How Does Digital Economy Drive Export Trade of Chinese Cities?—Based on the Perspective of Influence Mechanism Analysis and Threshold Effect

Abstract

Driven by the digital transformation of global value chains and the digital economy strategy outlined in China’s 14th Five-Year Plan, international trade competition is increasingly centered on digital technology ecosystems. This study addresses the gap in existing research that often overlooks the critical role of cities as key nodes in digital technology and global value chains, as well as the mechanisms through which the digital economy influences urban export trade. Using panel data from 273 prefecture-level Chinese cities between 2006 and 2022, we apply fixed effects, mediation, and multiple threshold regression models to analyze the mechanisms and nonlinear effects of the digital economy on urban export performance. Results show that the digital economy significantly facilitates urban export trade, with its marginal impact moderated by regional development levels and factor endowment structures. Eastern and central cities exhibit stronger export-driving effects, benefiting from resource agglomeration. Technological innovation and human capital accumulation are identified as the main transmission channels through which the digital economy promotes export upgrading. Additionally, the relationship between digital economy development and export trade demonstrates significant nonlinear characteristics across different regional economic development stages. The study emphasizes bridging the regional digital divide and enhancing technological innovation and human capital efficiency to drive digital transformation and boost urban export vitality.

1. Introduction

Against the background of the global digital economy wave profoundly reconstructing the international trading system, China has established the digital economy as the core strategic direction of the 14th Five-Year Plan. Data from the China Academy of Information and Communications Technology show that China’s digital economy scaled to CNY 53.9 trillion in 2023, representing 42.8% of GDP, and it has maintained the second global rank for five successive years [1]. This major strategic positioning highlights the significant role of the digital economy as the engine of new productivity, which is accelerating the change of the traditional trade paradigm through technology penetration and factor innovation [2]. Research shows that digital technology injects new kinetic energy into export trade by reducing transaction costs, breaking through geographical restrictions, and reshaping industrial chain division of labor [3]. In 2024, China’s export scale exceeded CNY 25 trillion for the first time, with a year-on-year increase of 7.1%, showing the strong vitality of China’s export trade. However, under the complex situation of the deep adjustment of the global economic structure, the sustainable development of China’s export trade is facing the dual challenges of internal momentum transformation and external environmental constraints. On the one hand, external factors such as the reconstruction of international trade rules and the intensification of external demand fluctuations constitute severe external pressures; on the other hand, internal structural problems such as insufficient core competitiveness of industries and the need to improve the position in the global value chain remain prominent. These factors restrict the sustainable development of regional export trade to varying degrees. Therefore, how to effectively stimulate the potential of the digital economy, break the long-standing structural constraints facing foreign trade in China, and drive its strategic transformation from scale expansion to quality upgrading has emerged as a critical issue requiring urgent resolution [4].

The current study primarily emphasizes the enhancement of the digital economy in export commerce and the augmentation of foreign trade competitiveness. Generally, the digital economy is regarded as having a beneficial effect on export commerce, altering its structural characteristics and substantially enhancing the quality of trade [5,6]. The research indicates that the technological spillover effect advances the global value chain to a higher echelon and enhances the worldwide competitive advantage of digital service commerce [7,8]. The digital economy facilitates the high-quality development of regional foreign trade by promoting industrial structure upgrading [9]. Particularly in service trade, the digital economy demonstrates significant regional linkage effects [10]. International research data show that in China, the digital economy significantly contributes to the export of services and high-tech products [11,12]. The emergence of the digital economy has markedly broadened the range of service commerce and the categories of exportable commodities and services [13]. For instance, in China and other nations, new business models such as big data analysis, cloud services, and cross-border electronic commerce have emerged as new engines for export growth [14]. In addition, the mechanism through which the digital economy affects export trade has become increasingly clear. Global trade can be generally stimulated by the widespread adoption of Internet technology [15], and the export competitiveness of enterprises can be significantly improved by the application of digital technology, which improves innovation efficiency, optimizes corporate governance structure, and augments human capital [16].

Its core mechanisms mainly manifest in two aspects: First, digital technologies (such as big data and cloud computing) significantly reduce costs in international trade, including those related to information search, contract execution, and cross-border payments; break through geographical restrictions; and empower small- and medium-sized enterprises to participate in global competition. Second, they reconstruct the division of labor in the global value chain, promoting enterprises to transform from a “cost-driven” model to a “data-driven” one and enhancing supply chain flexibility. On this basis, the digital economy further enhances the technical complexity of export products and the competitiveness of enterprises by promoting technological innovation and optimizing the structure of human capital (such as improving the digital skills of the labor force and accelerating knowledge spillover), ultimately driving the upgrading of urban export trade.

However, the existing research on the impact of the digital economy on export trade has generally confirmed its role in promoting exports at the national or provincial level [17,18]. However, it frequently disregards the unique attributes of cities as key nodes of digital technology diffusion and important carriers of global value chains. The city is the primary spatial entity in which the components of the digital economy congregate and industrial activities take place, and the impact mechanism and degree of digital economy in export trade may differ from those at the national or provincial level due to its internal structure, resource endowment, and connection with external networks. Simultaneously, in the extant research, the theoretical interpretation and empirical testing on the mechanism of action are still insufficient, particularly in terms of the absence of a systematic analysis of the internal transmission path of the digital economy that impacts urban export commerce. In addition, is the relationship between digital economy and urban export trade linear? Is there a threshold effect or nonlinear relationship caused by the difference of urban characteristics (such as economic development level, industrial structure, etc.)? These questions have not been fully answered in the existing literature.

Therefore, this paper concentrates on the city level, utilizing panel data from 273 prefecture-level cities in China spanning from 2006 to 2022, and it constructs a fixed effect model, an intermediary effect model, and a multiple threshold regression model to systematically elucidate the mechanisms and nonlinear impacts of the digital economy on export trade. It aims to systematically explore and answer the following core questions: How does the digital economy affect urban export trade through specific mechanisms, especially through the paths of technological innovation and human capital accumulation? Does its impact exhibit significant regional heterogeneity? Does the level of regional economic development exert a nonlinear regulatory effect on the relationship between the digital economy and export trade? This study aims to provide a solid empirical basis for China to formulate differentiated foreign trade policies by deepening the understanding of these issues.

Marginal contributions of the research include the following: (1) The research concentrates on the city level, elucidating the micro-mechanisms and heterogeneity of the digital economy’s impact on export trade with greater precision, addressing the limitations of current studies regarding spatial scale, and offering a novel perspective for comprehending the regional intricacies of the relationship between the digital economy and trade. (2) This study examines the influence of the digital economy on urban export trade by employing an intermediary effect model, and it thoroughly analyzes the transmission pathways of the digital economy through two critical dimensions: technological innovation and human capital, thereby offering a more nuanced theoretical framework and empirical evidence to elucidate the specific effects of the digital economy on urban export trade. (3) This paper systematically examines how the characteristics of urban economic development levels modulate the impact of the digital economy on export trade by employing a multiple threshold regression model, thereby uncovering potential nonlinear relationships or threshold effects that transcend the constraints of the conventional linear model. The growth of this method facilitates a more thorough and dynamic comprehension of the intricate mechanisms of the digital economy impacting urban export trade, which offers a significant empirical foundation for developing differentiated and precise urban foreign trade strategies.

2. Theoretical Analysis and Research Hypothesis

The advancement of the digital economy serves as a significant catalyst for the expansion of urban export commerce, primarily manifesting through two key dimensions: the reconfiguration of the global value chain’s labor division and the diminution of international trade transaction costs. On the one hand, digital technology, particularly the platform economy, big data analytics, and artificial intelligence, transcends the constraints of conventional geographical boundaries, compelling firms to transition from a traditional “cost-driven” export model to a more adaptable “data-driven” approach. The digital platform facilitates direct participation of small- and medium-sized enterprises (SMEs) in global competition, which allows production links to more precisely address specific market demands and encourages the value chain to extend along both ends of the “smile curve,” thereby optimizing the labor division within the global value chain [19].

On the other hand, digital technology has markedly diminished the expenses associated with information retrieval, contract execution, and cross-border payments in international trade by fundamentally altering the configuration of production factors, enhancing resource allocation efficiency, and lowering market transaction barriers [20]. From the perspective of transaction cost theory, the cost reduction enhances the operating efficiency of major firms while significantly lowering the barriers for small- and medium-sized enterprises (SMEs) to engage in foreign trade, which fosters the diversification of export participants [21]. Simultaneously, digital technology augments supply chain flexibility, allowing firms to react more swiftly to market fluctuations [22]. Although the operation of digital technology generates certain carbon footprints and associated costs, from the perspective of sustainable development, its precise matching of demand can reduce resource waste [23]. Moreover, the continuous advancement and application of green digital technology are gradually mitigating its environmental impacts [24].

The digital economy serves as the fundamental impetus for the expansion of scale and enhancement of efficiency in urban export trade by optimizing labor division within global value chains, minimizing transaction costs, empowering small- and medium-sized enterprises and fostering supply chain flexibility. Consequently, this paper anticipates that the city’s digital economy maturity correlates positively with the scale and efficiency of its export commerce. According to this rationale (Figure 1), we propose the following hypotheses to assess the fundamental influence of the digital economy on urban export trade:

H1. 

The development level of the digital economy has a significantly positive impact on urban export trade.

The promotion effect of the digital economy on urban export trade is spatially different. According to the global value chain theory, as the key node of digital technology diffusion and an important carrier of the global value chain, a city’s own development endowment and external environment will significantly regulate the exertion of digital economic effects. Specifically, there are significant differences in economic development level, industrial structure, factor endowment, and policy environment among different cities, which together shape the path and intensity of the digital economy, affecting export trade.

From a regional perspective, the eastern region relies on the digital industrial clusters (such as Hangzhou e-commerce and Suzhou intelligent manufacturing) and policy dividends of the Pilot Free Trade Zone (such as Shanghai Port Data Cross-border Flow Pilot) to form a dual advantage of “technology-system,” and its digital economy far exceeds that of the central and western regions in promoting exports [25]. Previous studies indicated that the export promotion impact of the digital economy in the eastern region can be 2.2 times greater than that in the middle and western regions, demonstrating a pronounced Matthew effect [26]. However, other research shows that the enhancement of export quality through the digital economy is more pronounced in the central and western regions, attributed to the late-mover advantage and favorable policies. Conversely, the eastern regions may experience a “siphon effect” that stifles innovation due to an overconcentration of resources [27]. In addition, the urban administrative level also constitutes another important dimension: the heterogeneity effect. Due to their distinct administrative resource endowments and development priorities, municipalities under central government jurisdiction and first-tier cities generally demonstrate higher digital economy penetration and operational efficiency. As a result, the mechanisms through which the digital economy influences export trade, along with its specific impacts, show notable disparities compared to other urban areas. In light of the preceding analysis, this paper proposes:

H2. 

There is significant regional heterogeneity in the promotion of digital economy development to urban export trade.

The role of the digital economy in promoting urban export trade is not simply a superposition of traditional factors but is realized by reshaping its core driving paths. Among them, the ability to stimulate and utilize technological innovation is the primary engine for the digital economy to empower exports. The new economic growth theory emphasizes that technological progress and human capital are the endogenous drivers of sustained economic growth. Studies have shown that government expenditures on science and technology and enterprise R&D investments can form a positive cycle mechanism of “R&D–patent–export” through the commercial application of digital technologies. Different from traditional innovation models, the digital economy has greatly improved R&D efficiency and innovation accuracy through factor-driven data, algorithm optimization, and platform-based collaboration. In addition, the mechanism not only mitigates the strain of diminishing marginal returns from conventional production components but also efficiently transforms national-level innovation accomplishments into competitive advantages in international trade through digital collaboration inside global value chains. At present, China’s export trade sustainability trajectory has transitioned from reliance on resource endowment to a dynamic rivalry framework focused on technical innovation. On the one hand, technological input creates structural driving forces by increasing the technical complexity of export products, and the tendency is especially evident in high-tech sectors [28]. Conversely, micro-level company technical innovation markedly enhances total factor productivity, and it directly reinforces the worldwide competitive advantages of export commodities [29]. In addition, the in-depth application of digital technologies (such as industrial internet and artificial intelligence) has accelerated the full-process digital transformation of enterprises, improved the efficiency of innovation output, and thereby enhanced export competitiveness [30,31].

H3A. 

The digital economy can promote urban export trade by stimulating technological innovation.

The accumulation and upgrading of human capital are closely coupled with and mutually supportive of the path of technological innovation. The robust advancement of the digital economy has generated new knowledge-intensive service trades, including cloud computing and digital content export, and it emphasizes the pivotal role of human capital in these sectors. Similarly, the new economic growth theory points out that human capital is the key to technology absorption and innovation. The popularization of digital technologies not only requires the labor force to upgrade their professional capabilities through digital skills training but also gives rise to an urgent demand for high-quality talents with data analysis and cross-border integration capabilities. Enterprises, in turn, enhance their innovation capabilities by introducing high-quality talents. The digital economy can enhance Guangxi’s foreign trade competitiveness by leveraging human capital [32]. From the perspective of export sophistication, digital infrastructure can indirectly enhance the export sophistication of digital services via the human capital pathway [33]. Moreover, with the support of digital platforms and virtual communities, the talent agglomeration effect has been amplified, and the speed of knowledge spillover has accelerated, promoting an average annual growth of 2.1% in the complexity of regional export technologies [34]. Furthermore, high-quality human capital is a prerequisite for effectively applying digital technologies such as big data and artificial intelligence to achieve disruptive technological innovation, while cutting-edge digital technologies, such as intelligent training systems and collaborative R&D platforms, can empower talents and accelerate their knowledge updating and skill improvement, forming a virtuous cycle of “talent introduction, cultivation, and innovation—innovation empowering talents.” Based on Figure 2, this paper proposes the hypothesis H3b:

H3B. 

The digital economy can promote urban export trade by accelerating human capital accumulation.

The influence of the digital economy on export trade is marked by differentiation and dynamic nonlinearity as urban economic growth progresses. In the early phase of economic development, the preliminary implementation of the digital economy predominantly exerts a significant influence on export commerce by bridging the deficiencies of traditional transactions and mitigating information asymmetry. The research indicates that the combined influence of policy support and technological diffusion may result in an “increasing marginal effect” of the digital economy on export trade at this stage [35]. The influence of the digital economy on export trade in economically deprived regions exhibits a threshold effect. Upon overcoming the bottleneck, its advancement will increase markedly [36]. The enhancement of digital infrastructure and the integration of technology into the upper echelons of the industrial chain, along with the escalating complexity of resource integration and the saturation of local markets, result in a deceleration of marginal contribution. The findings of empirical research vary: data from prefecture-level cities indicate that the influence of the digital economy on exports follows a nonlinear pattern of “initial enhancement followed by subsequent decline” [37], whereas provincial studies reveal a continuous “marginal increase” in its driving effect [38]. The firm heterogeneity trade theory provides a micro perspective for understanding this nonlinear relationship, which holds that productivity differences among firms are the key determinant of their export behavior. As digital technology penetration nears saturation, the digital transformation of enterprises devolves into homogenized competition, resulting in overlapping data monopolies and conflicts in cross-border regulations, which impedes the marginal contribution of the digital economy to exports. Relevant research indicates that the advancement of the digital economy in foreign commerce exhibits an inverted U-shaped trajectory characterized by initial enhancement followed by subsequent decline [39], and that technological spillovers demonstrate regional asymmetry [37]. Based on the above analysis, it is proposed that:

H4. 

The relationship between the digital economy and export trade follows a threshold effect contingent on urban economic development levels.

3. Research Design

3.1. Variable Selection

3.1.1. Explained Variables

The export trade volume (LnED) is measured by the natural logarithm of the export value (millions of dollars) of prefecture-level cities. The increase in the value of this indicator reflects the increased activity of export trade. The basic data are all quoted from the statistical database of the China Economic Network.

3.1.2. Core Explanatory Variables

To objectively and scientifically assess the development level of China’s urban digital economy, the entropy method is employed to compute the digital economy comprehensive index (DIG) by impartially weighting multidimensional sub-indicators. In the field of multi-index comprehensive evaluation, selecting an appropriate weighting method is crucial. Compared with subjective weighting methods such as the analytic hierarchy process (AHP) or expert scoring method, the entropy method has significant advantages. Based on the theory of information entropy, this method determines weight coefficients by quantifying the degree of dispersion of each indicator’s data. Its objective weighting mechanism can effectively overcome the arbitrariness and preference that may be brought by subjective methods, thus ensuring the scientificity and robustness of evaluation results [9,40,41,42]. The applicability of the entropy method has been widely verified and applied in recent studies evaluating complex socioeconomic concepts (such as the digital economy and green development level). Its core advantage lies in its ability to perform “data-driven” weight allocation based on the inherent variation characteristics of indicator data, which is particularly suitable for comprehensive evaluation systems with rich connotations, multiple dimensions, and complex relationships between dimensions, such as the digital economy. Based on prior research [43], the advancement of the digital economy is assessed through three dimensions: digital infrastructure development, digital industry growth, and the digital economy’s developmental environment, which encompasses six secondary indicators. Refer to

Table 1. Index system of digital economy development level.

First-Class Index	Secondary Index	Index Calculation	Indicator Attribute
Digital infrastructure construction	Linear density of long-distance optical cable	Long-distance optical cable line length/administrative area	Forward
	Per capita Internet broadband access port	Internet broadband access port/total population	Forward
Digital industry development	Proportion of employees in information transmission, computer services, and software industry	Proportion of employees in urban units of information transmission, computer services, and software industry	Forward
	Per capita telecom business income	Total revenue of telecommunications business/total population	Forward
Development environment of digital economy	Mobile phone penetration rate	Number of mobile phone users/total population	Forward
	Internet penetration rate	Number of Internet broadband access users/total population	Forward

(Source: Designed by the authors).

for a detailed description of the indicators; all original data are sourced from the National Bureau of Statistics. Table 1’s index system utilizes annual data from 273 prefecture-level cities in China spanning from 2006 to 2022, with the original data sourced from the official statistics of the National Bureau of Statistics and the White Paper for the Development of Digital Economy in China.

3.1.3. Mediation Variables

The level of science and technology (sci) is used as an intermediary variable. This paper argues that advancements in science and technology can foster technical innovation, refine manufacturing techniques, and augment product value, which could bolster export competitiveness. The enhancement of scientific and technological standards can facilitate information development, which is intricately linked to the growth of the digital economy and establishes an indirect pathway for the digital economy to influence exports. In this paper, science and technology expenditure/general government financial expenditure is selected to measure the level of science and technology [44], and the data comes from the “China Economic Network Database”.

Human capital level (hum) is an intermediary variable. Human capital is the key intermediary for the digital economy to empower export trade. Its accumulation indirectly strengthens export competitiveness by improving labor skills, optimizing resource allocation efficiency, and enhancing innovation ability. High-quality human capital enables enterprises to absorb digital technology more effectively, improve management, and develop high-value-added products, thus forming differentiated advantages in the international market. At the same time, human capital also accelerates the process of digital transformation, and it is easier for high-quality laborers to master digital tools, so as to enhance the synergy between enterprise informationization and the supply chain, which constitutes another key intermediary path for the digital economy to affect exports. This paper selects the number of students in ordinary colleges/the total population at the end of the year to measure the level of human capital [45].

3.1.4. Control Variables

The level of economic development (lnpgdp) is measured in the logarithmic form of per capita GDP. The degree of economic development not only indicates the efficiency of regional economic output and the well-being of people but also signifies its developmental stage. Economically developed areas usually have higher total factor productivity, a better industrial supporting system, and a more optimized institutional environment. These factors form basic support for export trade by enhancing the international competitiveness of enterprises and reducing transaction costs. This variable aims to control the inherent driving effect of regional economic bases on trade expansion [46].

Population size (lnpop): Measured by the logarithm of the total population at the end of the year. Population size is the key factor affecting the scale and structure of economic activities. Logarithmic processing is helpful to alleviate the heteroscedasticity of data and capture the scale effect. A huge population may bring broader market demand, but it may also affect the pattern of resource distribution and the economic growth model. This variable controls the general influence of population factors on the regional economy [47].

Education expenditure ratio (edu) is defined as the proportion of government financial education expenditure to total expenditure. This index is an important dimension to measure the government’s investment in education and policy orientation. Investment in education has a far-reaching impact on long-term economic growth by improving the skill level of the labor force, promoting the accumulation of human capital, and then improving productivity and innovation ability. The purpose of including this variable is to control the long-term shaping effect of government investment in education on regional economic and trade potential [45].

Market size (mar): It is measured by the ratio of total retail sales of social consumer goods to GDP, reflecting regional consumption capacity and market potential. Areas with larger market scale usually have stronger consumption driving force, reflecting higher economic vitality and domestic demand support. Under the background that consumption is increasingly becoming the main engine of economic growth, this variable controls the possible substitution or complementary relationship between internal market demand and export trade.

Infrastructure level (inf): Assessed by the ratio of fixed asset investment to GDP, it indicates the magnitude and intensity of regional infrastructure development. The excellence of infrastructure is a crucial foundation for economic development, which directly influences production efficiency, transaction costs and overall economic competitiveness. To conclude, this variable seeks to regulate the essential influence of infrastructure conditions on regional economic efficiency and trade expenses.

Foreign investment level (inv): Quantified by the ratio of actual foreign direct investment to GDP, it indicates the capacity to attract external capital and the extent of economic openness in the region. Foreign investment frequently brings technology, managerial expertise, and additional elements, resulting in spillover effects on local firms and fostering economic progress. This variable is incorporated to regulate the possible influence of foreign capital inflow on regional economic structure, technological advancement, and export capability [48].

Table 2. Definition and design of variables in benchmark model.

Variable Type	Variables	Variable Definition	Measuring Method
Explained variable	InED	Export trade volume	Export trade volume of prefecture-level cities in China (millions of US dollars)
Core explanatory variable	DIG	Development level of digital economy	Entropy method
Control variable	lnpgdp	Level of economic development	Logarithm of GDP per capita
	lnpop	Population size	Logarithm of the total population at the end of the year
	edu	Proportion of education level	Education expenditure/general government expenditure
	mar	Market size	Total retail sales of social consumption/regional GDP
	inf	Infrastructure construction	Investment in fixed assets/regional GDP
	inv	Foreign investment level	Actual utilization of foreign capital/regional GDP
Mediating variable	sci	Scientific and technological level	Science and technology expenditure/general government financial expenditure
	hum	Human capital level	Number of students in general college/total population at the end of the year
Threshold variable	lnpgdp	Level of economic development	Logarithm of GDP per capital

(Source: Designed by the authors).

describes the construction of the specific indicator system and the data sources.

3.2. Model Setting

3.2.1. Baseline Regression Model

In order to empirically test the mechanism of digital economy development on export trade, based on the previous analysis and drawing on the research results [38], this paper constructs a benchmark econometric model, setting export trade volume (LnED) as the dependent variable and digital economy development index (DIG) as the core independent variable, as shown in (1):

$$\ln {\mathrm{ED}}_{it}={\alpha }_0+{\alpha }_1{\mathrm{DIG}}_{\mathrm{it}}+{\gamma }_1{\mathrm{lnpgdp}}_{\mathrm{it}}+{\gamma }_2{\mathrm{lnpeo}}_{\mathrm{it}}+{\gamma }_3{\mathrm{edu}}_{\mathrm{it}}+{\gamma }_4{\mathrm{mar}}_{\mathrm{it}}+{\gamma }_5{\mathrm{lnf}}_{\mathrm{it}}+{\gamma }_6{\mathrm{lnv}}_{\mathrm{it}}+{\eta }_{\mathrm{i}}+{\lambda }_{\mathrm{t}}+{\epsilon }_{\mathrm{it}}.$$

(1)

Among these terms, $i$, $t$ represent different prefecture-level cities and different years; ${\mathrm{DIG}}_{\mathrm{it}}$ represents the digital economic index calculated by entropy method; ${\eta }_{\mathrm{i}}$ is an individual fixed effect; ${\lambda }_{\mathrm{t}}$ is a time fixed effect; and ${\epsilon }_{\mathrm{it}}$ is a random disturbance term.

The baseline regression model employs a panel data fixed effects approach, which is particularly suitable for this research as it effectively controls for unobserved time-invariant characteristics across cities and common time shocks affecting all regions. This methodological choice ensures more reliable estimates of the digital economy’s impact on export trade by mitigating potential omitted variable bias.

3.2.2. Mediating Effect Model

In order to verify the transmission path of digital economy affecting export trade, this paper draws lessons from the intermediary effect analysis framework of former scholars and constructs a model to test the intermediary role between technological progress (sci) and human capital accumulation (hum) [49]. The specific model building process is presented by Formulas (2) to (3):

$${\mathrm{Mediator}}_{\mathrm{it}}={\beta }_0+{\beta }_1\cdot {\mathrm{DIG}}_{\mathrm{it}}+\psi Contro{l}_{\mathrm{it}}+{\eta }_{\mathrm{i}}+{\lambda }_{\mathrm{t}}+{\epsilon }_{\mathrm{it}}$$

(2)

$$\ln {\mathrm{ED}}_{\mathrm{it}}={\theta }_0+{\theta }_1\cdot {\mathrm{DIG}}_{\mathrm{it}}+{\theta }_2{\mathrm{Mediator}}_{\mathrm{it}}+\Phi Contro{l}_{\mathrm{it}}+{\eta }_{\mathrm{i}}+{\lambda }_{\mathrm{t}}+{\epsilon }_{\mathrm{it}}$$

(3)

Among these, ${\mathrm{Mediator}}_{\mathrm{it}}$ is an intermediary variable, representing the level of science and technology (sci) and the level of human capital (hum); and $Contro{l}_{\mathrm{it}}$ is a series of control variables.

The mediating effect model is employed to systematically identify and quantify the transmission channels through which the digital economy influences export trade. This approach is particularly valuable for our research as it allows us to disentangle the direct effects of digital economy development from its indirect effects operating through technological progress and human capital accumulation, providing a more comprehensive understanding of the underlying mechanisms.

3.2.3. Threshold Regression Model

Threshold regression, as a nonlinear analysis method, aims to capture the asymmetric influence relationship among variables. The fundamental aspect of empirical analysis is to pinpoint structural mutation points—when threshold variables surpass a particular critical value, the interaction mechanism between the core explanatory factors and the dependent variables undergoes considerable alteration. This study, under the benchmark model framework, utilizes insights from previous academics’ research design, incorporates the degree of economic development (lnpgdp) as the threshold variable, and formulates a nonlinear econometric model as depicted in Equation (4) [36]. This model empirically examines the heterogeneous effects of the digital economy on export commerce across several stages of economic development, thereby elucidating the features of trade growth driven by digital factors at different times.

$$\ln {\mathrm{ED}}_{it}={\alpha }_0+{\phi }_1{\mathrm{DIG}}_{\mathrm{it}}\cdot I(lnpgd{p}_{\mathrm{it}}\le \theta )+{\phi }_2{\mathrm{DIG}}_{\mathrm{it}}\cdot I(lnpgd{p}_{\mathrm{it}}>\theta )+\psi Contro{l}_{\mathrm{it}}+{\eta }_{\mathrm{i}}+{\lambda }_{\mathrm{t}}+{\epsilon }_{\mathrm{it}}$$

(4)

In Formula (4), the threshold variable is $lnpgd{p}_{\mathrm{it}}$; the threshold is $\theta$, represented by the threshold value $I(\theta )=I(lnpgd{p}_{\mathrm{it}}\le \theta )$ is a set dummy variable; and $I(\theta )$ is a Bernoulli variable.

The specific formula of the double threshold regression model is:

$$\begin{array}{l}\ln {\mathrm{ED}}_{it}={\alpha }_0+{\phi }_1{\mathrm{DIG}}_{\mathrm{it}}\cdot I(lnpgd{p}_{\mathrm{it}}\le {\theta }_1)+{\phi }_2{\mathrm{DIG}}_{\mathrm{it}}\cdot I({\theta }_1<lnpgd{p}_{\mathrm{it}}\le {\theta }_2)\\ +{\phi }_3{\mathrm{DIG}}_{\mathrm{it}}\cdot I({\theta }_2<lnpgd{p}_{\mathrm{it}}\le {\theta }_3)+{\phi }_4{\mathrm{DIG}}_{\mathrm{it}}\cdot I(lnpgd{p}_{\mathrm{it}}>{\theta }_3)+\psi Contro{l}_{\mathrm{it}}+{\eta }_{\mathrm{i}}+{\lambda }_{\mathrm{t}}+{\epsilon }_{\mathrm{it}}\end{array}$$

The threshold regression methodology is particularly appropriate for this study as it allows us to identify potential nonlinear relationships and structural breakpoints in the impact of digital economy on export trade. By incorporating economic development level as a threshold variable, this approach enables us to capture how the relationship between digital economy and export trade may change across different development stages, providing more nuanced policy implications than traditional linear models.

3.3. Data Sources

This study analyzes panel data from 273 prefecture-level cities in China from 2006 to 2022. The sample was selected based on data availability, excluding Tibet, Hong Kong, Macao, and Taiwan. All economic indicators were adjusted for inflation using 2006 as the base year to enhance data comparability. The primary data sources for this study include the “China Urban Statistics Yearbook,” the “China Urban Construction Statistics Yearbook” (https://data.cnki.net/; accessed on 1 October 2023), and the Easy Professional Superior (EPS) database (http://olap.epsnet.com.cn; accessed on 1 October 2023). Additionally, data were extracted from statistical yearbooks of specific provincial-level administrative regions and prefecture-level cities. The evaluation system for the development level of the digital economy was constructed based on the China Economic and Social Development Statistics Database (https://www.nbsti.net/CSYDMirror/Yearbook; accessed on 1 September 2023), integrating indicators related to digital infrastructure, industrial digitization, and digital technology innovation. For cases where data were missing for individual cities, linear interpolation was used to fill in the missing values. All data underwent rigorous cleaning and validation procedures to ensure quality and consistency throughout the study period.

4. Analysis of Empirical Research Results

4.1. Regression Analysis of Baseline Results

4.1.1. Descriptive Statistics

The findings of the descriptive statistical analysis indicate that the standard deviation of the digital economy index exceeds the mean value, as illustrated in

Table 3. Descriptive statistics of the major variables.

Variables	Count	Mean	Sd	Min	Max
LnED	4448	6.779	2.095	0.113	12.862
DIG	4448	0.028	0.032	0.002	0.719
lnpgdp	4448	10.500	0.722	4.595	13.056
lnpop	4448	5.885	0.696	2.868	8.136
edu	4448	0.180	0.042	0.010	0.377
mar	4448	0.368	0.106	0.000	1.013
inf	4448	0.782	0.366	0.000	2.772
inv	4448	0.018	0.019	0.000	0.229

(Source: Authors’ calculation using Stata 16.0).

, highlighting significant disparities in the development levels of the digital economy across various cities.

4.1.2. Multicollinearity Test

To ensure the robustness of the model estimation results, this study uses the variance inflation factor to diagnose the multicollinearity of the core explanatory variables and control variables.

Table 4. Variance inflation factors for independent variables.

Variable	VIF
lnpgdp	1.440
lnpop	1.350
DIG	1.330
edu	1.270
mar	1.210
inv	1.100
inf	1.050
MeanVIF	1.250

(Source: Authors’ calculation using Stata 16.0).

shows that the VIF values of all variables are far lower than the critical value of 10, and the maximum value is only 1.44. This indicates that there is no significant multicollinearity among the explanatory variables, the regression coefficients of the model are not disturbed by collinearity, and the estimation results are reliable and valid.

4.1.3. Benchmark Regression Results

Results of the model setting test shown in

Table 5. F test and Hausman test.

	F Test	Hausman Test
F-value	66.79
Chi-square test value		327.2
p-value	0	0

(Source: Authors’ calculation using Stata 16.0).

show that both the F test and Hausman test are significant at the 1% level, which rejects the original hypothesis of a random effect. Therefore, this study adopts the fixed effect model for benchmark analysis.

The empirical test results for research hypothesis H1 (see

Table 6. Regression and robustness check results.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)
Variables	LnED	LnED	LnED	LnED	LnED	LnED	LnED
DIG	26.1701 ***	11.068 ***	10.7065 ***	10.7348 ***	9.8381 ***	9.4442 ***	8.7404 ***
	(5.9460)	(4.1667)	(4.1753)	(4.2305)	(4.0291)	(3.9758)	(3.7442)
lnpgdp		1.9479 ***	2.1571 ***	2.1811 ***	2.2968 ***	2.2426 ***	2.0282 ***
		(9.7244)	(15.7399)	(15.2788)	(16.8688)	(16.4121)	(14.6606)
lnpop			1.4899 ***	1.4632 ***	1.2976 ***	1.2849 ***	1.1842 ***
			(16.8058)	(15.7714)	(13.6091)	(13.2555)	(12.9031)
edu				1.4886	1.8201	1.5891	2.9091 **
				(1.0052)	(1.3171)	(1.1644)	(2.1986)
mar					3.0233 ***	2.9564 ***	2.7470 ***
					(4.2136)	(4.1003)	(4.0048)
inf						−0.3127 **	−0.4819 ***
						(−2.1545)	(−3.4142)
inv							17.2306 ***
							(6.2938)
_cons	5.4780 ***	−12.987 ***	−23.6786 ***	−24.0186 ***	−25.1921 ***	−24.3800 ***	−22.2020 ***
	(42.2976)	(−6.9111)	(−17.9920)	(−17.1090)	(−18.0105)	(−16.7972)	(−15.2620)
ID Fixed	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year Fixed	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	4448	4448	4448	4448	4448	4448	4448
r2	0.1860	0.4206	0.6629	0.6636	0.6802	0.6822	0.7023

Note: T-statistics are in parentheses. ** denotes p < 0.05, *** denotes p < 0.01. (Source: Authors’ calculation using Stata 16.0).

for details) indicate that the digital economy exerts a consistent and considerable beneficial influence on urban export trade. The regression coefficient of the digital economy index exhibits a declining trend, decreasing from 26.1701 in the basic model (1) when control variables are progressively incorporated, although it maintains statistical significance at the 1% level across all model configurations. The findings from the comprehensive model (7) indicate that each standard unit rise in the digital economy index can enhance the export scale of cities by 8.7404%. This outcome robustly corroborates the research hypothesis H1 and affirms that the advancement of the digital economy can substantially enhance export trade.

The underlying mechanism of this driving influence may reside in the following factors: Firstly, the inventive utilization of digital technology, including cloud computing, big data, and artificial intelligence, enhances production efficiency and equips organizations with novel tools to optimize operations and save costs. Secondly, the digital economy significantly transforms the supply chain system, enhances the efficiency of order processing and transportation, and diminishes transaction costs via digital management, e-commerce, and cross-border payment. Moreover, the proliferation of the digital economy augments the market competitiveness and brand influence of firms, while simultaneously elevating brand awareness through digital marketing strategies. Ultimately, digital platforms (including e-commerce platforms and online payment systems) offer simple avenues for firms to “go global,” significantly diminishing barriers to entrance in the foreign market and facilitating the expansion of export scale. To conclude, these systems collaboratively form the internal logic of the digital economy, enhancing export trade.

The regression results of control variables reveal the influence of other key factors on urban export trade:

Economic development level (lnpgdp): The influence coefficient of economic development level on export trade development is positive. When per capita GDP increases by 1%, export trade will increase by about 2.03%. With stronger technology absorption capacity, more developed industrial chain supporting facilities (such as the Yangtze River Delta and Pearl River Delta industrial clusters), and perfect infrastructure, economically developed regions can transform technology into products more efficiently, reduce the marginal cost of export enterprises, and thus enhance their international competitiveness [46].

Population size (lnpop): The coefficient is significantly positive. For every 1% increase in population size, exports increase by about 1.18%. Large population size provides sufficient human resources, which helps to reduce labor costs and is attractive to labor-intensive industries. At the same time, its “local market effect” can reduce export prices through economies of scale and enhance the overall competitiveness of local enterprises.

Proportion of education expenditure (edu): The coefficient is significantly positive, and exports increase by about 2.91% for every unit increase in the proportion of education expenditure. The increase in investment in education has improved the skill level of the labor force and promoted the export of technology-intensive products (such as software services and precision instruments). Technology transformation between universities and scientific research institutions (such as patent authorization, industry–university research cooperation) also promotes enterprise innovation and enhances the competitiveness of export products.

The market scale (mar) coefficient shows a very prominent positive relationship. For every unit increase in the market scale, the export growth rate is about 2.75%. Under a prosperous domestic demand market, enterprises will meet the domestic demand by expanding production capacity and reduce the unit cost by virtue of scale effect, so as to improve the export price competitiveness. Taking the rapid development of household appliances and the automobile industry as an example, this enables export enterprises to meet the international market demand with lower cost and higher efficiency. In addition, the domestic market gives local brands room for trial and error. When local brands accumulate to a certain scale in the market, they can expand overseas and form global brand influence.

The coefficient of infrastructure (inf) is negative, indicating that excessive infrastructure investment may have a resource crowding effect on the export sector, leading to an imbalance in resource allocation. The underlying mechanism lies in the insufficient synergy efficiency of the infrastructure network. For example, although the Zhengzhou Airport Economy Zone has complete hardware facilities, it once suffered from “data silos” and process barriers in the air–rail transit link, resulting in an average cargo connection time of up to 12 h, which seriously weakened its regional advantage as an inland opening-up highland. This is precisely the specific manifestation of the “resource crowding effect” at the micro-operational level—physical assets are not utilized efficiently, and instead, additional “time costs” and “transaction costs” are generated due to poor processes. To address this, the local government has built an “air–rail intermodal digital platform,” opened up data interfaces, realized “single-order” customs clearance and full-process visual tracking, and increased cargo turnover efficiency by more than 60%. This case convincingly proves that the key to solving the “resource crowding effect” is to improve the synergy efficiency and utilization rate of the logistics network through digital means, and truly transform the “stock advantage” of infrastructure into the “flow advantage” of exports.

The coefficient of foreign investment (inv) shows a positive trend. When foreign investment increases by one unit, exports will increase by about 1.87%. Foreign businessmen bring advanced technology and management experience, which improves the efficiency of the local supply chain, shortens production cycles, and reduces costs. The global sales network of foreign-funded enterprises also helps local products enter the international market. In addition, foreign investment has eased the financing constraints of local enterprises, expanded export production capacity, and promoted the growth of export trade [48].

4.2. Heterogeneity Analysis

Based on benchmark regression’s confirmation of the direct driving effect of the digital economy on export commerce, this study empirically analyzes Hypothesis H2 and investigates the effect’s regional heterogeneity. Considering the substantial disparities in the development levels of China’s regional digital economy and foreign trade foundations, this study perpetuates the grouping logic established in the current situation analysis and devises a multi-dimensional classification framework. Meanwhile, we systematically examine the variances in the marginal contributions of the digital economy to export trade across cities with varying regional development gradients, which aims to precisely identify the regional distribution characteristics of digital dividends.

4.2.1. Regional Heterogeneity: Significant Differentiation Between Eastern and Central and Western Regions

The regression results by sample (refer to

Table 7. The results of sub-sample regression in different dimensions of each city.

	(1)	(2)
Variables	Eastern Regions	Central and Western Areas
DIG	9.388 ***	2.227
	(2.147)	(2.125)
lnpgdp	1.843 ***	1.488 ***
	(0.151)	(0.168)
lnpop	0.920 ***	1.220 ***
	(0.146)	(0.103)
edu	4.976 ***	−4.935 ***
	(1.447)	(1.433)
mar	1.325 **	1.502
	(0.660)	(0.923)
inf	−0.872 ***	0.0185
	(0.277)	(0.163)
inv	13.27 ***	17.27 ***
	(2.574)	(3.529)
Constant	−17.91 ***	−15.99 ***
	(1.614)	(1.872)
Observations	1600	2848
R-squared	0.792	0.586

Note: T-statistics are in parentheses. ** denotes p < 0.05, *** denotes p < 0.01. (Source: Authors’ calculation using Stata 16.0).

, columns 1–2) indicate that the impact of the digital economy on export commerce varies significantly between the eastern and central and western areas. The eastern region exhibits a robust and significant promotional effect, whereas the middle and western regions lack statistical significance [26]. This disparity is attributable to the “engine effect” generated by the eastern core urban agglomeration and the “digital divide” typically encountered by the middle and western regions. The eastern core urban agglomerations (Beijing–Tianjin–Hebei, Yangtze River Delta, and Pearl River Delta) have generated a substantial engine effect due to their twin roles as national economic hubs and worldwide digital innovation centers. Node cities like Beijing, Shanghai, and Shenzhen depend on superior digital infrastructure and intricately integrated digital trade models, including cross-border e-commerce platforms, digital service trade, and advanced manufacturing enhanced by the industrial Internet, to establish robust global connectivity, positioning the digital economy as the primary catalyst for export expansion. Shenzhen exemplifies electronic information, Hangzhou represents e-commerce ecology, and Shanghai is characterized by financial technology and high-end industry. Take Hangzhou as an example: As a representative city in the innovation-driven period, Hangzhou relies on the advantages of the entire digital economy industry chain to form an export growth model of “platform empowerment—technology penetration—ecological reconstruction.” In 2024, the added value of Hangzhou’s core digital economy industries exceeded CNY 630 billion, accounting for 28.8% of GDP, and the transaction volume of the cross-border e-commerce comprehensive pilot zone accounted for 21% of the national total. Alibaba International Station has driven 72% of small, medium, and micro enterprises to achieve exports through the “platform going global” model. In addition, 72 unicorn enterprises in the Binjiang High-Tech Zone have built a digital ecosystem, and Dahua Technology’s AI quality inspection system has reduced the defective rate of export products to 0.3%, driving the annual growth of intelligent security equipment exports by 32%.

Conversely, the overall impact in the central and western regions is negligible, which underscores the limitation of the “digital divide” [25]. The particulars of the performance are as follows: (1) digital infrastructure and industrial clustering are underdeveloped, hindering technological spillovers; (2) policy and resource distribution remain skewed towards conventional industries, with inadequate digital skill reserves; and (3) the forms of foreign capital consumption vary. Foreign investment in the central and western areas predominantly focuses on low-end processing, exhibiting minimal synergy with digital technology, whereas foreign investment in the eastern regions is more focused on high-tech sectors. While certain core cities in the central and western regions, such as Zhengzhou, Chengdu, and Wuhan, are making notable advancements, a considerable disparity exists between the numerous peripheral cities and those in the eastern regions regarding infrastructure, technological integration, enterprise digitalization capacity, talent availability, and international connectivity. Consequently, the potential influence of core cities remains underutilized, leading to a minimal overall empowerment effect. Take Zhengzhou and Xi’an as examples: As inland cities in the efficiency-driven phase, Zhengzhou leverages the cross-border e-commerce “1210 customs model” and digital clearance systems at its aviation port zone to boost exports of labor-intensive goods like apparel and home furnishings. In 2023, Zhengzhou’s cross-border e-commerce exports surged by 99.7% year-on-year, where digital technology played a fundamental role in reducing international market entry costs for small- and medium-sized enterprises. Meanwhile, Xi’an promotes agricultural and traditional industrial product exports through its “Digital Silk Road” node infrastructure (e.g., smart logistics platforms at Xi’an International Inland Port), though digital adoption remains concentrated in marketing applications (such as cross-border livestreaming), with supply chain digitization still lagging behind.

4.2.2. Heterogeneity at City Level: Central Cities Lead and Peripheral Cities Benefit

The regression analysis of sub-samples (refer to

Table 8. The results of sub-sample regression in different dimensions of each city.

	(3)	(4)
Variables	Central Cities	Peripheral Cities
DIG	8.507 **	8.320 ***
	(2.926)	(3.111)
lnpgdp	1.198 **	2.051 ***
	(0.507)	(0.143)
lnpop	1.097 ***	1.191 ***
	(0.305)	(0.106)
edu	−1.172	2.850 **
	(6.128)	(1.388)
mar	−3.661 **	2.895 ***
	(1.590)	(0.706)
inf	−1.193	−0.468 ***
	(0.754)	(0.145)
inv	8.635	17.58 ***
	(4.938)	(2.836)
Constant	−9.233	−22.50 ***
	(6.637)	(1.535)
Observations	192	4256
R-squared	0.797	0.657

Note: T-statistics are in parentheses.** denotes p < 0.05, *** denotes p < 0.01. (Source: Authors’ calculation using Stata 16.0).

, columns 3–4) demonstrates that the advancement of the digital economy positively influences export trade in both central and peripheral cities, suggesting that the capacity of the digital economy to enhance exports is prevalent in both urban centers and surrounding areas. The goodness-of-fit for the central city model is markedly superior to that of the peripheral cities, with a slightly bigger coefficient, affirming the center city’s pivotal role as the main engine of digital trade and substantiating hypothesis H2.

The leading and hub function of the central city is the key to its high goodness of fit. They are not only the giants of their own digital trade, but also the source and distribution hub of national and even global digital trade rules, technology, capital, and information. The digital economy is deeply embedded in their export competitiveness. For example, megacities such as Beijing, Shanghai, and Shenzhen, as well as regional central cities such as Tianjin, Chongqing, Chengdu, and Wuhan, all rely on the national strategy to actively build regional digital hubs and significantly drive local exports.

The notable advancement of peripheral cities signifies that the benefits of the digital economy are extending to the hinterland, primarily due to the spillover effects of technology, platforms, capital, and orders from core cities, alongside localized digital initiatives rooted in resource endowment, such as rural e-commerce and industrial internet in specialized towns. The explanatory power of the peripheral city model is limited, indicating that traditional elements, institutional environment, market scale, and industrial support facilities remain fundamental factors influencing its exports, while the digital economy plays a significant yet potentially secondary role.

To conclude, the influence of the digital economy on export trade exhibits significant regional and urban heterogeneity, particularly in eastern and central cities. Conversely, central, western, and peripheral cities derive benefits, albeit constrained by foundational conditions and transmission mechanisms, necessitating enhancements in both the intensity and scope of their effects.

4.3. Robustness Test

In order to ensure the reliability of research conclusions and overcome the possible errors in model setting or index measurement, this study has carried out a systematic robustness test.

4.3.1. Change the Method of Measuring the Development Level of Digital Economy

To enhance the robustness of research findings, this study employs principal component analysis (PCA) to remeasure the digital economy development level, conducts panel data regression analysis consistent with previous academic methodologies [50], and derives subsequent regression results. As illustrated in

Table 9. Robustness test for changing measurement method.

Variables	LnED	LnED
DIGe	0.765 ***	0.280 ***
	(0.0867)	(0.0484)
lnpgdp		1.926 ***
		(0.140)
lnpop		1.194 ***
		(0.0903)
edu		2.955 **
		(1.291)
mar		2.591 ***
		(0.680)
inf		−0.430 ***
		(0.138)
inv		17.02 ***
		(2.670)
Constant	6.435 ***	−20.92 ***
	(0.137)	(1.499)
Observations	4448	4448
R-squared	0.250	0.711

Note: T-statistics are in parentheses. ** denotes p < 0.05, *** denotes p < 0.01. (Source: Authors’ calculation using Stata 16.0).

, the regression outcomes after reconstructing digital economic indicators via PCA demonstrate high congruence with the benchmark model. In the baseline model without control variables, the elasticity coefficient of the digital economy index on export trade reaches 0.765 and is statistically significant. When control variables are incorporated, the coefficient of the primary explanatory variable decreases to 0.28 while still maintaining statistical significance. The stability of these results across different measurement approaches strongly validates the persistent positive impact of the digital economy on export trade.

4.3.2. Rejection of Municipality Regression

In order to test whether the core conclusions are driven by specific types of city samples, this study further eliminates the samples of central municipalities (Beijing, Shanghai, Tianjin, Chongqing) with particularity in administrative structure and economic functions and only uses the data of ordinary prefecture-level cities for panel data regression analysis. The operational design aims to eliminate potential sample heterogeneity interference, thereby enabling a more pure exploration of the correlation mechanism between the digital economy and export trade at the general urban level, which enhances the universality and robustness of research conclusions.

The regression outcomes, following the exclusion of samples from municipalities directly governed by the Central Government (refer to

Table 10. Robustness test for excluding municipalities.

Variables	LnED
DIG	7.967 ***
	(2.722)
lnpgdp	2.037 ***
	(0.140)
lnpop	1.182 ***
	(0.0941)
edu	2.991 **
	(1.336)
mar	2.806 ***
	(0.694)
inf	−0.480 ***
	(0.142)
inv	17.46 ***
	(2.789)
Constant	−22.31 ***
	(1.484)
Observations	4384
R-squared	0.684

Note: T-statistics are in parentheses. ** denotes p < 0.05, *** denotes p < 0.01. (Source: Authors’ calculation using Stata 16.0).

), indicate that the coefficient of the digital economy on export trade remains significantly positive, and it demonstrates a high degree of consistency with the benchmark regression results in both direction and significance. This result indicates that the primary findings of this study are not only influenced by the specific sample of municipalities directly governed by the Central Government, but also encompass a broader sample base, which enhances the robustness and generalizability of the research conclusions. This test further substantiates the overall enhancement of the digital economy on urban export trade and offers supplementary empirical evidence for Hypothesis H1.

4.4. Endogenous Test

In order to alleviate the potential endogenous bias, the study further implemented an instrumental variable test. Although the benchmark regression has adopted the fixed effect model to control some unobserved heterogeneity, the technology spillover effect of export activities may form a reverse causal chain—the overseas technology feedback acquired by enterprises in the export process may promote the development of the local digital economy. Therefore, based on the basic characteristics of the digital economy and its time lag effect on trade, the research draws lessons from the previous methodology framework [51], and selects the digital economy indexes of one period (DIG_lag1) and two periods (DIG_lag2) as tool variables. This design not only meets the correlation requirements of instrumental variables, but also weakens the reverse causal correlation through time difference processing, thus providing double guarantee for alleviating endogenous problems in econometrics. The results are shown in

Table 11. Instrumental variable regression.

	(1)	(2)
Variables	LnED	LnED
DIG_lag1	16.14 ***
	(2.443)
DIG_lag2		17.44 ***
		(2.861)
lnpgdp	1.882 ***	1.858 ***
	(0.142)	(0.148)
lnpop	1.196 ***	1.204 ***
	(0.0898)	(0.0907)
edu	3.009 **	3.103 **
	(1.303)	(1.327)
mar	2.558 ***	2.514 ***
	(0.681)	(0.684)
inf	−0.406 ***	−0.374 ***
	(0.139)	(0.139)
inv	16.85 ***	16.66 ***
	(2.786)	(2.949)
Constant	−21.08 ***	−21.04 ***
	(1.520)	(1.592)
Observations	4170	3892
R-squared	0.707	0.704

Note: T-statistics are in parentheses. ** denotes p < 0.05, *** denotes p < 0.01. (Source: Authors’ calculation using Stata 16.0).

.

The initial stage regression results indicate that the estimation coefficient of the instrumental variable DIG_lag1 is considerably positive at the 1% level, confirming a substantial correlation between the instrumental variable and the endogenous variable, which satisfies the validity criterion for the instrumental variable. The second stage regression findings indicated that the estimated coefficient of the endogenous variable DIG_lag2 remained significantly positive at the 1% significance level, and the sign was consistent with the benchmark regression results. This result not only validates that the advancement of the digital economy exerts a consistent long-term positive influence on urban export trade, but also demonstrates that the research conclusion remains valid even after accounting for endogenous factors, thereby augmenting the reliability and robustness of the measurement results.

4.5. Mediating Effect Analysis

4.5.1. Mediating Effect Test of Scientific and Technological Level

Table 12. The mediating effect test of the level of science and technology.

	(1)	(2)	(3)
Variables	LnED	sci	LnED
DIG	8.740 ***	0.093 ***	5.941 ***
	(3.744)	(0.019)	(1.467)
sci			23.457 ***
			(2.457)
lnpgdp	2.028 ***	0.010 ***	1.359 ***
	(14.661)	(0.001)	(0.053)
lnpop	1.184 ***	0.003 ***	1.083 ***
	(12.9031)	(0.000)	(0.032)
edu	2.909 ***	0.024 ***	2.406 ***
	(2.199)	(0.008)	(0.549)
mar	2.747 ***	0.001	1.551 ***
	(4.005)	(0.002)	(0.200)
inf	−0.482 ***	−0.003 ***	−0.950 ***
	(−3.4142)	(0.001)	(0.055)
inv	17.231 ***	0.157 ***	19.966 ***
	(6.294)	(0.013)	(1.354)
Constant	−22.202 ***	−0.113 ***	−15.017 ***
	(−15.262)	(0.006)	(0.591)
Observations	4448	4448	4448
R-squared	0.682	0.403	0.701
Sobel test	2.183 *** (z = 10.8)
Indirect effect	2.183 *** (z = 10.8)
Direct effect	5.941 *** (z = 9.5)
Total effect	8.125 *** (z = 12.8)
Proportion of total effect that is mediated	0.269 ***

Note: T-statistics are in parentheses. *** denotes p < 0.01. (Source: Authors’ calculation using Stata 16.0).

illustrates the empirical test results regarding the intermediary effect of technological progress, revealing the following characteristics: The fundamental regression Model (1) establishes a considerable promotional effect of the digital economy on export commerce, satisfying the initial criterion for the mediation effect assessment. Additional examination of Model (2) in Table 12 indicates that the advancement of the digital economy exerts a strong beneficial influence on technological progress and successfully meets the second criterion for intermediary effect. The regression results of Model (3) indicate that, after controlling for technological progress variables, the digital economy continues to play a direct role in enhancing export trade [29]. Additionally, the coefficient for technological progress is significantly positive, confirming that scientific and technological advancement serves as a substantial partial intermediary in the relationship between the digital economy and export trade. This discovery offers empirical validation for Hypothesis H3.

4.5.2. Test of Mediating Effect of Human Capital Level

Table 13. The mediating effect test of human capital level.

	(1)	(2)	(3)
Variables	LnED	Hum	LnED
DIG	8.740 ***	0.170 ***	7.422 ***
	(3.744)	(0.029)	(1.788)
hum			4.171 ***
			(0.961)
lnpgdp	2.028 ***	0.009 ***	1.552 ***
	(14.661)	(0.001)	(0.053)
lnpop	1.184 ***	0.003 ***	1.135 ***
	(12.9031)	(0.001)	(0.032)
edu	2.909 ***	−0.084 ***	3.346 ***
	(2.199)	(0.007)	(0.553)
mar	2.747 ***	0.034 ***	1.450 ***
	(4.005)	(0.003)	(0.208)
inf	−0.482 ***	−0.005 ***	−1.012 ***
	(−3.4142)	(0.001)	(0.055)
inv	17.231 ***	0.194 ***	22.860 ***
	(6.294)	(0.018)	(1.390)
Constant	−22.202 ***	−0.099 ***	−17.245 ***
	(−15.262)	(0.007)	(0.563)
Observations	4448	4432	4432
R-squared	0.682	0.350	0.682
Sobel test		0.708 *** (z = 4.3)
Indirect effect		0.708 *** (z = 4.3)
Direct effect		7.422 *** (z = 11.4)
Total effect		8.130 *** (z = 12.8)
Proportion of total effect that is mediated		0.087 ***

Note: T-statistics are in parentheses. *** denotes p < 0.01. (Source: Authors’ calculation using Stata 16.0).

illustrates the test findings of the intermediary mechanism of human capital deepening, highlighting the following critical characteristics: The benchmark regression Model (1) confirms the promotion of the digital economy to export commerce through the initial criterion of the intermediate effect test. An in-depth examination of Model (2) reveals that the advancement of the digital economy exerts a strong beneficial influence on the accumulation of human capital, hence satisfying the second criterion for the intermediary effect assessment. The regression outcomes of Model (3) indicate that, even after accounting for human capital variables, the digital economy continues to play a direct role in enhancing export trade [34]. Furthermore, the coefficient of human capital variables is significantly positive, confirming that human capital serves as a substantial partial intermediary in the relationship between the digital economy and export trade. This discovery offers empirical validation for investigating the transmission mechanism of Hypothesis H3, which posits that the digital economy propels trade expansion via the optimization of human capital, hence enhancing the robustness of the theoretical hypothesis.

The difference analysis of the contribution degree of the two-path mediating effect shows that the level of science and technology promotes export trade significantly higher than the level of human capital, and the difference of contribution degree between them is 18.2% (p < 0.01). This discovery suggests that in the digital economy era, policy design should follow the differentiation principle of “technology-led manpower coordination”.

4.6. Threshold Effect

4.6.1. Threshold Effect Test

In order to test the threshold effect of economic development level proposed in Hypothesis H4, this study constructs a nonlinear threshold regression model. The model utilizes economic development as the threshold variable to elucidate the stage-specific aspects of the digital economy’s influence on export commerce as it progresses with economic development stages. Refer to

Table 14. Threshold effect test results.

H0	H1	F-Value	Threshold Value	95% Conf. Interval	Conclusion
No threshold	Single threshold	311.58	9.3070 *** (0.0000)	[9.2332,9.3682]	Reject H0
Single threshold	Double threshold	220.77	10.2628 *** (0.0000)	[10.2257,10.3028]	Reject H0
Double threshold	Triple threshold	122.82	10.3747 (0.6667)	[10.3559,10.3909]	Fail to reject H0

Note: T-statistics are in parentheses. *** denotes p < 0.01. (Source: Authors’ calculation using Stata 16.0).

for specifics about parameter estimate and threshold test outcomes. The test results indicate that both the single threshold and double threshold are significant at the 1% level. However, the three thresholds are not significant; thus, the double threshold model is ultimately chosen for estimation. The result indicates that the degree of economic development serves as a critical structural inflection point influencing the interaction between the digital economy and export commerce.

4.6.2. Analysis of Threshold Regression Results

This study further examines the asymmetric impact of the digital economy (DIG) on export commerce across various stages of economic growth utilizing the double threshold model. Based on the estimation results in

Table 15. Regression results of the threshold effect.

Variables	Coefficient	Std.Err	t-Value	p-Value
DIG (lnpgdp ≤ 9.3070)	0.538 ***	0.052	10.44	0.000
DIG (9.3070 < lnpgdp ≤ 10.2628)	0.413 ***	0.053	7.85	0.000
DIG (lnpgdp > 10.2628)	0.295 ***	0.057	5.18	0.000
lnpop	2.105 ***	0.420	5.02	0.000
edu	0.242	0.780	0.31	0.757
mar	−0.704	0.254	−0.28	0.782
inf	0.164 *	0.721	2.27	0.024
inv	0.360	1.066	0.34	0.736
Constant	−4.383 *	2.992	−1.72	0.086

Note: T-statistics are in parentheses. * denotes p < 0.1, *** denotes p < 0.01. (Source: Authors’ calculation using Stata 16.0).

, we divided the sample cities into three development stages using per capita GDP as the threshold variable: low development stage (per capita GDP < CNY 77,000), medium development stage (CNY 77,000 ≤ per capita GDP < CNY 196,000), and high development stage (per capita GDP ≥ CNY 196,000). The empirical results clearly reveal that the role of the digital economy in promoting export trade presents a significant nonlinear decreasing characteristic, with marked differences in its internal mechanism of action, effect quality, and core constraints [37,39].

(1)

Urban agglomerations in the low development stage: the “basic empowerment” effect of the digital economy is significant

This stage includes urban agglomerations such as the Guanzhong Plain, Beibu Gulf, Harbin-Changchun, Liaozhongnan, Qianzhong, Dianzhong, Hohhot-Baotou-Ordos-Yulin, Lanzhou-Xining, Ningxia Yellow River Basin, and Northern Tianshan. The promotion coefficient of the digital economy on exports is the highest at 0.53. The core mechanism is to drive export growth by filling the gaps in traditional trade and reducing information asymmetry, with “popularization of digital infrastructure + penetration of cross-border e-commerce” as the core. For example, relying on the “Digital Silk Road” nodes (such as the smart logistics platform in Xi’an International Port Area), the Guanzhong Plain urban agglomeration has improved customs clearance efficiency by 30% through digital customs declaration, effectively promoting the export of agricultural products and traditional industrial products. The Beibu Gulf urban agglomeration, leveraging the policy of cross-border e-commerce comprehensive pilot zones, uses digital platforms to connect with the ASEAN market, confirming the mechanism of “the digital economy breaking geographical restrictions”.

Industries at this stage are mainly labor intensive and resource based, and the digital economy mainly reduces export costs through basic applications such as online marketing and order matching, rather than promoting export trade through technological upgrading. However, its development is constrained by weak digital infrastructure and a shortage of digital skill talents, resulting in effects concentrated in low-value-added links, making it difficult to extend to the high end of the industrial chain. Therefore, urban agglomerations in the low development stage need to strengthen “digital inclusion,” such as expanding cross-border e-commerce coverage in the Guanzhong Plain and improving digital logistics hubs in the Beibu Gulf, to alleviate the “digital divide” problem.

(2)

Urban agglomerations in the medium development stage: the “efficiency improvement” effect of the digital economy is prominent

This stage includes urban agglomerations such as the Central Plains, Shandong Peninsula, Middle Reaches of the Yangtze River, and Chengdu-Chongqing. The digital economy coefficient drops to 0.413, and the core mechanism shifts to “supply chain digitalization + industrial cluster linkage,” with the effect intensity being lower than that in the low stage but higher in quality. The digital economy is deeply integrated with regional leading industries. For example, the Central Plains urban agglomeration (Zhengzhou) has adopted the “cross-border e-commerce + industrial belt” model, integrating production and export links through a digital middle platform, with the export volume of the industrial belt increasing by 38% in 2023. The Shandong Peninsula urban agglomeration, combined with the marine economy, has optimized the supply chain collaboration in shipbuilding and aquatic product processing using digital technologies, improving export efficiency by 25%.

The development stage is dominated by mid-end manufacturing, and the digital economy enhances product competitiveness through “intelligent manufacturing and digital logistics.” For example, the intelligent terminal industrial park in the Chengdu-Chongqing urban agglomeration has optimized the production process through the industrial internet, significantly increasing the proportion of high-tech product exports, which reached 63.3% in 2025, confirming the conclusion that “the digital economy promotes the improvement of the complexity of export technologies.” The constraints lie in the “digital divide” and “skill mismatch,” as traditional industrial workers lack digital skills and some enterprises lag in digital transformation, reflecting the constraint of insufficient human capital accumulation on the exertion of effects.

(3)

Urban agglomerations in the high development stage (per capita GDP ≥ CNY 196,000): the “innovation-driven” effect of the digital economy dominates

This stage includes the Beijing–Tianjin–Hebei, Yangtze River Delta, and Guangdong–Hong Kong–Macau Greater Bay Area. The digital economy coefficient further drops to 0.295, with the core mechanism being “technological innovation + data factor flow.” Although the effect intensity is low, the quality is the highest, which is consistent with the characteristic of “diminishing marginal returns but improved empowerment level in the high development stage.” The digital economy deeply empowers high-end industries and trade in services. For example, the Yangtze River Delta has realized cross-border data flow through the “Digital Free Trade Zone” pilot (such as Shanghai Lingang), driving the export of integrated circuits and financial technology. In the Guangdong–Hong Kong–Macau Greater Bay Area, relying on the “Guangzhou–Shenzhen–Hong Kong–Macau Science and Technology Corridor,” the digital economy drives cross-border e-commerce and software service exports to account for more than 40%, reflecting the “dual drive of digital industrialization and industrial digitalization.” Industries at this stage are dominated by high-end manufacturing and digital services, and the digital economy improves the added value of exports through digitalized R&D and precise market positioning. For example, the industrial Internet platform in the Yangtze River Delta has shortened the R&D cycle by 30% and increased the patent conversion rate to 45%, reflecting the mechanism through which the digital economy promotes export upgrading by improving R&D efficiency.

Facing “technical bottlenecks” and “rule barriers,” such as the dependence of chip design in Beijing–Tianjin–Hebei on external technologies and the restriction of cross-border data flow in Guangdong–Hong Kong–Macau by international rules, it is necessary to break through the constraint of insufficient core technological competitiveness. Urban agglomerations in the high development stage need to focus on “innovation breakthroughs,” such as increasing investment in digital technology R&D in Beijing–Tianjin–Hebei and exploring the marketization of data factors in the Yangtze River Delta to address the constraint of “the need to improve the position in the global value chain”.

4.7. Discussion of Results

Heterogeneity analysis (Table 7 and Table 8) clearly shows that the absolute effect of the digital economy on export trade in eastern regions (with a regression coefficient as high as 9.388) is significantly stronger than that in central and western regions (with a coefficient of 2.227). This reflects that in terms of stock scale, eastern regions have formed a huge “volume advantage” in driving exports through the digital economy by virtue of their solid digital infrastructure, high-end talent reserves, and mature industrial clusters. This advantage is the result of long-term accumulation and a direct manifestation of its high contribution to national exports at the current stage.

On the other hand, the threshold effect analysis (Table 15) reveals the “marginal diminishing” challenge hidden behind this “volume advantage” from the perspective of dynamic efficiency. When the per capita GDP crosses the threshold of the high development stage of CNY 196,000, the marginal promoting effect of the digital economy on exports begins to weaken. The internal logic is that the application of digital technology in the eastern region has tended to be saturated, the dividends from basic and universal digital transformation have been basically released, and the marginal output of continued investment naturally declines. At this time, the bottleneck of export growth has shifted from “insufficient technology application” to “insufficient technological innovation” and “institutional constraints,” such as facing challenges such as core technology being “stuck” and barriers to international data rules. In contrast, although the per capita GDP in the central and western regions is in the low and medium development stages, theoretically in the “golden rising period” of the marginal effect of the digital economy, due to their “shortcomings in basic conditions” in terms of digital infrastructure, talent attraction, and industrial supporting facilities, it is difficult to transform the theoretically high marginal effect into a high absolute effect in reality. Therefore, although the potential return rate brought by “each additional unit of digital economy input” is higher, the base and quality of their “total digital economy input” are far lower than those in the eastern region, and the contribution to the total export volume (absolute effect) is still low.

5. Conclusions and Policy Recommendations

5.1. Main Conclusions

This paper conducts an empirical analysis utilizing panel data from 273 Chinese cities spanning from 2006 to 2022, employing both the intermediary effect model and threshold model to systematically elucidate the mechanisms and intricate characteristics of the digital economy’s influence on export trade. For the first time, it reveals the nonlinear characteristics and stage-specific differences of this relationship through the dual mediation effect and threshold effect models, filling the gap in the existing literature regarding the understanding of the mechanism by which the digital economy empowers export trade, culminating in the following core conclusions:

(1)

The digital economy plays a crucial and consistent role in enhancing export trade. Benchmark regression results indicate that advancements in the digital economy considerably enhance the growth of urban export commerce volume. The conclusion remains valid across many robustness tests, including variable replacement, sample reduction, and the instrumental variable technique, affirming the pervasive positive impact of the digital economy on export commerce.

(2)

The influence of the digital economy on export trade exhibits considerable geographical variability, establishing a “core–edge” differentiation pattern. Heterogeneity study indicates that the advancement of the digital economy in facilitating export trade in the eastern region is approximately 4.2 times greater than in the central and western regions, with central cities exhibiting a much superior effect compared to periphery cities. This disparity originates from the eastern region and central cities’ superior digital infrastructure and industrial chain integration, policy, and resource concentration, underscoring the tangible contradiction between the regional digital divide and the uneven distribution of resources. It reflects the “Matthew effect” of China’s digital economy empowering exports—advantaged regions, by virtue of their first-mover accumulation, can more effectively capture and convert digital dividends. This divergence is rooted in the regional imbalance of factor endowments such as digital infrastructure, industrial foundation, talent reserves, and internationalization level, confirming the real-world manifestation of the “digital divide” in the trade sector.

Compared to existing studies that often focus on the average national effect, our detailed regional and city-tier heterogeneity analysis provides a more nuanced and contextualized understanding, revealing the “Matthew effect” in digital trade empowerment and highlighting the risk of increasing regional disparities despite overall growth.

(3)

Scientific and technological innovation, along with human capital, are the primary conduits by which the digital economy enhances export trade. The intermediary effect test demonstrates that the digital economy facilitates the advancement of technical complexity in export products and bolsters the international competitiveness of enterprises by enhancing the efficiency of scientific and technological investments and optimizing human capital structure. Investment in science and technology enhances R&D capabilities and the transformation of achievements, while the accumulation of human capital enables firms to satisfy international high-end market demands by augmenting skill adaptability. The collaboration between them expedites the transition of export trade from a factor-driven to an innovation-driven model.

Our mediation analysis moves beyond establishing a mere correlation to unpack the “black box” of how the digital economy affects exports. The quantification of the contribution difference between technological progress (26.9%) and human capital (8.7%) offers a rare and valuable insight into the principal mechanism at China’s current development stage, which is a significant gain over previous studies that often treat these channels in isolation. At the current stage, the core mechanism of China’s digital economy empowering exports is more inclined to “technology substitution” and “efficiency improvement” rather than “skill complementarity.” This may be related to China’s position in the global value chain, industrial structure characteristics, and stage-specific features of digital transformation, i.e., enterprises tend to reduce costs and improve production efficiency by introducing automated and intelligent equipment, while the transformation of systematically upgrading labor skills lags relatively behind.

(4)

There exists a nonlinear threshold effect in the influence of the digital economy on export trade, which initially increases and subsequently decreases with the level of economic progress. However, the quality of driving effects progressively upgrades from low-value-added to high-value-added drivers, with significant variations in core mechanisms and industrial foundations across development stages. Low-development-stage city clusters (such as Guanzhong Plain and Beibu Gulf) rely on a foundational enabling mechanism of “digital infrastructure proliferation + cross-border e-commerce penetration,” focusing on labor-intensive and resource-based industries to drive exports by reducing information asymmetry and trade costs. Mid-development-stage clusters (such as Central Plains and Chengdu-Chongqing) employ an efficiency enhancement mechanism through “supply chain digitization + industrial cluster synergy,” concentrating on mid-range manufacturing to enhance product competitiveness via smart manufacturing and logistics optimization. High-development-stage clusters (such as Beijing–Tianjin–Hebei and Yangtze River Delta) depend on an innovation-driven mechanism powered by “technological innovation + cross-border data flow,” specializing in high-end manufacturing and digital service trade to promote export upgrading through improved R&D efficiency and data mobility. Constraints exhibit hierarchical differences: low-development stages face limitations from weak digital infrastructure and skill shortages; mid-development stages encounter challenges of digital divides and skill mismatches; and high-development stages confront core technological bottlenecks and international regulatory barriers.

5.2. Theoretical Implications

This research enriches the theoretical framework of digital economics and international trade in three main ways: First, it integrates the concept of “threshold effects” into the study of digital economy and exports, proposing a more dynamic and nonlinear theoretical perspective. Second, it provides empirical evidence for the “core–edge” theory and “Matthew effect” within the context of digital trade in a large developing country. Third, it dissects and compares the efficacy of different mediation channels (technology vs. human capital), adding granularity to the theoretical understanding of transmission mechanisms.

5.3. Practical and Policy Recommendations

(1)

Address the regional digital disparity and establish a distinct developmental trajectory.

Considering the attributes of regional heterogeneity, the developed eastern regions should prioritize advancements, facilitate the profound integration of digital technology research and development with high-end service trade, establish national digital technology research and development centers, and endorse the implementation of avant-garde technologies in areas such as cross-border payment and digital supply chains, exemplified by the “Digital Free Trade Zone” pilot project in the Yangtze River Delta. In the central and western regions, it is imperative to execute the strategy of “digital infrastructure addressing deficiencies,” prioritizing the establishment of new infrastructures such as 5G and computing power centers. This entails accommodating the computing power demands of the east through “computing from the east to the west,” constructing green data center clusters in Guizhou and Inner Mongolia, and enticing enterprises to relocate through tax incentives. Simultaneously, create a “core-radiation” technology exchange platform, incentivize center cities to provide data and algorithm resources to surrounding cities, implement the “digital enclave” policy, and enhance regional collaboration.

(2)

Implement dynamic adaptation measures to surpass the threshold of economic progress.

Based on threshold effect analysis, differentiated strategies should be implemented across regions at varying development stages. For low-development city clusters, enhancing digital inclusion and strengthening foundational support is critical: accelerate infrastructure upgrades by prioritizing 5G deployment and smart logistics hubs while expanding fiber-optic coverage and cross-border e-commerce pilot zones (e.g., Guanzhong Plain extending e-commerce services to county-level specialty industries; Beibu Gulf improving ASEAN-oriented digital logistics corridors). Implement “Digital Skills for All” training programs through partnerships with e-commerce platforms (e.g., cross-border operations, digital customs clearance) to alleviate talent shortages. Leverage regional resources to cultivate “digital + specialty industry” export models (e.g., digital agriculture exports in Central Yunnan; digital energy product trade along Tianshan North Slope).

For mid-development clusters, advance digital–industry–labor synergy to upgrade efficiency: focus on core industries’ digital transformation through “industrial belts + digital platforms” integration (e.g., home appliances/textile belts in Central Plains linking with e-commerce platforms; marine industries in Shandong Peninsula converging with industrial IoT). Optimize human capital by aligning vocational education with digital economy demands (e.g., adding smart manufacturing/industrial software programs in Chengdu-Chongqing vocational institutes; developing enterprise–academy customized training). Eliminate regional digital barriers by promoting data sharing and supply chain coordination (e.g., establishing cross-city logistics data platforms in Yangtze River Mid-Reach clusters to enhance export efficiency).

High-development clusters must overcome innovation bottlenecks to lead high-end upgrading: increase R&D investment in “chokepoint technologies” like semiconductors, industrial software, and cross-border data governance, establishing national digital innovation hubs (e.g., Beijing–Tianjin–Hebei joint semiconductor design initiatives; Yangtze River Delta pilots for data factor marketization). Deepen alignment with digital trade rules through pilot free trade zones (e.g., Shanghai Lingang, Shenzhen Qianhai) to test cross-border data flows and digital service trade, building internationally compatible regulatory frameworks. Advance “digital industrialization + industrial digitization” dual drivers to cultivate high-end digital clusters (e.g., AI and fintech in Greater Bay Area to boost high-value service trade exports).

(3)

Enhance the intermediary transmission between scientific and technological innovation and human capital.

Concentrate on the dual primary channels of scientific and technological innovation and human capital, while augmenting investment. Establish a dedicated R&D fund for the digital economy, emphasizing support for export-oriented technological research, fostering a collaborative model of production, education, and research including enterprises, universities, and government funding, while expediting the technology industrialization process. Enhance the configuration of human capital, facilitate the profound integration of vocational education with the digital economy, establish majors relevant to digital trade, and collaborate with prominent firms to implement order-based training. Execute the “Digital Talent Settlement Plan” in the central and western areas to attract and retain skilled individuals. Establish a cohesive national digital skills certification framework to enhance labor market adaptability.

(4)

Strengthen the “digital + foreign trade” synergy to enhance resilience in coping with external shocks.

The research in this paper confirms that the digital economy is the core engine driving the growth of urban exports. To cope with the external uncertainties brought about by the adjustment of the global economic structure, this internal driving force should be transformed into a solid barrier against external risks. Specifically, enterprises should be guided to actively explore diversified international markets relying on the core paths such as digital platform empowerment and element-driven data as revealed in this paper, and use big data analysis to build a global demand early warning mechanism, so as to transform the inherent advantages of the digital economy in empowering exports into the practical ability to cope with external shocks and ensure the stability and resilience of the foreign trade system.

5.4. Research Outlook

This study reveals the internal driving mechanism of the digital economy on export trade based on panel data of Chinese cities. Given China’s unique position in the global digital economy and trade landscape, the conclusions of this study provide an important reference for understanding the digital transformation and foreign trade growth of emerging economies. Although the findings of this study may have limitations when extended to other countries, they have strong reference value for other emerging economies facing similar development stages and structural characteristics.

Firstly, the rise of China’s digital economy has distinct policy-driven characteristics. The strong investment and strategic guidance of the central and local governments in digital infrastructure (such as the “Eastern Data and Western Computing” project) have formed a unique top-down digital governance model, which is relatively rare in other countries, especially in developed countries with a higher degree of marketization. Secondly, there is a significant regional gradient within China. The huge differences between the eastern coastal areas and the central and western inland areas in terms of resource endowments, industrial structure, and opening-up level not only amplify the unbalanced effect of the digital economy but also make the relationship between “absolute effect” and “marginal effect” revealed in this study rooted in China’s unique regional development pattern. Thirdly, there is a deeply nested “chain–group synergy” between China’s manufacturing clusters and digital platforms (such as e-commerce platforms and live broadcast bases; e.g., the ‘factory zone + live broadcast base’ model in the Yangtze River Delta region). This unique digital transformation path is the product of the integration of a specific industrial ecosystem and digital technology, which is difficult to simply replicate.

Therefore, the conclusions of this study provide an important reference for understanding the digital transformation and foreign trade growth of emerging economies. However, due to differences in institutional environments, industrial structures, digital infrastructure, and other aspects among countries, their applicability needs to be verified through further empirical research. Looking forward, we suggest that subsequent studies can be extended to cross-country comparative analyses, especially focusing on countries along the “Belt and Road Initiative.” By examining digital economy practices under different institutional environments, cultural backgrounds, and industrial bases, we can explore the cross-cultural adaptation mechanisms and universal laws of the digital economy empowering export trade, thereby providing richer empirical evidence for the formulation of global digital trade policies.