }}_2{{\displaystyle \mathrm{EP}}}_{\mathrm{it}}\times \mathrm{I}\left({{\displaystyle {{\displaystyle \xi }}_1<\mathrm{DE}}}_{\mathrm{it}}\right)+{{\displaystyle X}}_{\mathrm{it}}^{\mathrm{T}}\beta +{{\displaystyle \lambda }}_{\mathrm{i}}+{{\displaystyle \nu }}_{\mathrm{t}}+{{\displaystyle \epsilon }}_{\mathrm{it}}$$

(4)

in this equation, DE_it is the threshold variable and EP_it is the variable affected by the threshold variable. ξ_q (q = 1, 2) are the thresholds. I(.) represents an index function; if DE_it meets the specific requirement of ξ_q, its value is 1; otherwise, it assumes a value of 0.

3.2. Variables Design

3.2.1. Regional Innovation (RI)

Regional innovation (RI) is the embodiment of a region’s knowledge transformation ability [44]. In the existing literature, the measurement of RI often relies on the number of patents accepted [45]. However, it should be noted that this approach has two disadvantages: First, while some patents may be accepted, they may not ultimately be formally approved and authorized, and using the number of accepted patents as the measurement index could lead to an overestimation of the regional innovation output level [39]. Second, a spatial spillover effect exists, whereby the emergence and dissemination of innovation in one region can influence and benefit other regions [42,46]. However, this important factor is ignored if patent acceptance volume is used to characterize regional innovation directly.

Hence, this paper chooses to employ the number of patent grants (Gpat) in each region as a surrogate for regional innovation to address the first deficiency; it is an effective index to measure the final innovative result [16]. Additionally, we calculate the disparity between the number of patents awarded in area i and those in other regions to assess the spatial spillover impact of regional innovation, defined by the following function:

$${{\displaystyle \mathrm{RI}}}_{\mathrm{it}}={\displaystyle \sum _{i\ne j=1}^N\left({{\displaystyle \mathrm{Gpat}}}_{\mathrm{it}}-{{\displaystyle \mathrm{Gpat}}}_{\mathrm{jt}}\right)}$$

(5)

where Gpat is the number of patents granted (where j = 1, 2, …, N), if i = j, RI_it is equal to 0. The specific economic implications of the formula are as follows: when the quantity of patent grants in region i exceeds that in region j, this indicates that innovation activities in region i would have a positive effect on region j, driving innovation upgrading in region j. Conversely, in instances where the quantity of patents awarded in area i is inferior to that in region j, due to the siphon effect of region j, the innovation promotion of region i may be hindered to a certain extent, resulting in a lower level of innovation.

3.2.2. Digital Economy (DE)

As a complex concept, the measurement of the digital economy (DE) is the result of a comprehensive analysis of multiple factors. Existing literature widely acknowledges that digital infrastructure construction [47], industrial digitization, and digitized industrialization [48] can serve as indicators to measure the digital economy [49]. However, relying solely on these indicators cannot fully reveal the status of the digital soft environment, which is an important reflection of the external support from governments and other entities [16], which must be considered. Therefore, this paper incorporates the digital soft environment, which includes the digital governance environment, digital technological environment, and digital financial development environment. Taking the effectiveness of governance, the vitality of research activity, and the maturity of financial development into account, the existing system has been enriched and improved while focusing on assessing the stability and sustainability of the DE.

Concretely, the DE can be determined using digital infrastructure construction (Did-infra), digital industrialization (Dig-indus), digital integration development (Dig-integ), and the digital soft environment (Dig-envir). Furthermore, it is worth emphasizing that the digital financial development dimension is described by three indicators: breadth of coverage, depth of use, and degree of digitization, which derives from the digital financial development Environment Index compiled by Peking University [50]. In particular, the digital governance environment data comes from the ‘Provincial Government and Key City Government Comprehensive Service Capability Investigation and Evaluation Report’. The specific DE index system is shown in

Table 1. Digital economy’s evaluation system.

Dimension Layer	Indicator Layer	Concrete Indicator Layer
Digital infrastructure (+)	Network construction	Number of Internet users
		Number of domain names
		Number of broadband access ports
		Number of phone users
	Infrastructure construction	Length of long-distance optical cable line
Digital industrialization (+)	Digital scale	Number of employees in scientific research services
		Number of enterprises with e-commerce transactions
	Digital income	Technical service revenue
		Software product revenue
		Software business income
Digital integration development (+)	Agriculture’s digitization	Per capita expenditure on education, culture, and entertainment
		Per capita expenditure on transportation and communications
		Total power of agricultural machinery
		Rural broadband access users
	Industry’s digitization	Employees in the electronic industry
		Total assets of the electronic manufacturing industry
		R&D projects of industrial enterprises
		R&D funds of industrial enterprises
	Service industry’s digitization	E-commerce sales
		E-commerce purchases
Digital soft environment (+)	Governance environment	Overall Index of e-government services
		Number of government and enterprise broadband access
	Financial environment	Digitization’s level
		Usage’s depth
	Technological environment	Coverage’s breadth
		Enterprise technology assets
		The number of digital economy knowledge assets

as follows:

Based on a reference to existing research [51], we adopt the entropy weight TOPSIS method. First, the objective weights of each index are calculated using the entropy method. Then, the TOPSIS method is used to rank DE, which avoids the subjectivity of evaluation to a certain extent.

• Step 1: Standardize the indicators.

Considering the possible dimensional differences between the indexes, we use the max-minimum standardization to reduce the error of the results before concrete calculation. All the indicators in this paper are positive.

$${{\displaystyle \mathrm{Z}}}_{\mathrm{ip}}={\displaystyle \frac{{{\displaystyle \mathrm{X}}}_{\mathrm{ip}}-\min \left\{{{\displaystyle \mathrm{X}}}_{\mathrm{ip}}\right\}}{\max \left\{{{\displaystyle \mathrm{X}}}_{\mathrm{ip}}\right\}-\min \left\{{{\displaystyle \mathrm{X}}}_{\mathrm{ip}}\right\}}}$$

(6)

where X_ip is p-th indicator’ value (p = 1,2,…,P), $\max \left\{{{\displaystyle \mathrm{X}}}_{\mathrm{ip}}\right\}$ and $\min \left\{{{\displaystyle \mathrm{X}}}_{\mathrm{ip}}\right\}$ are the maximum and minimum of p indicator across all regions, and Z_ip is the value after the standardization. The value of Z_ip is 0~1.

• Step 2: Calculate the entropy weight of the index.

Entropy is an important metric to evaluate the degree of disorder in a system. In 1948, C.E. Hannon combined the concept of entropy with information theory to provide an essential basis for the quantification of information. Specifically, the smaller the information entropy of an index, the greater the degree of variation in the system, and the more information value it can embody. On the contrary, when the information entropy is larger, the index contains less information value. The formula for calculation is as follows:

$$\begin{array}{l}{{\displaystyle \mathrm{E}}}_{\mathrm{p}}=-{\displaystyle \frac{1}{\ln \mathrm{N}}}{\displaystyle {\sum }_{i=1}^N\left({{\displaystyle \mathrm{H}}}_{\mathrm{ip}}\ln {{\displaystyle \mathrm{H}}}_{\mathrm{ip}}\right)}\\ {{\displaystyle \mathrm{H}}}_{\mathrm{ip}}={\displaystyle \frac{{{\displaystyle \mathrm{Z}}}_{\mathrm{ip}}}{{\displaystyle {\sum }_{i=1}^N{{\displaystyle \mathrm{Z}}}_{\mathrm{ip}}}}}\end{array}$$

(7)

Furthermore, the more information an indicator reflects, the more important it is, and the greater its weight in indicator evaluation. Conversely, the greater the information entropy, the smaller the index weight. The index weights are calculated as follows:

$${{\displaystyle \omega }}_p={\displaystyle \frac{1-{{\displaystyle \mathrm{E}}}_{\mathrm{p}}}{{\displaystyle \sum _{p=1}^P(1-{{\displaystyle \mathrm{E}}}_{\mathrm{p}})}}};{\displaystyle \sum _{p=1}^P{{\displaystyle \omega }}_p}=1$$

(8)

where ${{\displaystyle \omega }}_p$ denotes the p-th weighting coefficient.

• Step 3: Conduct a comprehensive evaluation via the TOPSIS model.

Firstly, the weighted decision matrix is obtained by weighting the entropy weight matrix (v_ip):

$${{\displaystyle v}}_{ip}={{\displaystyle \omega }}_p\times {{\displaystyle Z}}_{ip}$$

(9)

Second, the positive and negative ideal solutions are calculated according to the weighted decision matrix:

$${{\displaystyle v}}_p^{+}=\underset{p}{\max }\left\{{{\displaystyle v}}_{ip},p=1,2,\dots ,P\right\};{{\displaystyle v}}_p^{-}=\underset{p}{\min }\left\{{{\displaystyle v}}_{ip},p=1,2,\dots ,P\right\}$$

(10)

Third, the Euclidean distance from the region i to the positive and negative ideal solutions is calculated. The closer the distance is to the positive ideal solution, the higher the level of digital economy in the region; the closer the distance is to the negative ideal solution, the lower the level of digital economy development. The formula is as follows:

$${{\displaystyle \mathrm{S}}}_{\mathrm{i}}^{+}=\sqrt{{\displaystyle \sum _{p=1}^P{{\displaystyle \omega }}_p{{\displaystyle ({{\displaystyle v}}_{ip}-{{\displaystyle v}}_p^{+})}}^2}};{{\displaystyle \mathrm{S}}}_{\mathrm{i}}^{-}=\sqrt{{\displaystyle \sum _{p=1}^P{{\displaystyle \omega }}_p{{\displaystyle ({{\displaystyle v}}_{ip}-{{\displaystyle v}}_p^{-})}}^2}}$$

(11)

Fourth, the relative proximity between the index of region i and the positive and negative ideal solutions is calculated:

$${{\displaystyle \mathrm{C}}}_{\mathrm{i}}={\displaystyle \frac{{{\displaystyle \mathrm{S}}}_{\mathrm{i}}^{-}}{{{\displaystyle \mathrm{S}}}_{\mathrm{i}}^{+}+{{\displaystyle \mathrm{S}}}_{\mathrm{i}}^{-}}}$$

(12)

The higher the value of ${{\displaystyle \mathrm{C}}}_{\mathrm{i}}$, the better the condition of the DE. The lower the value, the worse the condition.

3.2.3. Environmental Performance (EP)

Environmental performance (EP) is a crucial reflection of regional sustainable development, which is not only a key goal for the growth of the digital economy, but an essential prerequisite for the stable improvement of regional innovation. The existing research has not formed a unified standard about the measurement of environmental performance. Some scholars utilize single indicators to directly characterize environmental performance [52], which reflects only one aspect of environmental performance; other scholars use composite indicators, like the ESG score [53]. However, these evaluations predominantly focus on the corporate level, and these do not accurately and effectively describe the regional overall environmental performance. In contrast, as a comprehensive indicator, Air Quality Index (AQI), can reflect the environmental performance of a region to a certain extent [54,55], which aligns with our research focus on regional environmental performance as a holistic outcome—consistent with common practices in macro-level studies exploring broad economic-environmental relationships. Therefore, we use the reciprocal of the AQI to characterize environmental performance, which gives a key consideration to the varied impacts of pollutants on human health [56], ecological environment, and social activities. Furthermore, PM_2.5 serves as a critical metric for assessing air quality, and environmental protection departments in various countries monitor and control it to protect public health and the ecological environment [57]. Thus, to strike a balance between aggregate measurement and targeted assessment, PM_2.5, as a key specific pollutant, has been integrated into the study framework of the DE and RI to supplement the multidimensional validation analysis of EP.

3.2.4. Control Variables

The degree of regional innovation development reflects a region’s overall capabilities, which are influenced by various elements such as foreign investment, industrial structure, and educational level. First of all, foreign investment reflects the degree of openness of the region, which can provide strong economic support for regional development and generate positive technology spillover effects [58], thus improving the overall competitiveness of the region [59]. We select the ratio of total investment to GDP to assess foreign investment (Fore). Secondly, a reasonable and perfect industrial structure contains more emerging industries [16], which can promote regional innovation through optimizing resource allocation. Industrial structure (Str) can be calculated as the ratio of the added value of the secondary industry to the added value of the tertiary industry [16]. Third, convenient transportation facilities are a basic condition for regional innovation, which determines the technology’s spread and the flow of human resources between regions [60]. The degree of transport infrastructure is assessed by the per capita road area of each region (Tra). Fourth, good educational conditions (Edu) can ensure the talent reserve of the region, provide necessary knowledge and labor support for regional innovation and development [61], and be characterized by the ratio of education funding to regional GDP. Finally, the government can directly encourage innovation activities through policy formulation [62], financial support, and other measures, so the proportion of regional government consumption expenditure and regional GDP is used as a measure of government intervention (Gov). It should be underlined that the enhancement of regional innovation is influenced by additional elements. However, only the aforementioned five characteristics have been examined because of the data’s restricted availability and representativeness.

3.3. Data Presentation

This research relies on data at the provincial level in China (2014–2021). The China Statistical Yearbook, China Environmental Statistical Yearbook, and EPS Global analytic platform are the main sources of statistics for 30 Chinese provinces. Macao, Hong Kong, Xi Zang, and Taiwan are not included. The China Environmental Monitoring Station’s nationwide urban air quality real-time release platform provides the AQI data. In addition, part of the missing data is interpolated by a three-step moving average.

Table 2. Descriptive statistics.

Variables	Definition	Obs	Mean	Std	Min	Max
RI	Regional innovation	240	0.000	0.025	−0.035	0.139
DE	Digital economy	240	0.137	0.139	0.010	0.847
EP	Environmental performance	240	0.015	0.005	0.007	0.033
Fore	Foreign investment	240	0.089	0.403	0.006	5.118
Tra	Transportation infrastructure	240	1.207	0.154	0.614	1.428
Str	Industrial structure	240	1.371	0.723	0.666	5.297
Gov	Government intervention	240	0.040	0.014	0.021	0.076
Edu	Educational level	240	0.134	0.045	0.058	0.296

displays the major variables’ summary statistics.

Furthermore, all the data used in this study were sourced from publicly available sources, and we strictly adhered to the principle of legal acquisition and use of data. During the data storage process, appropriate measures have been taken to ensure data security and confidentiality. In terms of data interpretation, we always maintain objectivity and transparency, and avoid any biased analysis. In addition, all relevant data are presented in the main text. Any other questions or information regarding the original data can be submitted to the corresponding author for acquisition.

In Figure 2, the results of the measurements taken at the provincial level in China on the digital economy are shown. The degree of digital economy varies significantly across various locations, which indicates that the culture, natural resources, and economic conditions of a region may be closely related to the construction of a digital economy system. The top three provinces with the highest digital economy level are Guangzhou, Beijing, and Jiangsu during the mean from 2014 to 2021. However, the top average annual growth rate (AAGR) is Shaanxi, Fujian, and Heilongjiang, showing that regions with lower levels of economy have higher growth rates and greater room for growth. The above cross-over phenomenon reveals the diversity of advantages and potentials of different regions in the process of unlocking the dividends of the digital economy.

Figure 3 shows the proportional contribution of the four dimensionalities of the DE. Sub-dimensional heterogeneity between regions is revealed. From the top six cities with the greatest degree of digital economy development, it is apparent that digital industrialization and digital integration account for the greatest percentage, but how each dimension plays a part in the process of stimulating regional innovation is unknown. In light of this, it is essential to take the influence of sub-dimensional and regional elements into account while conducting the empirical study.

Furthermore, Figure 4 shows the degree of innovation in 2015 and 2020, respectively, which provides a preliminary verification of regional differences in innovation development. Specifically, in 2015 (Figure 4a), the highest level of regional innovation was in Guangzhou, Jiangsu, and Zhejiang, while most other regions remained at a low level of innovation. In 2020 (Figure 4b), these three cities will continue to lead the way in innovation development. During this period, although the innovation capacity of a few regions has been greatly improved, the overall innovation development gap between regions is widening. Too large a development gap will hinder regional coordinated development, so it is necessary to conduct heterogeneity analysis, which can provide some reference for formulating development strategies according to local conditions and narrowing the innovation development gap.

4. Empirical Results

4.1. Baseline Effects

Following the verification of Hausman and LM, the double fixed effect panel model is used for subsequent empirical analysis.

Table 3. Baseline regression result.

Variables	RI	RI	RI	RI	RI	RI
	(1)	(2)	(3)	(4)	(5)	(6)
DE	0.255 ***	0.254 ***	0.246 ***	0.249 ***	0.257 ***	0.267 ***
	(16.66)	(16.46)	(15.29)	(15.90)	(16.79)	(16.92)
Fore		−0.105	−0.099	0.087	0.124	0.145
		(−0.670)	(−0.640)	(0.540)	(0.790)	(0.930)
Tra			−0.025	−0.029 *	−0.016	−0.016
			(−1.610)	(−1.910)	(−1.080)	(−1.050)
Str				−0.015 ***	−0.016 ***	−0.015 ***
				(−3.510)	(−4.010)	(−3.770)
Edu					0.703 ***	0.423 *
					(3.860)	(1.960)
Gov						0.097 **
						(2.350)
Fixed effect	Y	Y	Y	Y	Y	Y
R2	0.58	0.58	0.59	0.61	0.64	0.65

Note: * p <0.1, ** p <0.05, *** p < 0.01; and the t statistics are shown in brackets. The sample size was 240. The same is below.

lists the results of the direct impact that the digital economy (DE) made on regional innovation (RI). Generally, DE has a stable promoting impact on regional innovation, and the results are tested by a progressive regression strategy that incorporates fixed effects for both the province and the year. Column (1) shows the single net impact effect from DE on RI is 0.255, and it passed the significance test at 0.01. Furthermore, upon the consecutive inclusion of control variables, the computed coefficients of DE to RI remain consistently significant and positive, indicating that the digital economy is an essential impetus for regional innovation [33]. Meanwhile, the DE has the largest influence coefficient on regional innovation and the best simulation fitting effect when all control variables are added. In light of this, the preceding portion of Hypothesis 1 is verified.

4.2. Mechanism Analysis

4.2.1. Mediating Effect

The analysis of the transmission mechanism of environmental performance (EP) is shown in

Table 4. Results of mediating effects.

Variables		Mediation
	RI	RI	EP	RI
	(1)	(2)	(3)	(4)
DE	0.015	0.267 ***	0.008 ***	0.260 ***
	(0.300)	(16.92)	(3.810)	(15.95)
DE×EP	0.947 ***
	(5.260)
EP				0.994 *
				(1.810)
Fore	0.072	0.145	0.038 *	0.107
	(0.500)	(0.930)	(1.880)	(0.690)
Tra	−0.025 *	−0.015	0.003 *	−0.019
	(−1.810)	(−1.05)	(1.72)	(−1.270)
Str	−0.014 ***	−0.015 ***	0.002 ***	−0.017 ***
	(−3.800)	(−3.770)	(3.380)	(−4.110)
Edu	0.431 **	0.422 *	−0.055 **	0.478 **
	(2.130)	(1.960)	(−1.980)	(2.200)
Gov	0.062	0.097 **	0.003	0.094 **
	(1.580)	(2.350)	(0.620)	(2.280)
Fixed effect	Y	Y	Y	Y
R2	0.69	0.92	0.97	0.92

. It can be concluded that the indirect impact of DE on RI existed under the influence of EP. Specifically, Column (1) presents regression results with DE×EP, indicating that, after considering the impact of environmental performance, the DE’s influence on RI is diminished, highlighting the importance of the indirect impact of environmental performance. In addition, columns (2) and (3) prove that DE has a significantly positive correlation with RI and environmental performance, indicating that the DE not only effectively promotes regional innovation development, but also greatly contributes to improving environmental performance. Meanwhile, Column (4) presents that there is a partial mediating effect (0.008 × 0.994) between the DE and RI. The indirect effect of environmental performance is verified.

However, it can be seen from the results that the impact of the digital economy on innovation is mainly direct rather than indirect. The reasons for this phenomenon may be based on the following two points: First, China’s industrial and innovation ecosystems are still adapting to the green transformation. Compared with the digital economy directly reducing innovation costs and accelerating resource allocation efficiency, the driving range of environmental effects is limited. Therefore, although environmental performance does play a mediating role at present, the marginal contribution of environmental improvement to innovation is relatively limited. Secondly, the driving effect of environmental effects may have stage differences [21], so a threshold effect analysis should be further conducted.

4.2.2. Threshold Effect

In light of the theoretical analysis, there is an obviously non-linear relationship between DE and RI under the consideration of environmental performance, and the threshold panel model can be used to employ a more specific analysis. Before estimating the nonlinear relationship between environmental performance and regional innovation through Equation (4), it is necessary to determine whether the threshold value exists and its quantity. According to

Table 5. Results of the threshold value’s test.

Threshold Variable	Type	F Value	p Value	Critical Value			Threshold Value	95% Confidence Interval
				1%	5%	10%
DE	Single	131.4	0.000	43.95	31.05	26.90	0.367	[0.348, 0.394]
	Double	41.57	0.123	372.8	255.4	124.6	0.450	[0.430, 0.501]
							0.501	[0.440, 0.557]

, there is a single threshold for the impact of environmental performance on regional innovation at different levels of the digital economy, and it is significant at 1%. Testing the threshold shows that the single threshold is 0.450. In practical terms, DE = 0.450 marks a critical stage where regions transition from “basic digital development” to “advanced digital maturity”—characterized by robust digital integration into traditional industries, well-established digital governance, and sufficient digital financial support, which are consistent with the core drivers of regional innovation identified in our findings.

The results of the

Table 6. Estimation results for the threshold values of the digital economy.

RI	Coefficient	t	p Value	95% Confidence Interval
EP (DE ≤ 0.450)	1.898 ***	(3.11)	0.004	[0.650, 3.147]
EP (DE > 0.450)	4.658 ***	(5.16)	0.000	[2.811, 6.505]
Controls	Y
Fixed effect	Y
R2	0.49

indicate that when the DE index falls below 0.450, the impact coefficient of Environmental Performance (EP) on Regional Innovation (RI) is 1.898. Conversely, when the DE index surpasses or equals 0.450, this coefficient increases significantly to 4.658, passing the significance test. It indicates that as the digital economic system continually evolves and matures, it facilitates the widespread dissemination of digital dividends, thereby amplifying the stimulatory effect of environmental performance on regional innovation.

Notably, contextualizing this threshold within our sample reveals practical implications for China’s regional development. The DE index’s average value across the sample period (0.134) in Table 2 is far below the threshold of 0.450. Figure 2 illustrates that only Guangdong Province has surpassed the 0.450 threshold, while Beijing and Jiangsu are approaching it (around 0.40). The majority of other provinces remain significantly below this level. These observations highlight that the digital economy, as a powerful engine, has yet to fully unleash its potential in most regions. Accelerating digital economy development—particularly in improving digital integration, optimizing the digital soft environment, and strengthening infrastructure—will be critical to crossing this threshold. It can amplify the environmental performance dividends, thereby driving regional innovation and sustainable development more vigorously. Furthermore, these results underscore the pivotal role of environmental performance in fostering sustainable growth of regional innovation. Consequently, these results provide empirical support for Hypothesis 2.

4.3. Examination of Robustness

This section verifies the reliability of the findings by three methods: the decomposition of sub-dimensions for digital economy (DE), the design of the quantile for regional innovation (RI), and the replacement of core variables.

4.3.1. The Impact of Different Dimensions of DE and Different Levels of RI

From the decomposition of the sub-dimensions of DE: digital infrastructure (Dig-infra), digital industrialization (Dig-indus), digital integration development (Dig-integ), and digital soft environment (Dig-envir); the first four columns of

Table 7. Robustness checks by different dimensions of DE and different levels of RI.

Variables	RI				Tau = 0.25	Tau = 0.50	Tau = 0.75
	(1)	(2)	(3)	(4)	(5)	(6)	(7)
Dig-infra	0.091 ***
	(3.53)
Dig-indus		0.188 ***
		(12.40)
Dig-integ			0.266 ***
			(18.12)
Dig-envir				0.223 ***
				(16.02)
DE					0.122 ***	0.151 ***	0.189 ***
					(35.10)	(29.34)	(27.96)
Control variables	Y	Y	Y	Y	Y	Y	Y
Fixed effect	Y	Y	Y	Y	Y	Y	Y
R2	0.18	0.51	0.67	0.62	0.64	0.63	0.68

show the results. Overall, the four dimensions of the digital economy all contribute to enhancing regional innovation level significantly, and the improvement of Dig-integ exerts the strongest driving effect on RI, followed by the digital soft environment and digital industrialization, while the construction of digital facilities shows the least promoting effect. The building of digital infrastructure is a fundamental need for innovative activities, but it does not directly engage in such activities. Therefore, the other three variables play a more direct and substantial role in encouraging regional innovation. Furthermore, different aspects of digital economy growth have varying effects on regional innovation, requiring government policy formulation to be nuanced and tailored accordingly.

Quantile regression analysis can verify the asymmetric impact of RI. Columns (5) to (7) of Table 7 report the three quantiles of conditional results, and the asymmetry of the elasticity of digital economy to regional innovation is demonstrated. It also reveals that, with the regional innovation level gradually improving, the driving force on regional innovation capabilities is also gradually strengthening from the digital economy. To sum up, the tests reveal that the digital economy continues to play a significant role in the development of regional innovation, thus again supporting Hypothesis 1. Meanwhile, it is consistent with the results in Table 3, which demonstrates the robustness of the baseline results.

4.3.2. Changing the Measuring Method

To strengthen the credibility of the findings presented in this paper, the Digital Economy Index (DE-PCA) was devised through principal component analysis and utilized in a regression analysis, as per the methodology employed by Zhao et al. [63]. The outcomes of this analysis are summarized in Columns (1) and (2) of

Table 8. Robustness checks by replacing the variables and endogenous checks.

	Replacing DE		Replacing RI		Endogenous Checks
Variables	RI		PAT		DE	RI
	(1)	(2)	(3)	(4)	(5)	(6)
DE-PCA	0.258 ***	0.264 ***
	(19.23)	(19.34)
DE			0.450 ***	0.573 ***
			(4.26)	(5.310)
L.DE					1.068 ***
					(25.89)
Pre-DE						0.295 ***
						(5.580)
Control variables	N	Y	N	Y	Y	Y
Fixed effect	Y	Y	Y	Y	Y	Y
N	240	240	240	240	210	210
R2	0.65	0.70	0.88	0.89	0.93	0.93
Kleibergen-Paap LM					4.720 **
Kleibergen-Paap Wald F					670.2

. Specifically, Column (1) underscores the positive impact of the digital economy on enhancing regional innovation capabilities. When five control variables are introduced, including foreign investment (Fore), transportation infrastructure (Tra), industrial structure (Str), education investment (Edu), and government intervention (Gov), the significance of the digital economy to unlock the potential of innovation remains, as evident in Column (2). Moreover, the independent variable was substituted with the number of authorized invention patents (PAT). Columns (3) and (4) of Table 8 reveal that regardless of whether control variables are included, the digital economy continues to bolster regional innovation, thereby reinforcing the conclusion that the initial findings are indeed robust.

4.4. Endogenous Checks

When conducting regression analysis to explore the interplay between the digital economy and regional innovation, an endogeneity concern may arise due to the potential bidirectional influence. Advancements in the digital economy can stimulate innovation across various sectors, while the extensive application of innovative achievements and ongoing technological advancements contribute to digital improvement. Recognizing this bidirectional endogenous relationship [16], this paper employs the two-stage least squares regression method to address the issue. Columns (5) and (6) show that the Kleibergen-Paap LM statistic is 4.720 and significant in Table 8, indicating that there is an endogeneity problem in regression analysis when other factors are controlled. The Kleibergen-Paap Wald F statistic is used to test the strength of the instrumental variable, with a value of 670.2, indicating that the instrumental variable is a strong instrumental variable, that is, L.D is not correlated with the error term, which increases the confidence of the results of the IV estimation. Furthermore, the validity of Hypothesis 1 is further supported by the fact that the estimated coefficient of DE remains significant in the face of the endogeneity problem.

5. Additional Analysis

5.1. Heterogeneity Test of Region

China shows complexity and diversity in geography, natural resources, and conditions of economic development [64], indicating that the potential for digital economy growth under different conditions may exhibit diverse characteristics. In economic geography research, although the division of “economically coherent regions” can better reflect the functional economic connections between regions, such analyses essentially classify regions based on economic development stages [65]. However, the threshold effect analysis in this study has intrinsically captured the heterogeneity at the economic level, and the policies for the construction of Digital China are mostly implemented by provincial administrative units [66]. For this reason, this paper further conducts heterogeneity analysis based on administrative coherent regions.

Meanwhile, Figure 2 and Figure 4 confirm the presence of heterogeneity. Thus, it is necessary to conduct heterogeneity tests. The three geographical plates (East, Central, and West) are taken into consideration for revealing the different regional characteristics. This division is derived from the official definition of the four major geographical regions by the National Bureau of Statistics of China [67]. Due to the limited sample size in the northeastern region and referring to existing literature [68], this paper merges the northeastern region into the eastern and central regions. The three major regions reflect fundamental regional and developmental differences [68,69].

Table 9. Heterogeneity regression results based on different areas.

Variables	Three Plates
	East	Central	West
	(1)	(2)	(3)
DE	0.254 ***	0.131 *	0.156 ***
	(8.09)	(1.84)	(12.48)
Control variables	Y	Y	Y
Fixed effect	Y	Y	Y
N	104	48	88
R2	0.66	0.70	0.97

reveals the development difference between the three major sectors, in which the enhancement impacts of DE on RI are shown as follows: East > West > Central. It shows that the eastern region, with its advantages in digital system construction and economic development, can make more effective use of the digital economy to tap innovation potential. At the same time, the central and western regions are relatively backward. There is still much space for improvement in both the central and western regions of the DE, indicating that the formulation of development strategies should take various local factors into account and fully stimulate the pillar role of the DE on RI. The second half of Hypothesis 1 is tested.

5.2. Heterogeneity Test from Environmental Pollution

The AQI, as a comprehensive metric for assessing environmental performance, primarily encompasses six pollutants: PM_2.5, PM₁₀, NO, NO₂, O₃, and SO₂. However, given the significant variations in air pollution sources, climatic conditions, and environmental backgrounds, the calculation methods and standards of AQI may vary across different countries or regions, leading to certain limitations when comparing environmental performance with other countries. Therefore, this paper incorporates PM_2.5 into the analytical framework, which is widely recognized at the international level for its ability to comprehensively reflect pollution levels [68]. This kind of particulate matter can pose a direct threat to health, and its sources cover a wide range, including industrial emissions, traffic exhaust, and dust. Analyzing its role in the progress of regional innovation provides a more objective and scientific basis for international environment comparison and assessment.

Table 10. Results of heterogeneity from environmental pollution.

Variables		Mediation		Dimension Decomposition
	RI	PM2.5	RI	PM2.5
	(1)	(2)	(3)	(4)	(5)	(6)	(7)
DE	0.267 ***	−0.423 ***	0.257 ***
	(16.92)	(−3.940)	(15.85)
PM2.5			−0.024 **
			(−2.290)
Dig-infra				−0.008
				(−0.070)
Dig-indus					−0.350 ***
					(−3.980)
Dig-integ						−0.366 ***
						(−3.490)
Dig-envir							−0.356 ***
							(−3.880)
Control variables	Y	Y	Y	Y	Y	Y	Y
Fixed effect	Y	Y	Y	Y	Y	Y	Y
R2	0.92	0.92	0.92	0.39	0.63	0.65	0.65

presents the results. First, the PM_2.5 index is used as an intermediation to investigate the link between the DE and RI. Its findings, which are shown in Columns (1) through (3), demonstrate that the digital economy effectively limits the increase in PM_2.5 levels so as to enhance the capacity of regional innovation. Furthermore, regression analyses are built to further examine which dimension is the primary source of the inhibitory impact of the DE on PM_2.5. Columns (4) to (7) show that the growth of digital integration has the most inhibiting impact on PM_2.5, followed by digital industrialization and digital soft environment, and digital infrastructure has the weakest governance effect. This emphasizes how crucial it is for traditional fields and new technology to be deeply integrated to support sustainable growth. The robustness of Hypothesis 2 is further verified by analyzing the transmission mechanism of PM_2.5.

5.3. External Impact: From COVID-19

Regional innovation development issues pose challenges due to the uncertainty in the external economic development environment. The emergence of the COVID-19 outbreaks in 2019 has had far-reaching implications for global development, posing serious challenges to the global public health system and leading to significant disruptions in the global economy [21,70]. Taking 2019 as the time node, regression analysis was conducted before and after the epidemic.

Table 11. The effects of exogenous shocks.

Variables	Before 2019	After 2019
	(1)	(2)
DE	0.219 ***	0.370 ***
	(16.22)	(6.120)
Control variables	Y	Y
Fixed effect	Y	Y
N	150	90
R2	0.76	0.57

shows that, after the pandemic, the effect of the digital economy in improving regional innovation capacity has been enhanced, indicating that the development of regional innovation is more dependent on the digital economy while facing external shocks, verifying the strong resilience of the digital economy, and showing greater potential in supporting economic recovery and stimulating regional innovation vitality. At the same time, the robustness of the results in this paper was further verified.

6. Conclusions and Implications

6.1. Discussion and Conclusions

This paper proposes a multi-transition framework based on theoretical analyses of the digital economy and regional innovation while taking environmental performance into account. The mediating and threshold effect models experimentally explore direct effects and transmission mechanisms, utilizing panel data from China’s provinces from 2014 to 2021. This paper may be distinctive in the following respects when compared with previous research: first, we incorporate the soft environment dimension into the index system to more accurately and comprehensively access the digital economy, which provides evidence to distinguish the dimensional heterogeneity that stimulates regional innovation vitality in detail. Second, by innovatively considering the environmental performance characterized by AQI, it expands the path for effectively promoting regional innovation via the digital economy. Third, a complete empirical framework is constructed to quantitatively analyze the impact of regional heterogeneity and exogenous shocks in the digital economy in unlocking innovation potential.

It is further revealed that scientifically and reasonably guiding the collaborative advancement of environmental performance and digital economy can stimulate greater regional innovation benefits. In this light, several crucial findings are as follows:

(1)

Different components of the digital economy show obvious differences in the direct promotion of regional innovation. In particular, the further integration of the traditional industries and digital technology can tap the potential of regional innovation growth to the greatest extent. Furthermore, the construction of the digital environment plays a strong effect in enhancing the capacity of regional innovation, which is also an indispensable supporting factor in the innovation process.

(2)

The multi-transmission effects of environmental performance are confirmed in the model relationship based on digital economy and regional innovation. Firstly, the mediating effect indicates that environmental performance is an important intermediary in the task of enhancing regional innovation, and the digital economy can indirectly enhance regional innovation ability by improving environmental performance. Secondly, there exists a nonlinear linkage between environmental performance and regional innovation under the context of digital growth. Specifically, when the digital economy index exceeds the 0.45, the positive effect of EP on RI increases exponentially. Third, digital integration has contributed the most to the process of controlling PM_2.5 emissions, thus injecting a strong impetus to regional innovation.

(3)

From the perspective of spatial heterogeneity and exogenous influence, it is found that the digital economy’s influence on promoting regional innovation is geographically imbalanced, with East > West > Central. In contrast to the eastern regions, which have active financial markets and excellent technical support, the digital economy has less motivation to foster regional innovation in the middle and western areas. Furthermore, in terms of external influence, regional innovation has become increasingly reliant on the digital economy during the COVID-19 pandemic, demonstrating that the digital economy is resilient and a major engine for revitalizing regional innovation.

6.2. Policy Implications

The strength, depth, extent, and efficiency of optimizing environmental performance and deeply integrating the digital economy should be promoted to a higher level; it is a critical engine to stimulate regional innovation potential. According to the findings of this paper, policy recommendations are put forward to further stimulate the release of regional innovation momentum:

(1)

The primary aim is stimulating the advancement of the digital economy according to local conditions. The policy support, ecological, economic, and other conditions of the country or region should be comprehensively assessed to formulate reasonable and feasible development strategies. First, the central and western regions in which the development is lagging should focus on promoting the digital upgrading of traditional fields and actively attracting capital and policy support to create a high-quality development environment for driving innovative development. Second, for the eastern region with a more mature innovation system, the government should encourage business owners to share technology and export talents to the surrounding areas, forming a virtuous cycle of mutual promotion and ensuring the steady progress of regional innovation.

(2)

Enhance the governance of environmental performance by incorporating environmental improvement into the strategy of the digital economy. Specifically, enterprises should persist in the extension and expansion of their digital product chains and accelerate the green transformation of polluting enterprises, promoting regional innovation efficiency. For the government, at the beginning of the construction of the digital economy system, scientific support and guidance is essential to fully stimulate the environmental performance dividend and promote regional innovative development in the later stage. Meanwhile, it is important to establish a comprehensive environmental performance monitoring and evaluation system to strictly control PM_2.5 emissions at the source, thereby achieving a win-win scenario for a region or nation in terms of both innovation benefits and environmental benefits.

(3)

Enhance regional resilience to risks. Unforeseen public emergencies are inevitable in the improvement of regional innovation. Hence, it is a crucial task to continuously consolidate the construction of regional innovation systems. First, the government should encourage regional innovation collaboration and the exchange of cutting-edge digital technologies and experiences, therefore improving overall anti-risk capabilities. Secondly, measures should be taken such as improving resource utilization efficiency and establishing a digital environmental supervision system to guarantee sufficient resource supply and a high-quality environment for innovation construction, which will enable regions to better adapt to the potential uncertainties and challenges of the future.

6.3. Limitations

In studying the influence of DE on RI within the constraints of EP, we strive to comprehensively reveal the intricate relationships among the three, yet inevitably encounter several limitations and challenges that point to future research directions. First, due to the limitation of data acquisition, this paper only conducted a one-way analysis on the promotion impacts. However, the enhancement of innovation ability can drive the growth of the digital economy, so there may be a two-way causal problem. We plan to adopt the Toda–Yamamoto causal test method and combine it with Fourier approximation technique in our following research to accurately distinguish the causal direction. Second, digital economy, as an emerging economic form, presents a dynamic evolution law in the development process, but this paper is limited to data availability and technical means, and fails to reflect its dynamic change law in the index design. Thus, the idea of a sliding window can be introduced in the future to comprehensively and accurately reflect its dynamic characteristics, in order to explore the complex impacts between digital economy and regional innovation from a broader perspective and provide a scientific basis and forward-looking guidance for policy making. Third, it is equally important to distinguish the specific impact of the digital economy on individual pollutants such as carbon dioxide. This article has not yet delved deeply into the mechanism of action of specific pollutants. In future research, we plan to incorporate more detailed pollutant data (including carbon dioxide, industrial wastewater, waste, solid waste emissions, etc.) to conduct more targeted analyses of the impact of the digital economy on different types of pollutants, thereby providing more targeted policy implications for environmentally sustainable development.