Big Data Innovative Development Experiments, Sci-Technology Finance Ecology, and the Chinese Path to Sustainable Modernization—A Quasi-Natural Experiment Based on SDID and DML

Abstract

Modernization in developing countries such as China has long been unsustainable. As a result, China has set the goal of achieving sustainable modernization characterized by harmony between humanity and nature. Against this backdrop, in this study, we apply spatial difference-in-differences (SDID) and double machine learning (DML) models using panel data from 30 provincial-level regions in China from 2009 to 2021. We examine the impacts of the National Big Data Comprehensive Pilot Zone policy and sci-technology financial ecology on the Chinese Path to Sustainable Modernization. The results show that big data pilot zones significantly enhance modernization and generate positive spatial spillover effects through demonstration and diffusion. Sci-technology financial ecology improves sustainable modernization and amplifies the role played by pilot zones. Heterogeneity tests reveal stronger effects in eastern provinces and in areas implementing urban–rural integration or green finance reforms. The results of the mechanism analysis show that big data innovation promotes modernization by strengthening sci-technology financial ecology, raising government attention, fostering inclusive intelligence development, enhancing green innovation efficiency, and upgrading industrial structures.

1. Introduction

The Report of the 20th National Congress of the Communist Party of China emphasized that the Chinese path to modernization is characterized by a vast population, common prosperity, coordinated material and cultural–ethical progress, harmony between humanity and nature, and peaceful development. China is currently in the initial phase of building a modern socialist country [1]. Therefore, it is crucial to understand both the theoretical foundation and practical significance of this path. This requires clarifying the challenges it faces, seizing emerging opportunities, and identifying effective strategies.

Existing domestic studies mainly focus on interpreting the theoretical implications of modernization, tracing its historical evolution, and connecting it to policy practice. Quantitative empirical research remains scarce, especially from a regional perspective, yet regional balance is essential, as achieving modernization depends on reducing development gaps [2]. China’s social reality and policy objectives include coordinating modernization across regions. Promoting such coordination is therefore vital to achieving national modernization [3].

In September 2015, the State Council issued the Action Plan for Promoting the Development of Big Data (the “Plan”). It called for accelerating big data innovation, expanding applications, and building a new ecosystem for the data industry. In 2016, eight regions—including Guizhou Province, the Beijing–Tianjin–Hebei region, and the Pearl River Delta—were approved as National Big Data Comprehensive Pilot Zones (hereinafter referred to as “pilot zones”). These zones served different purposes. Cross-regional pilot zones have promoted inter-regional flows of production factors using data as a key input. Demonstration zones focused on resource integration and industrial clustering to achieve spillover effects. Infrastructure-oriented zones aimed to restructure local economies and transform growth models by leveraging local natural resource advantages. Pilot zones have achieved notable results in terms of raising productivity, fostering innovation, and improving energy efficiency [4,5,6]. However, existing research has not explained how these policies influence the Chinese Path to Sustainable Modernization; this study aims to fill this research gap.

Building a modern economic system and implementing a new development philosophy require the joint development of the real economy, technological innovation, and modern finance. The concept of the Science and Technology Finance Ecology, which refers to the integration of scientific and financial innovation within an ecosystem framework, plays a key role in optimizing resource allocation and supporting high-quality growth as China advances toward innovation-driven development [7]. Although prior research confirms that science and technology finance promotes high-quality economic growth [8], its role in advancing Chinese-style modernization remains underexplored.

Since the introduction of the Pilot Zone policy, many regions have aligned their measures with the Plan and undertaken supporting reforms. Guizhou Province, for instance, has developed over 200 big data standards, built 23 data centers, and established China’s first big data exchange. Shanghai has promoted data openness through regulations such as the Measures for the Management of Public Data and Integrated Online Services and the Guidelines for the Classification and Grading of Public Data Openness. The Beijing–Tianjin–Hebei Pilot Zone has built a collaborative processing center and innovation platform, applying big data to energy, biotechnology, mining, and transportation. These improvements show that big data innovation depends on both financial support and technological progress. The Pilot Zone policies and the Science and Technology Finance Ecology reinforce each other. Accordingly, this study integrates big data innovation, the financial ecology, and Chinese-style modernization into a unified analytical framework to explore their dynamic interactions and mechanisms, thereby contributing by empirically identifying the impact of Pilot Zone policies on sustainable modernization, revealing the mediating role played by the Science and Technology Finance Ecology, and highlighting regional heterogeneity to provide new evidence for promoting coordinated and sustainable modernization in developing countries.

2. Literature Review and Theoretical Framework

2.1. Big Data Innovation Pilot Zones and the Chinese Path to Sustainable Modernization

From the perspective of New Structural Economics, realizing the Chinese path to modernization entails achieving a range of interrelated goals: modernizing the national governance system and capacity; enhancing indigenous technological innovation; synchronizing income growth with economic growth and labor compensation with productivity; improving public service infrastructure and the social security system; enhancing urban–rural living environments; and advancing ecological civilization [9]. National Big Data Comprehensive Pilot Zones contribute to the achievement of these objectives by facilitating making progress across several of these domains.

First, big data allows for government-accessible digital information to coalesce into large-scale, networked systems. Through the integration and sharing of data across administrative levels, regions, and systems, big data helps us to eliminate blind spots in regulation. In areas such as administrative operations, government transparency, and process optimization, digital platforms enable the development of automated systems and intelligent service portals, substantially improving document workflows, administrative efficiency, and transparency in public governance [10]. Second, by leveraging big data technologies and massive data resources, a dynamic mechanism involving five interconnected elements—data, information, platforms, collaboration, and security—has emerged. This facilitates real-time public sentiment monitoring and dynamic risk management, contributing to the creation of an intelligent, data-driven governance system [11]. Moreover, it promotes accessible, equitable, and efficient public service delivery in the employment, healthcare, and education sectors [12]. Big data pilot zones also receive targeted policy support and increased innovation investment. As a core production factor, data drive enterprise innovation, transform innovation models, and strengthen the national innovation ecosystem [13]. Data-driven approaches enhance market transparency, alleviate information asymmetries, and mitigate resource misallocation. These pilot zones aggregate resources, improve academia–industry collaboration, and integrate data with human and intellectual capital. They also enhance intellectual property protection [14], encourage the internationalization of enterprise R&D [15], and stimulate innovation at scale.

Furthermore, big data improves regional total factor productivity through human capital investment and industrial upgrading [16]. The demand for digital talent fosters human capital accumulation and boosts resource allocation efficiency. The rising prominence of knowledge and information reduces transaction costs, improves managerial efficiency, broadens industrial specialization, and accelerates industrial digitalization. In their study on national-level big data pilot zones, Bu Han et al. [17], in their study on national-level big data pilot zones, found that big data can significantly increase labor income shares via direct effects of task creation and indirect effects such as automation and easing financial constraints. Structurally, big data provides strong positive externalities that reinvigorate China’s “Troika” of economic growth [18]. First, investment in data-intensive sectors boosts market confidence. With rising domestic substitution in high-tech, security-sensitive industries—like chip manufacturing and telecommunications—traditional industries also face urgent needs for digital transformation, offering broad prospects for the development of big data. Second, big data enhances consumer expectations and broadens consumption channels. With the growing integration of recommendation systems, mobile payment, and smart logistics, it expands market capacity, reduces entry barriers, and improves transaction efficiency and user experience. Third, by linking domestic and international markets through low-cost, high-speed, and diversified channels, big data enhances the global competitiveness of Chinese products and services. These combined effects support China’s effort to overcome the “middle-income trap” through reoriented investment, consumption upgrading, and export advantages.

According to the 14th Five-Year Plan and the 2035 Vision Outline of the People’s Republic of China, the nation has tasked pilot zones with exploring new digital production relations, building leading big data industrial hubs, and fostering a data value ecosystem. From their inception, these zones have been designed to generate spillover effects, facilitating coordinated regional development through resource integration and industrial clustering. Given their inherent policy demonstration function, we posit that big data innovation pilots will exert positive spillover effects on neighboring regions’ modernization processes.

Based on this analysis, we propose the following hypotheses:

H1. 

Big data innovation pilot programs significantly enhance the level of the Chinese Path to Sustainable Modernization.

H2. 

Big data innovation pilot programs generate significant positive spatial spillover effects on the Chinese Path to Sustainable Modernization.

2.2. Science and Technology Finance Ecology and the Chinese Path to Sustainable Modernization

As China embarks on a new journey toward building a modern socialist country, both science and finance—key drivers of economic growth—must undergo deep integration [19]. While existing research largely focuses on how the Science and Technology Finance Ecology affects high-quality development, innovation, and green transitions, these themes are essential pathways to achieving Chinese-style modernization. This section explores how this ecology supports each of modernization’s sub-tasks to evaluate its overall impact on sustainable modernization.

First, financial development lowers transaction, information, and regulatory costs while enabling intertemporal resource allocation through market mechanisms. Price signals help coordinate supply–demand mismatches and manage macroeconomic challenges in the context of China’s massive population [20]. As intermediaries, financial institutions catalyze the digital transformation of traditional industries and expand upon the scope of wealth distribution. Simultaneously, the Science and Technology Finance Ecology introduces new tools for data analysis, risk forecasting, and crisis management, enhancing the resilience and quality of economic growth [21].

Second, following Schumpeter’s theory of innovation, a well-functioning financial system incentivizes technological innovation, generating supernormal profits and accelerating capital accumulation, thus triggering waves of innovation-driven economic expansion. As the ecology matures, it fosters the greater societal recognition of innovation: respect for scientists, encouragement of experimentation, and tolerance for failure. Increased policy support and rising government R&D expenditure have enabled tech-driven enterprises to expand financing access, promote knowledge production, and accelerate technological commercialization [22]. In terms of green development, the ecology fosters green innovation through labor-absorbing intelligent manufacturing and digital empowerment [23]. It also suppresses corporate pollution and improves carbon performance through talent agglomeration and clean technology adoption.

However, the spatial effects of the Science and Technology Finance Ecology are ambivalent: both siphon and spillover effects can occur across regions [24]. On the one hand, the spread of technology, knowledge, and talent from developed areas can inspire lagging regions and promote inclusive growth. On the other hand, technological agglomeration and market segmentation create barriers to resource mobility, and resource-rich regions may attract disproportionate development, generating siphoning effects. Thus, the direction and magnitude of spatial spillovers remain theoretically and empirically uncertain.

Based on this analysis, we propose the following hypothesis:

H3. 

The Science and Technology Finance Ecology significantly enhances the level of the Chinese Path to Sustainable Modernization.

2.3. Science and Technology Finance Ecology and Sustainable Modernization Under the Big Data Innovation Pilot Framework

Science and technology are primary productive forces, while finance is the lifeblood of the real economy. When combined, big data and the Science and Technology Finance Ecology offer a powerful engine for advancing Chinese-style modernization. National Big Data Comprehensive Pilot Zones, under supportive policy guidance and data-driven innovation, focus on enhancing regional economic growth, urban–rural coordination, cultural advancement, ecological sustainability, and overall social modernization.

Given the alignment between the goals of Chinese-style modernization and the objectives of pilot zones, several reinforcing mechanisms emerge. First, the government’s strong emphasis on big data innovation reflects a need to engage enterprises in digital transformation. In turn, firms responding to these goals may receive policy returns such as subsidies, easier access to financing, and simplified regulatory procedures, establishing a positive feedback loop [25]. Second, smart infrastructure (e.g., industrial robots) drives digital transformation in manufacturing, while digital platforms (e.g., digital finance) expand access to financial services, enhancing the inclusiveness and efficiency of resource allocation [26].

Simultaneously, reconciling ecological protection with economic growth is a core challenge of sustainable modernization. The deep integration of big data helps break “information silos” and facilitates the flow of technology, talent, knowledge, and capital into green R&D. This allows for the fusion of digital and traditional sectors and promotes structural upgrades toward a green economy via cross-sector, cross-regional factor mobility [27]. Lastly, under the big data innovation framework, the digital economy exerts multiplier effects: industrial digitization facilitates the intelligent transformation of traditional sectors, while digital industrialization drives innovation in business models and industrial restructuring [28].

Based on this analysis, we propose the following hypotheses:

H4. 

Big data innovation pilot programs amplify the positive effect of the Science and Technology Finance Ecology on the Chinese Path to Sustainable Modernization.

H5a. 

Big data innovation pilot programs promote Chinese-style sustainable modernization by improving the Science and Technology Finance Ecology.

H5b. 

Big data innovation pilot programs promote Chinese-style sustainable modernization by increasing government attention.

H5c. 

Big data innovation pilot programs promote Chinese-style sustainable modernization by advancing intelligent and inclusive development.

H5d. 

Big data innovation pilot programs promote Chinese-style sustainable modernization by enhancing green innovation capacity.

H5e. 

Big data innovation pilot programs promote Chinese-style sustainable modernization by optimizing the industrial structure.

3. Quasi-Natural Experiment Design

3.1. Model Construction

3.1.1. Spatial Difference-in-Differences (SDID) Model

This study investigates the impact of the implementation of the National Big Data Comprehensive Pilot Zone (NBDCPZ) policy and the development of Science and Technology Finance Ecology (SFE) on the Chinese Path to Sustainable Modernization (CPM) across 30 provinces, municipalities, and autonomous regions in mainland China (excluding Tibet) from 2009 to 2021. To better address spatial endogeneity and correlation, this paper follows the approach of Chagas [29], integrating the traditional Difference-in-Differences (DID) method with spatial econometric models to construct a Spatial Difference-in-Differences (SDID) framework.

$$\mathrm{Model}1:{\mathrm{C}\mathrm{P}\mathrm{M}}_{\mathrm{i}\mathrm{t}}=\mathrm{\rho }\mathrm{W}·{\mathrm{C}\mathrm{P}\mathrm{M}}_{\mathrm{i}\mathrm{t}}+{\mathrm{\alpha }}_1{\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}}+{\mathrm{\alpha }}_2{\mathrm{S}\mathrm{F}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+\mathrm{\Sigma }\mathrm{\beta }{\mathrm{X}}_{\mathrm{i}\mathrm{t}}+{{\mathrm{\gamma }}_{\mathrm{i}}+{\mathrm{\mu }}_{\mathrm{i}}+\mathrm{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(1)

$$\mathrm{Model}2:{\mathrm{C}\mathrm{P}\mathrm{M}}_{\mathrm{i}\mathrm{t}}={\mathrm{\alpha }}_1{\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}}+{\mathrm{\alpha }}_2{\mathrm{S}\mathrm{F}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+\mathrm{\Sigma }\mathrm{\beta }{\mathrm{X}}_{\mathrm{i}\mathrm{t}}+{{\mathrm{\gamma }}_{\mathrm{i}}+\mathrm{\lambda }\mathrm{W}·{\mathrm{v}}_{\mathrm{i}\mathrm{t}}+{\mathrm{\mu }}_{\mathrm{i}}+\mathrm{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(2)

$$\mathrm{Model}3:{\mathrm{C}\mathrm{P}\mathrm{M}}_{\mathrm{i}\mathrm{t}}=\mathrm{\rho }\mathrm{W}·{\mathrm{C}\mathrm{P}\mathrm{M}}_{\mathrm{i}\mathrm{t}}+{\mathrm{\alpha }}_1{\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}}+{\mathrm{\alpha }}_2{\mathrm{S}\mathrm{F}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+\mathrm{\Sigma }\mathrm{\beta }{\mathrm{X}}_{\mathrm{i}\mathrm{t}}+\mathrm{\theta }\mathrm{W}·({\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}}+{\mathrm{S}\mathrm{F}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+\mathrm{\Sigma }{\mathrm{X}}_{\mathrm{i}\mathrm{t}}){{+\mathrm{\gamma }}_{\mathrm{i}}+{\mathrm{\mu }}_{\mathrm{i}}+\mathrm{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(3)

Equations (1)–(3) represent the integration of the DID method with the Spatial Autoregressive Model (SAR), Spatial Error Model (SEM), and Spatial Durbin Model (SDM), respectively. The dependent variable is the Chinese Path to Sustainable Modernization (CPM), while $\mathrm{X}$ represents a vector of control variables, ${\mathrm{\gamma }}_{\mathrm{i}}$ denotes time fixed effects, ${\mathrm{\mu }}_{\mathrm{i}}$ denotes spatial fixed effects, and ${\mathrm{\epsilon }}_{\mathrm{i}\mathrm{t}}$ denotes the error term. The coefficients ${\mathrm{\alpha }}_1,{\mathrm{\alpha }}_2\mathrm{a}\mathrm{n}\mathrm{d}{\mathrm{\alpha }}_3$ correspond to the policy dummy variable for NBDCPZ (DID), the Science and Technology Finance Ecology (SFE), and their interaction term, respectively, while $\mathrm{\beta }$ represents the coefficient vector for control variables. Subscript $\mathrm{i}$ refers to provinces, municipalities, or autonomous regions, and $\mathrm{t}$ indicates the time period.

To examine the joint effect of the NBDCPZ policy implementation and the development of SFE on sustainable modernization, interaction terms are introduced based on Models 1–3. Taking the Spatial Durbin Model within the DID framework as an example, as shown in Equation (4):

$$\begin{gathered} \mathrm{Model}4:{\mathrm{C}\mathrm{P}\mathrm{M}}_{\mathrm{i}\mathrm{t}}=\mathrm{\rho }\mathrm{W}·{\mathrm{C}\mathrm{P}\mathrm{M}}_{\mathrm{i}\mathrm{t}}+{\mathrm{\alpha }}_1{\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}}+{\mathrm{\alpha }}_2{\mathrm{S}\mathrm{F}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+{\mathrm{\alpha }}_3{{\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}}\times \mathrm{S}\mathrm{F}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+\mathrm{\Sigma }\mathrm{\beta }{\mathrm{X}}_{\mathrm{i}\mathrm{t}}+\mathrm{\theta }\mathrm{W}·({\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}}+{\mathrm{S}\mathrm{F}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+ \\ {{\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}}\times \mathrm{S}\mathrm{F}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+\mathrm{\Sigma }{\mathrm{X}}_{\mathrm{i}\mathrm{t}})+{{\mathrm{\gamma }}_{\mathrm{i}}+{\mathrm{\mu }}_{\mathrm{i}}+\mathrm{\epsilon }}_{\mathrm{i}\mathrm{t}} \end{gathered}$$

(4)

3.1.2. Double Machine Learning (DML) Model

The Double Machine Learning (DML) model combines machine learning techniques with econometric methods. This approach effectively addresses two major limitations of traditional Difference-in-Differences models: the need to pre-specify the functional form of covariates and the “curse of dimensionality.” Moreover, it overcomes the shortcomings of standard machine learning—such as biased estimation, inability to construct confidence intervals, and slow convergence rate—while offering substantial advantages in capturing nonlinear relationships among economic variables. This provides a more robust method for causal inference in economics. This paper follows the approach of Chernozhukov [30] to construct the DML models, as shown in Equations (5)–(7):

$$\mathrm{Model}5:{\mathrm{C}\mathrm{P}\mathrm{M}}_{\mathrm{i}\mathrm{t}}={\mathrm{\alpha }}_1{\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}}+\mathrm{g}\left({\mathrm{X}}_{\mathrm{i}\mathrm{t}}\right){+\mathrm{U}}_{\mathrm{i}\mathrm{t}},\mathrm{E}\left({\mathrm{U}}_{\mathrm{i}\mathrm{t}}\right|{\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}},{\mathrm{X}}_{\mathrm{i}\mathrm{t}})=0$$

(5)

$$\mathrm{Model}6:{\mathrm{C}\mathrm{P}\mathrm{M}}_{\mathrm{i}\mathrm{t}}={\mathrm{\alpha }}_2{\mathrm{S}\mathrm{F}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+\mathrm{g}\left({\mathrm{X}}_{\mathrm{i}\mathrm{t}}\right){+\mathrm{U}}_{\mathrm{i}\mathrm{t}},\mathrm{E}\left({\mathrm{U}}_{\mathrm{i}\mathrm{t}}\right|{\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}},{\mathrm{X}}_{\mathrm{i}\mathrm{t}})=0$$

(6)

$$\mathrm{Model}7:{\mathrm{C}\mathrm{P}\mathrm{M}}_{\mathrm{i}\mathrm{t}}={\mathrm{\alpha }}_3{\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}}\times {\mathrm{S}\mathrm{F}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+\mathrm{g}\left({\mathrm{X}}_{\mathrm{i}\mathrm{t}}\right){+\mathrm{U}}_{\mathrm{i}\mathrm{t}},\mathrm{E}\left({\mathrm{U}}_{\mathrm{i}\mathrm{t}}\right|{\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}}\times {\mathrm{S}\mathrm{F}\mathrm{E}}_{\mathrm{i}\mathrm{t}},{\mathrm{X}}_{\mathrm{i}\mathrm{t}})=0$$

(7)

To correct for the regularization bias in parameter estimation, an auxiliary equation is introduced, as shown in Equation (8):

$${\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}}=\mathrm{m}\left({\mathrm{X}}_{\mathrm{i}\mathrm{t}}\right){+\mathrm{V}}_{\mathrm{i}\mathrm{t}},\mathrm{E}\left({\mathrm{V}}_{\mathrm{i}\mathrm{t}}\right|{\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}},{\mathrm{X}}_{\mathrm{i}\mathrm{t}})=0$$

(8)

Here, $\mathrm{g}\left({\mathrm{X}}_{\mathrm{i}\mathrm{t}}\right)$ and $\mathrm{m}\left({\mathrm{X}}_{\mathrm{i}\mathrm{t}}\right)$ represent unknown functions, but their estimators $\widehat{\mathrm{g}}\left({\mathrm{X}}_{\mathrm{i}\mathrm{t}}\right)$ and $\widehat{\mathrm{m}}\left({\mathrm{X}}_{\mathrm{i}\mathrm{t}}\right)$ can be obtained via machine learning, ${\mathrm{U}}_{\mathrm{i}\mathrm{t}}$ and ${\mathrm{V}}_{\mathrm{i}\mathrm{t}}$ represent the error term. The estimated residual ${\widehat{\mathrm{V}}}_{\mathrm{i}\mathrm{t}}$ is used as an instrumental variable for ${\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}}$, resulting in the estimator ${\widehat{\mathrm{\alpha }}}_1$ as shown in Equation (9), where $\mathrm{n}$ denotes the total sample size.

$${\widehat{\mathrm{\alpha }}}_1={({\displaystyle \frac{1}{\mathrm{n}}}\sum _{\mathrm{i}\mathrm{ϵ}\mathrm{I},\mathrm{t}\mathrm{ϵ}\mathrm{T}}{\widehat{\mathrm{V}}}_{\mathrm{i}\mathrm{t}}{\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}})}^{-1}{\displaystyle \frac{1}{\mathrm{n}}}\sum _{\mathrm{i}\mathrm{ϵ}\mathrm{I},\mathrm{t}\mathrm{ϵ}\mathrm{T}}{\widehat{\mathrm{V}}}_{\mathrm{i}\mathrm{t}}({\mathrm{C}\mathrm{P}\mathrm{M}}_{\mathrm{i}(\mathrm{t}+1)}-\widehat{\mathrm{g}}({\mathrm{X}}_{\mathrm{i}\mathrm{t}}))$$

(9)

To further investigate the potential mediating effects among variables, this paper constructs a DML-based mediation testing mechanism, taking the Science and Technology Finance Ecology as the mediating variable. This approach is based on the studies by He Jin’an [31] and Ni Xuanming [32].

$$\mathrm{Model}8:{\mathrm{C}\mathrm{P}\mathrm{M}}_{\mathrm{i}\mathrm{t}}={\mathrm{\alpha }}_1{\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}}+\mathrm{g}\left({\mathrm{X}}_{\mathrm{i}\mathrm{t}}\right){+\mathrm{U}}_{\mathrm{i}\mathrm{t}},\mathrm{E}\left({\mathrm{U}}_{\mathrm{i}\mathrm{t}}\right|{\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}},{\mathrm{X}}_{\mathrm{i}\mathrm{t}})=0$$

(10)

$$\mathrm{Model}9:{\mathrm{S}\mathrm{F}\mathrm{E}}_{\mathrm{i}\mathrm{t}}={\mathrm{\beta }}_1{\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}}+\mathrm{g}\left({\mathrm{X}}_{\mathrm{i}\mathrm{t}}\right){+\mathrm{U}}_{\mathrm{i}\mathrm{t}},\mathrm{E}\left({\mathrm{U}}_{\mathrm{i}\mathrm{t}}\right|{\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}},{\mathrm{X}}_{\mathrm{i}\mathrm{t}})=0$$

(11)

$$\mathrm{Model}10:{\mathrm{C}\mathrm{P}\mathrm{M}}_{\mathrm{i}\mathrm{t}}={\mathrm{\alpha }}_1^{\prime }{\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}}+{\mathrm{\alpha }}_2{\mathrm{S}\mathrm{F}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+\mathrm{g}\left({\mathrm{X}}_{\mathrm{i}\mathrm{t}}\right){+\mathrm{U}}_{\mathrm{i}\mathrm{t}},\mathrm{E}\left({\mathrm{U}}_{\mathrm{i}\mathrm{t}}\right|{\mathrm{D}\mathrm{I}\mathrm{D}}_{\mathrm{i}\mathrm{t}},{\mathrm{X}}_{\mathrm{i}\mathrm{t}})=0$$

(12)

In this system, ${\mathrm{C}\mathrm{P}\mathrm{M}}_{\mathrm{i}\mathrm{t}}$ represents the Chinese Path to Sustainable Modernization indicator. Equation (10) estimates the total effect of the NBDCPZ policy on sustainable modernization; Equation (11) estimates the effect of the policy variable on the mediating variable, SFE; and Equation (12) estimates the effect of the policy on the outcome variable while accounting for the mediating role of SFE.

3.2. Variable Selection and Data Sources

3.2.1. Dependent Variable: Chinese Path to Modernization (CPM)

According to the report of the 20th National Congress of the Communist Party of China, the Chinese path to modernization encompasses five key dimensions: modernization for a large population, common prosperity for all people, coordination between material and cultural–ethical advancement, harmony between humanity and nature, and peaceful development. In designing the measurement framework for CPM, it is essential to fully incorporate the theoretical implications of these five dimensions while ensuring data availability and the selection of scientifically sound and quantifiable indicators.

Modernization for a large population reflects both China’s demographic realities and its people-centered development philosophy. This dimension is implicitly captured by using per capita economic indicators. The modernization path of peaceful development involves political, military, and diplomatic considerations. In the context of globalization, increasing openness to the outside world is a necessary choice [33], while enhancing economic and technological strength forms a solid foundation for safeguarding world peace. Therefore, indicators such as import and export volumes, foreign direct investment (FDI), and R&D expenditure and personnel are employed to reflect this dimension.

Following the framework proposed by Ren Baoping [34], this study constructs a comprehensive evaluation system for the level of Chinese-style modernization from five aspects: modernization of economic development, urban–rural coordination, cultural–ethical progress, ecological sustainability, and social progress (see

Table 1. Indicator System for Measuring the Level of Chinese Path to Modernization.

Secondary Dimension	Tertiary Dimension	Proxy Indicator	Attribute
Economic Development	Economic Performance	GDP per capita	+
	Economic Structure & Efficiency	Advanced industrial structure index	+
		Rationalized industrial structure index	+
	Marketization	Marketization index	+
	Openness to the World	FDI/GDP	+
		Total imports and exports/GDP	+
	Innovation Capacity	R&D expenditure/GDP	+
		Technology market transaction value/GDP	+
Urban–Rural Coordination	Urbanization Level	Urbanization rate	+
	Income Level	Urban residents’ per capita disposable income	+
		Rural residents’ per capita disposable income	+
	Income Gap	Urban–rural income Theil index	−
Cultural–Ethical Progress	Cultural Investment	Culture, sports & media expenditure/Fiscal spending	+
		Public library collections per capita	+
		Broadcast coverage rate	+
	Cultural Industry	Number of library-hosted public lectures	+
		Number of performances by art troupes	+
Ecological Sustainability	Resource Consumption	Water resources per capita	+
		Electricity consumption per unit GDP	−
	Pollution Control	SO2 emissions	−
		NOX emissions	−
	Environmental Governance	Daily urban sewage treatment capacity	+
		Harmless treatment rate of household waste	+
	Urban Greening	Green coverage ratio in built-up areas	+
		Forest coverage rate	+
Social Progress	Employment	Employment rate	+
	Education	Average years of schooling	+
		Student–teacher ratio in primary & secondary schools	−
		Public education expenditure	+
	Healthcare	Practicing physicians per 10,000 people	+
		Number of top-tier hospitals	+
	Social Security	Social security & employment expenditure/Fiscal spending	+
	Infrastructure	Urban road area per capita	+
		Public transport ridership	+
		Broadband access ports per 10,000 people	+
	Government Efficiency	Per capita public service expenditure	+

Note: “+” indicates that the indicator is positively associated with the level of Chinese modernization, while “−” indicates a negative association.

). Both overall and sub-dimensional indices are calculated using the entropy method. Data are primarily sourced from the China Statistical Yearbook, China Urban Statistical Yearbook, China Science and Technology Statistical Yearbook, China High-Tech Industry Statistical Yearbook, Compilation of Statistics on Science and Technology of Higher Education Institutions, and the iFinD database by Tonghuashun, covering the period from 2009 to 2021.

3.2.2. Independent Variables

(1)

National Big Data Comprehensive Pilot Zone Policy (DID)

To promote big data innovation and development, the National Development and Reform Commission (NDRC), the Ministry of Industry and Information Technology (MIIT), and the Cyberspace Administration of China approved Guizhou Province in February 2016 as the first national-level Big Data Comprehensive Pilot Zone. In October of the same year, the NDRC and MIIT expanded the policy to seven additional regions, including the Beijing–Tianjin–Hebei area. Accordingly, the treatment group in this study includes provinces where pilot zones were established: Guizhou, Beijing, Tianjin, Hebei, Guangdong, Shanghai, Henan, Chongqing, Shenyang, and Inner Mongolia.

To maintain consistency in measurement and align with the research objective, this paper uses provincial-level data from 30 mainland provinces (excluding Tibet). Although Shenyang is a sub-provincial city, it serves as a regional growth pole within multi-core provinces and promotes high-quality development through metropolitan integration [35]; thus, it is reassigned to Liaoning Province in the treatment group. Following the quasi-natural experiment design, provinces included in the pilot zones during the policy period are assigned a value of 1; all others are assigned 0.

(2)

Science and Technology Finance Ecology (SFE)

To define and measure the science and technology finance ecology, this paper draws on the studies of Bai Yujuan [36] and Chen Qiang [37], which adopt an ecological systems perspective to reflect the integration of science and technology finance with ecological thinking. The framework includes four components: producers, consumers, decomposers, and the abiotic environment, as shown in

Table 2. Indicator System for Measuring Science and Technology Finance Ecology.

Secondary Indicator	Tertiary Indicator	Proxy Variable	Attribute
S&T Finance Producers	Universities	Number of regular higher education institutions	+
		Full-time equivalent R&D personnel in universities	+
		Internal R&D expenditure of universities	+
		Number of R&D projects in universities	+
		Number of patent applications by universities	+
	Research Institutes	Number of R&D institutions	+
		Full-time equivalent R&D personnel in research institutes	+
		Internal R&D expenditure of research institutes	+
		Number of R&D projects in research institutes	+
	Enterprises	Number of R&D-active industrial enterprises (above scale)	+
		Full-time equivalent R&D personnel in enterprises	+
		Internal R&D expenditure of enterprises	+
		Number of R&D projects in enterprises	+
S&T Finance Decomposers	Financial Institutions	Employment in urban financial sector	+
		Tech loan balance of financial institutions	+
		Total stock market value	+
	Intermediaries	Number of university science parks	+
		Number of business incubators	+
		Number of venture capital institutions	+
		Intensity of venture capital investment	+
S&T Finance Consumers	Support for Tech Commercialization	Number of listed tech enterprises	+
		Total bank credit for technology	+
	Effectiveness of Tech Transfer	Sales revenue of new products (above-scale enterprises)	+
		Number of new product development projects	+
		Contract value of technology transactions	+
S&T Finance Environment	Infrastructure	Year-on-year growth in fixed investment in sci-tech services	+
		Internet penetration rate	+
		Number of academic conferences hosted by provincial S&T associations	+
	Market Support	FDI as % of GDP	+
		Marketization index	+
		Total output value of financial industry	+
	Government Support	Strength of IP protection	+
		Local government education expenditure as % of total	+
		Local government sci-tech expenditure as % of total	+

Note: “+” denotes a positive indicator (higher values indicate a higher level of science and technology finance ecology).

. To ensure data availability, comprehensiveness, and innovation, the indicator system based on ecological principles is further refined with reference to the works of Fang Lei [7] and Pan Jie [38].

The SFE index is calculated using the entropy weight method. Data are primarily sourced from China Statistical Yearbook, China Science and Technology Statistical Yearbook, China High-Tech Industry Statistical Yearbook, China Information Industry Yearbook, China Torch Statistical Yearbook, China Financial Statistics Yearbook, provincial statistical yearbooks and bulletins, the Peking University Digital Inclusive Finance Index, and the iFinD database by Tonghuashun.

3.2.3. Control Variables

To better isolate the effects of the Big Data Innovative Development Pilot and the science and technology finance ecology on the process of Chinese path to modernization, this study includes a set of control variables to eliminate interference from other potential influencing factors. These variables include high-tech employment agglomeration, high-tech enterprise agglomeration, environmental regulation, tax burden, social consumption level, and innovation level (see

Table 3. Control Variables.

Control Variable	Definition	Measurement Description
High-Tech Employment Agglomeration (TP)	Concentration of personnel engaged in high-tech industries within a region	$\mathrm{H}\mathrm{T}\mathrm{P}={\displaystyle \frac{{\mathrm{p}}_{\mathrm{i}}/{\mathrm{l}}_{\mathrm{i}}}{\mathrm{P}/\mathrm{L}}}$ ${\mathrm{p}}_{\mathrm{i}}$ and ${\mathrm{l}}_{\mathrm{i}}$ represent the number of high-tech industry employees and total employees in a given region; P and L represent the corresponding national figures
High-Tech Enterprise Agglomeration (TE)	Clustering of interrelated, complementary, or competing high-tech enterprises within a region	$\mathrm{H}\mathrm{T}\mathrm{E}={\displaystyle \frac{{\mathrm{e}}_{\mathrm{i}}/{\mathrm{l}}_{\mathrm{i}}}{\mathrm{E}/\mathrm{L}}}$ ${\mathrm{e}}_{\mathrm{i}}$ and ${\mathrm{l}}_{\mathrm{i}}$ represent the number of high-tech enterprises and total employees in a given region; E and L represent the corresponding national figures
Environmental Regulation (ER)	Regulations targeting pollution to protect the environment	$\mathrm{E}\mathrm{R}={\displaystyle \frac{\mathrm{I}\mathrm{n}\mathrm{v}\mathrm{e}\mathrm{s}\mathrm{t}\mathrm{m}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{i}\mathrm{n}\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{u}\mathrm{s}\mathrm{t}\mathrm{r}\mathrm{i}\mathrm{a}\mathrm{l}\mathrm{p}\mathrm{o}\mathrm{l}\mathrm{l}\mathrm{u}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}}{\mathrm{O}\mathrm{u}\mathrm{t}\mathrm{p}\mathrm{u}\mathrm{t}\mathrm{v}\mathrm{a}\mathrm{l}\mathrm{u}\mathrm{e}\mathrm{o}\mathrm{f}\mathrm{s}\mathrm{e}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{a}\mathrm{r}\mathrm{y}\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{u}\mathrm{s}\mathrm{t}\mathrm{r}\mathrm{y}}}$
Tax Burden (TL)	The extent of tax pressure	$\mathrm{T}\mathrm{L}={\displaystyle \frac{\mathrm{T}\mathrm{a}\mathrm{x}\mathrm{r}\mathrm{e}\mathrm{v}\mathrm{e}\mathrm{n}\mathrm{u}\mathrm{e}}{\mathrm{G}\mathrm{D}\mathrm{P}}}$
Social Consumption Level (CL)	The degree of material and cultural consumption by society	$\mathrm{C}\mathrm{L}={\displaystyle \frac{\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{r}\mathrm{e}\mathrm{t}\mathrm{a}\mathrm{i}\mathrm{l}\mathrm{s}\mathrm{a}\mathrm{l}\mathrm{e}\mathrm{s}\mathrm{o}\mathrm{f}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{s}\mathrm{u}\mathrm{m}\mathrm{e}\mathrm{r}\mathrm{g}\mathrm{o}\mathrm{o}\mathrm{d}\mathrm{s}}{\mathrm{G}\mathrm{D}\mathrm{P}}}$
Innovation Level (IL)	Capacity to generate new ideas, theories, or methods through technology and practice	$\mathrm{I}\mathrm{L}=\ln \left(\mathrm{N}\mathrm{D}\mathrm{I}\mathrm{P}\mathrm{A}\mathrm{A}\right)$

Note: NDIPAA = Number of Domestic Invention Patent Applications Accepted.

), aiming to address endogeneity problems caused by omitted variables. Specifically, high-tech employment agglomeration (TP) and high-tech enterprise agglomeration (TE) are calculated following the methodology of Huang Yongchun [39], environmental regulation (ER) is measured based on the formula provided by Liu Rongzeng [40], and data on tax burden (TL), social consumption level (CL), and innovation level (IL) are obtained from the China Statistical Yearbook.

3.2.4. Mechanism Variables

To align with the policy goals of the National Big Data Comprehensive Pilot Zone, this paper further investigates the mechanisms by which these policy events influence the Chinese Path to Sustainable Modernization, focusing on four dimensions: enhancing government attention, promoting intelligent and inclusive development, strengthening green innovation efficiency, and optimizing industrial structure.

Specifically, the proportion of keywords such as “digital economy” and “big data” in government work reports is used to measure Government Attention (GAT). The Intelligent and Inclusive Development mechanism is divided into two subcomponents: intelligent development and inclusive development. Intelligent development (INT) is measured by industrial robot installation density, following the estimation method of Lu Tingting [41], which first collects industrial robot installation data from the IFR alliance and adjusts it by the proportion of employment in each province within the corresponding economic sectors reported in the China Labour Statistical Yearbook. Inclusive development (INC) is measured using the Digital Inclusive Finance Index developed by Peking University.

Green Innovation Efficiency (GIN) is measured by the number of green patent applications. Advanced Industrial Structure (AIS) is calculated based on the formula proposed by Xu Min [42], which evaluates the transition of the industrial structure from lower to higher levels.

3.2.5. Spatial Weight Matrix

Due to the existence of demonstration, diffusion, and siphon effects, spatial spillovers may occur among neighboring or economically interconnected regions. To account for this, a spatial weight matrix is introduced.

Given that the relationships among the explanatory and dependent variables—namely the development of the National Big Data Comprehensive Pilot Zones, the sci-technology finance ecology, and Chinese-style sustainable modernization—are fundamentally economic, this study adopts an economic spatial weight matrix. Specifically, the weight is defined as the inverse of the absolute difference in real per capita GDP between two regions. This design reflects the principle that greater economic disparity implies a weaker connection, and thus a smaller corresponding weight.

$${\mathrm{w}}_{\mathrm{i}\mathrm{j}}=\left\{\begin{array}{c}{\displaystyle \frac{1}{|{\mathrm{d}}_{\mathrm{i}}-{\mathrm{d}}_{\mathrm{j}}|}}\\ 1\mathrm{i}=\mathrm{j}\end{array}\mathrm{i}\ne \mathrm{j}\right.$$

(13)

In Equation (13), ${\mathrm{w}}_{\mathrm{i}\mathrm{j}}$ represents an element in the economic spatial weight matrix; ${\mathrm{d}}_{\mathrm{i}}$ and ${\mathrm{d}}_{\mathrm{j}}$ denote the real per capita GDP of regions $\mathrm{i}$ and $\mathrm{j}$, respectively. The weight assigned to a region itself is set to 1.

4. Empirical Results and Analysis

4.1. Spatial Difference-in-Differences Model Analysis

4.1.1. Spatial Autocorrelation Analysis

We first use the global Moran’s I index to test whether the spatial distribution of Chinese-style sustainable modernization exhibits spatial autocorrelation, ensuring the suitability of applying spatial econometric models. The test results show that from 2009 to 2021, the global Moran’s I index of Chinese-style sustainable modernization passed the 1% significance level test each year, mostly displaying “high-high” or “low-low” clustering patterns between regions. This indicates significant spatial autocorrelation of the dependent variable—Chinese-style sustainable modernization—as shown in

Table 4. Global Moran’s I, p-values, and Z-values of Chinese-style sustainable modernization (2009–2021).

Year	Moran’s I	p Value	Z Value	Year	Moran’s I	p Value	Z Value
2009	0.277 ***	0.000	14.544	2016	0.251 ***	0.000	12.967
2010	0.276 ***	0.000	14.350	2017	0.244 ***	0.000	12.622
2011	0.271 ***	0.000	13.990	2018	0.226 ***	0.000	11.807
2012	0.273 ***	0.000	14.155	2019	0.212 ***	0.000	11.155
2013	0.266 ***	0.000	13.822	2020	0.206 ***	0.000	10.877
2014	0.267 ***	0.000	13.837	2021	0.199 ***	0.000	10.525
2015	0.256 ***	0.000	13.194

t statistics in parentheses, *** p < 0.01.

and Figure 1.

4.1.2. Identification, Selection, and Testing of Spatial Econometric Models

To determine which spatial econometric model should be combined with the difference-in-differences (DID) method, in this study, we conduct both pre-tests and post-tests to identify, select, and validate spatial models. The pre-test uses the LM test to judge whether spatial terms need to be introduced. The post-tests aim to select the appropriate model by applying LR tests, Wald tests, and Hausman tests, as shown in

Table 5. Identification, Selection, and Testing of Spatial Econometric Models.

Statistic	Value	p Value
LM test (SAR)	108.963 ***	0.000
Robust LM test (SAR)	46.072 ***	0.000
LM test (SEM)	126.989 ***	0.000
Robust LM test (SEM)	46.072 ***	0.000
LR test (SAR)	121.63 ***	0.000
LR test (SEM)	138.43 ***	0.000
Wald test (SAR)	96.06 ***	0.000
Wald test (SEM)	77.44 ***	0.000
Hausman test (Random Effects)	181.89 ***	0.000

t statistics in parentheses, *** p < 0.01.

.

The test results corresponding to these statistics are shown in Table 5. Under the economic spatial weight matrix, both the spatial lag model (SAR) and spatial error model (SEM) pass the LM and Robust LM tests at the 1% significance level, preliminarily indicating that the spatial Durbin model (SDM) is the most suitable choice. Both the Wald and LR tests reject the null hypothesis at the 1% significance level, demonstrating that the SDM will not degenerate into the SAR or SEM. Additionally, the Hausman test for random effects is significant at the 1% level, supporting the use of random effects in the SDM applied in this study. In summary, we determine that combining the difference-in-differences method with the spatial Durbin model (SDM-DID) is the appropriate spatial difference-in-differences model.

4.1.3. Regression Analysis of the Spatial Difference-in-Differences Model

In this study, we select the difference-in-differences spatial Durbin model (SDM-DID) with random effects as the spatial econometric model to examine the impact relationship among the big data innovation development pilot, the Science and Technology Finance Ecology (SFE), and Chinese-style sustainable modernization. To further analyze the spatial effects of the variables, this study reports local effects, neighboring effects, and total effects in the regression results. Local effects represent the impact of explanatory variables on the explained variable within the same region, including feedback effects; that is, the effect of the explanatory variable on other regions which then feedback to the original region. Neighboring effects measure how an explanatory variable in one region affects the explained variable in adjacent regions. Total effects are the sum of local and neighboring effects and can be interpreted as the average impact of a region’s explanatory variable changes on all regions’ explained variables. The parameter estimation results of the SDM-DID model are shown in

Table 6. Parameter Estimation Results for Model 3 and Model 4.

	Model 3			Model 4
Variable	Local Effect	Neighboring Effect	Total Effect	Local Effect	Neighboring Effect	Total Effect
DID	0.041 ***	0.343 **	0.384 **	−0.001	−0.090 **	−0.091 **
	(0.000)	(0.027)	(0.016)	(0.783)	(0.033)	(0.039)
SFE	1.248 ***	0.488	1.736 ***	0.917 ***	0.190	1.107 ***
	(0.000)	(0.388)	(0.002)	(0.000)	(0.598)	(0.002)
DID × SFE				0.610 ***	2.084 ***	2.694 ***
				(0.000)	(0.000)	(0.000)
TP	0.004	0.977 *	0.981 *	0.012	0.004	0.016
	(0.857)	(0.077)	(0.087)	(0.248)	(0.979)	(0.922)
TE	0.028	−0.774 *	−0.746	−0.001	0.015	0.014
	(0.163)	(0.080)	(0.104)	(0.918)	(0.913)	(0.922)
ER	0.905	10.899	11.804	−1.899 **	−27.601 ***	−29.500 ***
	(0.455)	(0.582)	(0.567)	(0.017)	(0.002)	(0.002)
TL	−0.369 **	6.094 **	5.725 *	−0.167	1.150	0.983
	(0.046)	(0.037)	(0.058)	(0.208)	(0.234)	(0.322)
CL	0.013	−0.357	−0.344	0.163 ***	2.345 ***	2.508 ***
	(0.878)	(0.808)	(0.822)	(0.004)	(0.001)	(0.001)
IL	−0.021 ***	−0.070	−0.092 *	−0.008 **	0.020	0.012
	(0.000)	(0.173)	(0.085)	(0.026)	(0.278)	(0.529)
rho	0.644 ***			0.446 ***
	(0.000)			(0.000)
lgt_theta	−2.023 ***			−1.827 ***
	(0.000)			(0.000)
sigma2_e	0.001 ***			0.000 ***
	(0.000)			(0.000)
N	390			390
R2	0.822			0.919

t statistics in parentheses, * p < 0.10, ** p < 0.05, *** p < 0.01.

.

In Model 3, which excludes interaction terms, the local effect coefficient of the national big data comprehensive pilot policy variable is significant at the 1% level, while the neighboring and total effect coefficients are significant at the 5% level. This indicates that the pilot policy not only promotes the level of Chinese-style modernization locally but also stimulates development in surrounding areas through spillover effects, confirming H1 and H2. The local and total effect coefficients of the Science and Technology Finance Ecology variable are significant at the 1% level, reflecting that improvements in SFE significantly enhance Chinese-style modernization, confirming H3. However, existing institutional barriers may hinder the flow of innovative factors across regions [43], resulting in the weak neighboring effect of SFE.

After introducing the interaction term between the pilot policy and SFE, Model 4’s results show that the local, neighboring, and total effect coefficients of the interaction term are 0.610, 2.084, and 2.694, respectively, all significant at the 1% level. This indicates that the pilot policy of national big data comprehensive experimental zones strengthens the effect that SFE has on promoting Chinese-style sustainable modernization, verifying H4.

4.2. Double Machine Learning Model Analysis

4.2.1. Baseline Regression of Double Machine Learning

Referring to the study conducted by Zhang Tao [44], in this study, we use a double machine learning model based on the random forest algorithm to estimate the parameters for Models 5–7, with the sample split ratio set at 1:4. In Models 5 and 6, the regression coefficients of DID, SFE, and their interaction term are all significantly positive at the 1% level, with values of 0.073, 0.837, and 1.281, respectively. This result further confirms H1, H3, and H4 and aligns with the parameter estimation results of the spatial difference-in-differences model, as shown in

Table 7. Parameter Estimation Results for Models 5–7.

Variable	Model 5	Model 6	Model 7
DID	0.073 ***		0.073 ***
	(0.016)		(0.016)
SFE		0.837 ***	0.837 ***
		(0.128)	(0.128)
DID × SFE			1.281 ***
_cons	0.000	−0.000	−0.001
	(0.002)	(0.002)	(0.002)
N	390	390	390
R2	-	-	-

t statistics in parentheses, *** p < 0.01.

.

4.2.2. Robustness Tests

(1)

Excluding Potential Regional Factor Interference

This study is closely linked to regional development conditions. Certain provinces may uniquely influence the results due to their characteristics. Qinghai and Ningxia are inland and have a low population density; Hainan is an island lacking land connectivity to mainland provinces, and it has a small market size, placing it at a disadvantage in terms of regional economic cooperation. Moreover, except for Tibet, Qinghai (CNY 334.663 billion), Ningxia (CNY 452.23 billion), and Hainan (CNY 647.52 billion) ranked lowest in terms of their GDP in 2021, with large gaps compared to Gansu (CNY 1024.33 billion), which ranked 27th. Economically, these provinces lag nationwide. Thus, we reran the tests excluding Qinghai, Ningxia, and Hainan, with the parameter estimates being largely consistent with Table 7, ensuring that the model and conclusions are robust.

(2)

Resetting the Double Machine Learning Sample Split Ratio

The ratio for training and testing sets in machine learning is usually set arbitrarily. To avoid bias from this setting, the sample split ratio was adjusted to 1:2 for re-estimation. The results remain consistent with those detailed in Table 7, confirming their robustness.

(3)

Replacing Machine Learning Algorithms

Machine learning offers a variety of algorithms for different scenarios. Besides the random forest algorithm used in baseline regression, we replace it with Lasso regression and neural network algorithms for parameter estimation. The results, shown in

Table 8. Robustness Tests.

Robustness Test Content	Adjusted Sample Range (Excluding Qinghai, Ningxia, Hainan)	Reset Sample Split Ratio (K-Folds = 3)	Replace ML Algorithm (Lasso Regression)	Replace ML Algorithm (Neural Network)
DID	0.070 ***	0.079 ***	0.058 ***	0.268 ***
	(0.016)	(0.017)	(0.012)	(0.072)
SFE	0.768 ***	0.938 ***	1.384 ***	0.716 ***
	(0.126)	(0.182)	(1.384)	(0.124)
DID × SFE	0.547 **	1.334 ***	0.533 ***	0.338 ***
	(0.265)	(0.221)	(0.054)	(0.073)
Control Variables	Yes	Yes	Yes	Yes
Time Fixed Effects	Yes	Yes	Yes	Yes
Regional Fixed Effects	Yes	Yes	Yes	Yes
N	351	390	390	390

t statistics in parentheses, ** p < 0.05, *** p < 0.01.

, are broadly consistent with those outlined in Table 7, further supporting their robustness.

4.2.3. Testing for Omitted Variable Bias

Although we employ the double machine learning (DML) and spatial difference-in-differences (SDID) methods in this study to control for observable and unobservable confounders, we also recognize that several major macro events during the sample period (2009–2021) may have exerted potential influences on big data development, financial ecology, and the process of sustainable modernization. These events include multiple rounds of quantitative easing and fiscal stimulus following the global financial crisis [45,46]; policy uncertainty and international spillover effects triggered by events such as Brexit [47]; the impact of the COVID-19 pandemic on the labor market and macroeconomic policy [48,49]; and the rise in ESG funds influencing expected returns on sustainable projects [50]. To account for this, we follow the method proposed by Oster (2019) [51] and compute the parametric bounds of the treatment effect (DID) to assess the stability of the coefficients after controlling for both observable and unobservable variables.

The specific steps are as follows:

(1)

Estimate two OLS models:

Restricted Model: This includes only the treatment variable (DID) and a simplified set of control variables without fixed effects.

Full Model: This includes the treatment variable, the full set of control variables, and time and province fixed effects.

(2)

Compute coefficients and $R^2$:

Obtain the DID coefficient $\widehat{\beta }$ and $R^2$ value $R_{restricted}^2$ from the restricted model.

Obtain the DID coefficient $\tilde{\beta }$ and $R^2$ value $R_{full}^2$ from the full model.

(3)

Set the safety factor:

Following the literature [52,53], we adopt $\prod =1.3$ and $\prod =2.0$ to compute $R_{max}^2=min\left\{1,\prod \times R_{full}^2\right\}$.

(4)

Calculate the parametric bounds ${\beta }^{*}$:

The boundary values of the treatment effect are calculated using the following formula:

$${\beta }^{*}=\tilde{\beta }-\left[{\displaystyle \frac{R_{max}^2-R_{full}^2}{R_{full}^2-R_{restricted}^2}}\right](\tilde{\beta }-\widehat{\beta })$$

We conduct the above test for the dependent variables “Chinese Path to Sustainable Modernization (CPM)” and “Sci-technology Finance Ecology (SFE)”, and the results are reported in

Table 9. Results of Coefficient Stability Test Based on Oster (2019) [51].

Dependent Variable	Restricted Model	Full Model	${\mathit{R}}_{\mathit{f}\mathit{u}\mathit{l}\mathit{l}}^2$	${\mathit{R}}_{\mathit{r}\mathit{e}\mathit{s}\mathit{t}\mathit{r}\mathit{i}\mathit{c}\mathit{t}\mathit{e}\mathit{d}}^2$	${\mathit{\beta }}^{\mathit{*}}$$(\prod =1.3$)	${\mathit{\beta }}^{\mathit{*}}$$(\prod =2.0)$	Zero Included?
CPM	0.073 ***	0.085 ***	0.919	0.712	0.071 ***	0.069 ***	No
SFE	0.837 ***	0.921 ***	0.901	0.698	0.830 ***	0.825 ***	No

t statistics in parentheses, *** p < 0.01.

.

4.2.4. Heterogeneity Analysis

(1)

Differences in “Location–Resource–Environment” Conditions

Given China’s vast territory and pronounced regional heterogeneity, eastern provinces generally benefit from a more favorable innovation environment, better infrastructure, stronger awareness of green development, higher marketization levels, and greater talent agglomeration. These advantages are conducive to advancing the Chinese Path to Sustainable Modernization. In contrast, the central and western regions, characterized by resource-based cities and industrial layouts, tend to lag behind in terms of economic technology, transportation, capital, and human resources. Their rigid and single-industry structures may expose them to the “resource curse”, resulting in greater resistance to economic transformation. Based on geographic divisions, we conduct heterogeneity tests by grouping 10 eastern and 20 central–western provinces. The results show that the policy coefficient is slightly lower in eastern provinces, possibly because the digital economy in these areas developed earlier, forming a stable foundation. Thus, the incremental effect of Big Data Pilot Zone policies may be harder to isolate. However, the coefficients of Science and Technology Finance Ecology (SFE) and its interaction term are higher in the east, confirming that the advantages in terms of location, resources, and environment can significantly enhance sustainable modernization.

(2)

Urban–Rural Integration Pilot Zones

Urban–rural integration plays a vital role in promoting social equity. The diffusion and application of big data help to expand rural markets by reducing information asymmetries and enabling precision production, thereby increasing income through e-commerce. Innovation within the Science and Technology Finance Ecology fosters technological upgrades in production and the widespread promotion of agricultural technologies, transforming traditional practices and improving efficiency. Meanwhile, financial institutions develop customized products to address rural financing constraints. Based on the 11 national urban–rural integration pilot zones designated by the National Development and Reform Commission in 2019, in this study, we perform heterogeneity analysis by group. The results indicate that the policy coefficients are significantly larger and more statistically robust in pilot provinces, confirming that urban–rural integration pilots are more conducive to promoting sustainable modernization, further validating their role in the Chinese path to modernization.

(3)

Green Finance Reform and Innovation Pilot Zones

Green finance fosters sustainable modernization by stimulating green technological innovation, facilitating the transition to new economic drivers, reducing pollution and energy consumption, and improving both resource utilization and environmental quality. Based on the green finance pilot zones established by the State Council since 2017 in six provinces and nine cities, heterogeneity analysis reveals significantly higher policy coefficients in pilot provinces than in non-pilot ones. This supports the assertion that green finance reform pilots positively contribute to the modernization process and affirms that green development is also a key pillar of the Chinese path to modernization, as summarized in

Table 10. Heterogeneity Analysis Results.

Heterogeneity Analysis Dimension	Location–Resource–Environment Differences		Urban–Rural Integration Pilot		Green Finance Reform and Innovation Pilot
	Eastern Provinces	Central and Western Provinces	Pilot Zone	Non-Pilot Zone	Pilot Zone	Non-Pilot Zone
DID	0.046 ***	0.059 ***	0.063 ***	0.031 *	0.080 ***	0.041 ***
	(0.011)	(0.016)	(0.010)	(0.017)	(0.014)	(0.010)
SFE	1.180 ***	0.702 ***	0.765 ***	0.728 ***	0.802 ***	0.616 ***
	(0.139)	(0.051)	(0.098)	(0.061)	(0.068)	(0.077)
DID × SFE	0.241 ***	0.203 ***	0.215 ***	0.117 ***	0.243 ***	0.202 ***
	(0.077)	(0.029)	(0.042)	(0.029)	(0.044)	(0.041)
Control Variables	Yes	Yes	Yes	Yes	Yes	Yes
Time Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes
Region Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes
N	130	260	143	247	78	312

t statistics in parentheses, * p < 0.10, *** p < 0.01.

.

4.2.5. Mechanism Analysis

To test whether sci-tech finance ecology and four specific channels—government attention, inclusive intelligence development, green innovation capacity, and industrial structure upgrading—mediate the relationship between the National Big Data Comprehensive Pilot Zone policy and the process of the Chinese Path to Sustainable Modernization, we estimate Models 8–10 using the double machine learning (DML) framework, following the methodological standard. Parameter estimates, significance statistics, and the proportion of mediated effects are reported in Table 10 and

Table 11. Parameter Estimates for Mechanism Testing.

Mechanism Path	Total Effect	Direct Effect	Indirect Effect	Sobel (Z)	Aroian (Z)	Goodman (Z)	Mediation Share
DID → SFE → CPM	0.102 ***	0.044 ***	0.058 ***	5.340 ***	5.338 ***	5.342 ***	57.22%
DID → GAT → CPM	0.102 ***	0.078 ***	0.024 ***	3.845 ***	3.830 ***	3.861 ***	23.31%
DID → INT → CPM	0.102 ***	0.002	0.100 ***	9.226 ***	9.221 ***	9.231 ***	98.03%
DID → INC → CPM	0.102 ***	0.049 ***	0.053 ***	6.837 ***	6.820 ***	6.855 ***	51.68%
DID → GIN → CPM	0.102 ***	0.060 ***	0.042 ***	5.964 ***	5.945 ***	5.983 ***	40.98%
DID → AIS → CPM	0.102 ***	0.085 ***	0.017 **	2.467 **	2.460 **	2.474 **	16.94%

t statistics in parentheses, ** p < 0.05, *** p < 0.01.

. The Sobel, Aroian, and Goodman tests are used to assess the statistical significance of the mediating effects, while the mediation proportion refers to the share of the total effect that is explained by the mediating path.

The regression results suggest that the Big Data Innovation Development Pilot exerts a positive impact on Chinese-style sustainable modernization through five key mechanisms: improving the sci-tech finance ecology, increasing government attention, advancing inclusive intelligence development, enhancing green innovation performance, and optimizing industrial structure. The coefficients of the mediating effects are 0.058, 0.024, 0.100, 0.053, 0.042, and 0.017, respectively. The corresponding mediation proportions are 57.22%, 23.31%, 98.03%, 51.68%, 40.98%, and 16.94%.

Among these, the mechanism via intelligent development is the most prominent, exhibiting an almost complete mediation effect. This is followed by the sci-tech finance ecology, inclusive development, green innovation, government attention, and industrial upgrading. These findings support H5a to H5e (see Table 11 and

Table 12. Bootstrap Results for Mechanism Paths.

Mechanism Path	Indirect Effect	Bootstrap SE	95% Percentile CI	95% Bias-Corrected CI	Z-Value	p-Value
DID → SFE → CPM	0.058 ***	0.013	[0.0314, 0.0844]	[0.0314, 0.0844]	4.36	0.000
DID → GAT → CPM	0.024 ***	0.005	[0.0131, 0.0338]	[0.0136, 0.0339]	4.63	0.000
DID → INT → CPM	0.100 ***	0.015	[0.0743, 0.1280]	[0.0746, 0.1292]	6.87	0.000
DID → INC → CPM	0.053 ***	0.007	[0.0398, 0.0651]	[0.0403, 0.0666]	7.85	0.000
DID → GIN → CPM	0.042 ***	0.006	[0.0305, 0.0537]	[0.0310, 0.0548]	7.11	0.000
DID → AIS → CPM	0.017 **	0.005	[0.0079, 0.0264]	[0.0079, 0.2688]	3.50	0.000

t statistics in parentheses, ** p < 0.05, *** p < 0.01.

for detailed results).

5. Conclusions and Policy Implications

5.1. Conclusions

In this study, we apply both a spatial difference-in-differences (SDID) model and a double machine learning (DML) approach to panel data from 30 provinces in China (2009–2021) under a quasi-natural experiment design, examining the interaction between the Big Data Innovation Development Pilot, the sci-tech finance ecology, and the Chinese Path to Sustainable Modernization.

The main findings of this study are as follows:

(1) The results of the SDID and DML models show that pilot regions under the Big Data Innovation Development Pilot have significantly advanced Chinese-style sustainable modernization through demonstration and diffusion effects, supported by location advantages, industrial foundations, resource endowments, policy guidance, and local actions. The sci-tech finance ecology further enhances modernization and amplifies the positive role played by pilot zones. Compared with existing studies that mainly focus on theoretical analyses or firm-level outcomes, this study provides provincial-level empirical evidence.

(2) The results of the heterogeneity analysis reveal that the effects vary across regions due to differences in resource endowments, development models, and policy environments, with stronger impacts in more developed provinces and in areas implementing urban–rural integration or green finance reforms. This highlights the regional contingencies of policy effectiveness.

(3) Mechanism analysis confirms that the Big Data Pilot policy mediates the impact of the sci-tech finance ecology on sustainable modernization through increasing government attention, fostering inclusive intelligence development, enhancing green innovation capacity, and upgrading industrial structures. Unlike prior research that discusses these mechanisms conceptually, our study quantifies the mediating pathways, providing new evidence on how digital and financial innovations jointly drive modernization.

5.2. Policy Implications

(1) Leverage regional strengths to reduce disparities: Pilot zones should use local advantages to drive reform and broader regional development, while non-pilot regions learn from their experiences. Tailored government policies and strong government–market coordination are essential to guide lagging regions and advance national big data innovation.

(2) Establish a unified big data platform: Expanding data collection, improving integration, and breaking information silos will enhance connectivity. High-tech talent and core technologies (blockchain, cloud computing, AI, and machine learning) should be focused on to support digital transformation and expand applications across industries.

(3) Promote the free flow of data, knowledge, and talent: Inter-regional and cross-sector cooperation should be strengthened to enhance positive spatial spillovers, while macro-level regulations should be implemented to prevent excessive siphoning and widening regional gaps.

(4) Strengthen the mediating role played by the sci-tech finance ecology: Financial innovation should fund enterprise innovation and digital transformation, with market mechanisms guiding resource allocation, identifying promising prospects, and improving risk management.

(5) Build a coordinated innovation–modernization system: Big data should be integrated with financial and technology markets to advance marketization, digitalization, intelligence, and innovation, supporting China’s innovation-driven, sustainable, and regionally coordinated modernization strategy.

5.3. Research Limitations

In this study, we employ advanced econometric methods to evaluate the impact of the National Big Data Comprehensive Pilot Zone policy on China’s path to sustainable modernization, yet there are certain limitations. First, unobservable macro-level shocks—such as the global financial crisis, geopolitical events, and the COVID-19 pandemic—may introduce heterogeneous effects that existing models cannot fully disentangle. Second, the measurement of underlying mechanisms is still insufficient, and micro-level transmission pathways of the integration between big data and sci-technology finance require further exploration. Finally, the analysis is conducted primarily at the provincial level, without examining the welfare and distributional effects of the policy on micro-level actors such as firms, households, and individuals. Future research could draw on more granular data and direct indicators to provide deeper insights into both the mechanisms and distributional outcomes.