Impact of Digital Economy on Inter-Regional Trade: An Empirical Analysis in China

Abstract

Unimpeded domestic inter-regional trade is an inevitable choice for a country to improve its economic development autonomy and internal stability. The booming development of the digital economy profoundly affects inter-regional trade exchanges and the construction of domestic trade patterns. Based on China’s inter-provincial panel data, this study analyzes the mechanism channels and regional heterogeneity of the digital economy’s impact on inter-regional trade. We found that firstly, the digital economy significantly promotes inter-regional trade outflows and inflows with positive spatial spillover effects. Secondly, the digital economy promotes inter-regional trade by reducing trade costs and stimulating market demand, while the role of the resource allocation effect and technological innovation effect needs to be enhanced. Thirdly, the promotion of inter-regional trade by the digital economy is more prominent and has the potential for less developed regions or non-border regions. In addition, the digital economy significantly boosts inter-regional trade in labor-intensive regions, while it has a limited effect on inter-regional trade in technology-intensive regions. Therefore, this study suggests increasing the construction of digital infrastructure in less developed regions, tapping the role of the digital economy in resource allocation, and promoting the in-depth integration of the digital economy with the region’s advantageous industries, so as to promote inter-regional trade and economic development.

1. Introduction

Since the reform and opening up, China has actively integrated into the international economic cycle by relying on the export-oriented development model. China has maintained high growth in foreign trade, especially after joining the World Trade Organization, and has grown to become the world’s largest trading country in goods. With the external environment uncertainty and internal economic restructuring, a new development pattern will gradually be created whereby domestic and foreign markets can boost each other, with the domestic market as the mainstay. However, China’s traditional comparative advantage has declined significantly, and the growth space of export trade, mainly processing trade, has been compressed, and there is an urgent need to explore new growth drivers [1]. As the momentum of international economic circulation has weakened, tapping domestic market demand and stimulating domestic trade potential have become important sources of economic growth for all countries. The domestic inter-regional trade (IRT) is an important manifestation and component of the domestic cycle, which can establish inter-regional economic ties and effectively promote the cross-regional flow of factors and the synergistic development of internal and external trade [2]. At the same time, China has the advantage of a super large-scale market and the most complete industrial advantage, which can both add momentum to China’s economic development and drive the world’s economic recovery. Therefore, making full use of the domestic consumer market to drive domestic circulation, which, in turn, drives and optimizes the circulation of the international economic system at a higher level, is the essential content of dual circulation. However, there is a significant market segmentation between regions in China, and its internal market is not unified for a long time. Local protection motives and disorderly competition have led to the intensification of regional trade barriers, which seriously hinder the trade cycle in the domestic market [3,4]. How to explore effective ways conducive, thus promoting smooth domestic circulation and inter-regional trade is an important issue facing China today and a strategic choice for future development.

As a new economic form in the new era, the digital economy (DT) has triggered a “digital butterfly” that plays an important role in production, consumption and distribution. This has provided new momentum for countries to promote inter-regional trade. With the deepening of the digital revolution in economic activities, the digital economy plays an important role in promoting production, rapid circulation and stimulating diversified demand. Digital technologies such as the Internet of Things, cloud computing and artificial intelligence link domestic market advantages and data advantages, which strengthens inter-regional trade by breaking down distance constraints as well as weakening local protections. The digital economy can accelerate the continuous optimization and mutual matching of both supply and demand, profoundly affecting the way and structure of the domestic trade cycle [5]. The digital economy with data as the core production factor has greatly changed the traditional inter-regional trade model. Based on this, can the digital economy as a new paradigm really better fuel domestic trade? Further, what are the channel mechanisms through which the digital economy affects inter-regional trade at the theoretical level? And are there differences in its impact due to the uneven development between regions and industries at the practical level? Answers to these questions will help clarify the effects of the digital economy on domestic inter-regional trade and provide theoretical support and implementation paths for deepening the effectiveness of the digital economy, promoting domestic market integration and realizing sustainable economic growth.

This study is based on the Web of Science database and CNKI Chinese database, which are searched by the themes of “digital economy”, “inter-provincial trade” and “inter-regional trade”, and in the CNKI database, there are more studies on inter-provincial trade and domestic trade, with the time span of 1993–2023, while there are fewer studies on digital economy and inter-regional trade, with the time span of 2021–2023. Web of Science database, more studies on inter-regional trade with the time span of 1962–2023, and less studies on digital economy and inter-regional trade with the time span of 2021–2023.

One strand of literature deals with the measurement of inter-regional trade flows. Earlier scholars used the well-known trade gravity model to establish domestic trade flows [6,7,8]. Some scholars have focused on domestic trade at a macro level. The methods included GDP minus total exports [9], total output minus total exports [10] and input-output tables [11,12,13]. Among them, more scholars choose input-output tables to measure a country’s inter-regional trade, but this data does not have continuity and cannot examine the stability of the impact effect over a certain period of time. In recent years, scholars have discussed interprovincial trade based on interprovincial rail freight volume data as countries continue to focus on domestic trade markets [14,15].

Another category of literature is the study of the trade effects of the digital economy. Existing studies have sorted out the basic mechanisms of the digital economy to boost domestic trade based on a double-loop perspective, showing from the theoretical level that the digital economy can stimulate demand, upgrade supply and reduce costs to smooth the domestic loop [16,17]. At the practical application level, the digital economy relies on digital technology, and the use of information and communication technology, digital platforms and blockchain technology can significantly reduce trade costs [18,19,20]. Some scholars have also explored the ability of the digital economy to enhance bilateral trade and significantly boost domestic trade flows by constructing a comprehensive digital economy index [21]. Given the high degree of intersection of the Internet and digitalization, scholars have explored the impact of the Internet on domestic and international trade [22]. Further, some studies have found that the digital economy causes more growth in domestic trade than international trade, and the scope for being able to enhance inter-regional connectivity remains huge [23].

Throughout the studies, a large amount of valuable literature has been accumulated on the measurement and characterization of domestic inter-regional trade, but there is a lack of literature on how to further promote inter-regional trade and analyze its influencing factors. At the same time, existing studies are more likely to use cross-sectional data to examine inter-regional trade, and there is a lack of characterization of inter-regional trade over consecutive periods, as well as a scarcity of literature that explores inter-regional trade from a two-way perspective of outflows and inflows. As a new driving force, existing studies examining the impact of the digital economy on inter-regional trade have focused only on the path of reducing trade costs. In fact, the digital economy deeply integrates industrial development and can affect inter-regional trade in various aspects. In addition, existing studies have only explored the heterogeneity of the effect of the digital economy on inter-regional trade under different levels of economic development, ignoring the heterogeneity of geographic location and resource endowment. Based on this, this study uses bilateral asymmetric trade data from 29 Chinese provinces and cities during 2006–2017 to clarify the differences between outflows and inflows of inter-regional trade in China and to examine the impact of the digital economy on inter-regional trade. It is found that the digital economy can significantly promote inter-regional trade, and the impact effect on trade outflow is always larger than trade inflow and has a significant positive spillover effect in space. The channel mechanism concludes that the digital economy significantly reduces trade costs and stimulates market demand thus promoting outflows and inflows, while the role of resource allocation effect and technological innovation efficiency needs to be further enhanced. The digital economy has a significant difference in inter-regional trade in areas with different geographical locations and different industry densities.

The marginal contributions of this paper are: first, based on the perspective of domestic trade circulation, the effectiveness of the digital economy in promoting inter-regional trade is verified from both trade outflows and inflows, which fills the gap of existing studies. Second, this study provides an in-depth discussion of the channels through which the digital economy affects inter-regional trade from the transportation channel, the demand side and the supply side, which broadens the channel mechanism. It Is found that the digital economy plays an important role in promoting regional trade on the demand side, while on the supply side, the digital economy optimizes the production efficiency of products and services, which needs to be further improved. Third, this study is based on coastal regions, inland regions and border regions, as well as economically developed regions (eastern) and less developed regions (central and western) and finds that for less developed regions or non-border regions, the digital economy has a more prominent and potential role in promoting inter-regional trade. This provides guidance on how regions with weak digital infrastructure in China and countries with strong domestic markets, such as India, can more effectively promote domestic consumption. In addition, studies based on labor-intensive, capital-intensive, resource-intensive and technology-intensive regions find that the digital economy significantly boosts the trade cycle in labor-intensive regions, while the impact effect on the trade cycle in technology-intensive regions is low. This provides guidance on how to make full use of comparative advantages among regions and how to promote the domestic technology cycle.

The remainder of this paper consists of the following sections. The theoretical mechanism and research hypotheses are detailed in Section 2. Section 3 presents the research design by introducing the model and describing the variables. Section 4 presents the results of the empirical tests, which include the benchmark test, spatial effect test, robustness test and endogeneity test. Section 5 is further research, including mechanism testing and heterogeneity testing. Finally, Section 6 provides study findings and policy recommendations.

2. Theoretical Mechanism

The rapid development of the digital economy can take full advantage of the scale of the domestic market, deepen domestic market reforms and break the barriers to the domestic trade cycle. From the perspective of intuitive factors affecting the domestic trade cycle, trade costs are an important factor impeding the smooth flow of domestic trade, including both objective trade costs, such as freight costs, and transaction costs caused by subjective factors such as differences in regional preferences, as well as institutional trade costs due to local trade protection [24]. The high permeability and low cost of the digital economy have broken the time and space limitations of market transactions, effectively reducing multiple forms of trade costs for trade transactions, and thus promoting the domestic trade cycle.

On the one hand, the deep integration of the digital economy and traditional industries effectively reduces the explicit costs of inter-regional trade. For example, in the logistics industry, digital technology is applied to the logistics system in order to promote the rapid development of digital logistics. It is able to plan and select efficient and visualized physical freight paths, significantly shorten the logistics journey and improve turnaround efficiency, ultimately reducing the transportation costs under traditional trade [25]. For the financial industry, the digital economy integrates with financial service systems to build an integrated online trade system, promote payment or settlement platforms to achieve convenience, and effectively reduce the transaction costs of inter-regional trade [26]. On the other hand, the digital economy uses digital technology and platforms to break through the limitations of time and geographic space, significantly cutting the hidden costs of inter-regional trade. The digital economy effectively integrates the massive domestic consumer demand with supply chain supply to achieve effective docking and matching of diversified demand and diverse supply, which weakens the information asymmetry of inter-regional trade exchanges [27]. At the same time, the digital economy builds application platforms to facilitate trade communication, which improves the efficiency of bilateral transaction negotiation for cross-regional trade and promotes the enhancement of inter-regional trade exchanges. Based on the above analysis, this paper proposes:

H1. 

The digital economy promotes inter-regional trade by reducing the cost of inter-regional trade.

Market demand is the core driver of the domestic trade cycle, and the digital economy helps unleash the huge potential of the consumer side and thus boosts the domestic trade cycle. First of all, the digital economy has led to the diversification of product and service demands, stimulating the flourishing of diversified and personalized new industries, expanding the categories of products and service transactions in the supply market, generating and nurturing a wide range of market sources and order demands and thus expanding the scale of market demand. Further, the digital economy promotes the smooth flow of market demand mechanism, expression mechanism and transmission mechanism through the digital platform, and forms a timely and effective iterative feedback loop with producers, which slows down the flow of domestic consumption out of the international market and thus promotes a large cycle of domestic trade [28]. Secondly, the digital economy promotes the convenience of trade in products and services. It enhances the convenience of product trade with disruptive changes in payment methods, consumption platforms and logistics and transportation, which greatly improves the convenience of consumption. At the same time, traditional service trade is deeply integrated with digital technology, such as the application of video conferencing technology and virtual reality technology, which enriches the types of service trade such as medical care, education and R&D, enhancing the convenience of service trade and thus promoting inter-regional trade. Finally, the digital economy promotes the productivity of low-skilled laborers and thus improves the income disparity of residents, promotes the consumption willingness of middle and lower classes consumers, access information on goods and services from other regions through the Internet and especially the increasing share of cross-regional spending on culture, education, sports and entertainment, which increases differentiated consumption and thus boosts domestic trade exchanges [29]. Based on the above analysis, this paper proposes:

H2. 

The digital economy promotes inter-regional trade by stimulating market demand.

The total allocation effect of the digital economy to promote the transfer of production factors from inefficient industries or regions to efficient industries or regions is an important reason to enhance production efficiency and thus promote inter-regional trade exchanges. Digital technology enables factors of production to be geographically and spatially widespread, facilitating the active allocation of factors of production between regions, thereby reducing factor market distortions and optimizing resource mismatches, enhancing regional productivity and promoting inter-regional trade [30]. On the one hand, the new model of digital finance bred by the digital economy, with its powerful network platform, integrates information flow to promote the efficient operation of capital, accelerates the flow of capital throughout society and provides more capital factors supply for the whole economic system. At the same time, the payment system of “Internet Finance” overcomes the time and space constraints and stock limitations of traditional transactions effectively circumventing the negative externalities of the financial market, improves the elasticity of capital accumulation and the matching of supply and demand, thus improving capital mismatch and promoting inter-regional linkages [31,32]. On the other hand, the digital economy can break the spatiotemporal constraints of labor market supply and demand, and effectively solve the problem of information asymmetry in labor market. Digital platforms enhance the transparency of labor market compensation, increase employment options for both current and potential employees, and help weaken negative impacts such as frictional and cyclical unemployment [33]. It enables the labor resource allocation effect to be greatly enhanced, which in turn drives the network trading of goods and services to promote inter-regional trade. Based on the above analysis, this paper proposes:

H3. 

The digital economy promotes inter-regional trade by improving the efficiency of resource allocation.

The digital economy promotes the integration and synergy of data, information and traditional factors, changes the original innovation structure and innovation organization and promotes intelligent distribution and reorganization of innovation factors within and across regions, thus releasing innovation momentum and improving innovation efficiency. At the same time, digital technologies help enterprises integrate and analyze effective information, promote the dissemination of innovation knowledge and generate an Increase in innovation activities, increase the stock of effective innovation knowledge and thus enhance innovation efficiency [34]. Further, the digital economy enables innovation spillover through digital application platforms, which gradually weakens innovation boundaries, shortens regional differences in technological innovation capabilities, thereby enriching the diversity of product offerings in each region and enhancing market competitiveness and promoting trade outflows from the province [35]. Enhancing product intelligence attributes and improving innovation efficiency provide the impetus for domestic trade outflows [36]. At the same time, the digital economy promotes increased local innovation demand by enhancing innovation efficiency, which significantly increases the demand for high-quality intermediate goods and key technology services, which in turn promotes trade inflows from other provinces to the province. Synthesizing the above analysis, this paper proposes:

H4. 

The digital economy promotes inter-regional trade by improving the efficiency of technological innovation.

In summary, Figure 1 depicts the channel mechanisms through which the digital economy affects inter-regional trade.

3. Research Design

3.1. Model Setting

3.1.1. Benchmark Regression Model

In order to examine the impact of the digital economy on inter-regional trade, the basic model (1) is set to conduct empirical analysis on this topic.

$$\ln IR{T}_{ijt}={\alpha }_0+{\alpha }_{1}D{T}_{it}+{\alpha }_2{\displaystyle \sum X_{ijt}+{\theta }_i+{\theta }_j+{\mu }_t+{\epsilon }_{ijt}}$$

(1)

where i and j denote provinces and t denotes year. $IR{T}_{ijt}$ denotes the inter-regional trade, including trade outflows from province I to province j and trade inflows from province j to province i, which are measured logarithmically. $D{T}_{it}$ denotes the digital economy, and $X_{ijt}$ denotes a series of control variables. Drawing on the research [37], the sending and arriving provinces are fixed. ${\theta }_i$ and ${\theta }_j$ denote Fixed effects of sending province and arriving province, ${\mu }_t$ and ${\epsilon }_{ijt}$ denote time effect and random error terms. ${\alpha }_1$ is the parameter to be estimated that the digital economy impacts the inter-regional trade.

3.1.2. Spatial Econometric Model

There is the spatial correlation between sending province and arriving province in the inter-regional trade cycle. The development of the digital economy has prompted the establishment of close ties of production cooperation or industrial division of labor across regions, enhancing the spatial association between producers and consumers across regions with the help of channels such as information resources and trading platforms, resulting in local trade outflows and inflows also being spatially influenced by the development of the digital economy in neighboring regions [38].

Firstly, there is a need to verify if there is spatial autocorrelation between the digital economy and inter-regional trade. This study uses the Global Moran’s I index to examine. The equation of the index is as follows:

$$Moran’s I=\frac{n}{{\displaystyle {\sum }_{i=1}^n{\displaystyle {\sum }_{j=1}^n{w}_{ij}}}}\times \frac{{\displaystyle {\sum }_{i=1}^n{\displaystyle {\sum }_{j=1}^n{w}_{ij}(x_i-\overline{x})(x_j-\overline{x})}}}{{\displaystyle {\sum }_{i=1}^n{(x_i-\overline{x})}^2}}$$

(2)

where $x_i$ is the observed value, $\overline{x}$ is the mean value and $w_{ij}$ denotes the element corresponding to the spatial weight matrix $W_i$. We employ four spatial weight matrixes in this study.$W_1$ set the sending province and the arriving province geographically adjacent to 1, otherwise 0. $W_2$ expressed using the reciprocal of the distance calculated by the latitude and longitude of each region. $W_3$, a larger economic gap between regions may cause more trade transactions, using the absolute value of the difference in per capita GDP between regions. The range of Moran’s I index is [−1, 1]. When the index < 0, it denotes negative spatial correlation; when it is equal to 0, it stands for no correlation; otherwise, it denotes positive spatial correlation.

Secondly, we refer to the method to set the multidimensional bilateral spatial weight matrix. $W_0$ is the original spatial weight matrix based on 29 provinces and cities, which is an n × n square matrix. The spatial weight matrix adjacent to the sending province is $W_i=W_0\otimes I_n$, where $I_n$ is a unit matrix of order n, then $W_i$ is N × N, N = n × n, i = 1, 2, 3. This study does not consider the interprovince trade transactions, so the diagonal elements are denoted as zero. The unity of comprehensive analysis, in this paper, only the spatial weight matrix of sending province is considered and based on the above three matrices for validation.

This study uses Spatial Durbin Methods to explore the spatial effect of DT and IRT. The model (3) used in this study are as follows:

$$\ln IR{T}_{ijt}={\alpha }_0+\rho W*\ln IR{T}_{ijt}+{\alpha }_{1}D{T}_{it}+\delta W*D{T}_{it}+{\alpha }_2{\displaystyle \sum X_{ijt}}+{\alpha }_{3}W{\displaystyle \sum X_{ijt}}+{\theta }_i+{\theta }_j+{\mu }_t+{\epsilon }_{ijt}$$

(3)

where W is the spatial weight matrix and $\rho$ measures the effect of represents the spatial autoregressive coefficient, and ${\alpha }_1$ is the spatial interaction terms of the digital economy.

3.1.3. Mediating Effect Model

In order to verify the possible channel mechanisms of the digital economy on inter-regional trade, according to the previous mechanism analysis, the mediation model (4) and (5) is set as follows:

$$M_{it}={\alpha }_0+{\alpha }_{1}D{T}_{it}+{\alpha }_2{\displaystyle \sum X_{ijt}+{\theta }_i+{\theta }_j+{\mu }_t+{\epsilon }_{ijt}}$$

(4)

$$\ln IR{T}_{ijt}={\alpha }_0+{\alpha }_{1}D{T}_{it}+\phi M_{it}+{\alpha }_2{\displaystyle \sum X_{ijt}+{\theta }_i+{\theta }_j+{\mu }_t+{\epsilon }_{ijt}}$$

(5)

where $M_{it}$ is the mediating variable, and other parts are the same as Formula (1).

3.2. Variable Description

In this section, dependent variables, independent variables, mediation variables, threshold variables and control variables are explained separately.

3.2.1. Dependent Variable

Inter-regional trade (IRT). We follow the research in estimating China’s inter-regional trade outflows and trade inflows using the gravity model plus the transport volume distribution coefficient method [39]. China publishes data on the freight exchange volume of national railway between administration regions from 1986 to the present, which are continuous, long-term, bilateral cargo traffic between regions at the provincial level. The data are counted in terms of freight volume, which is able to exclude price changes and therefore, to a certain extent, objectively reflect inter-regional trade exchanges. The specific equation is:

$$\ln IR{T}_{ijt}=Q_{ij}\cdot (s_{ij}{d}_j/{\displaystyle \sum s_i}) , Q_{ij}=H_{ij}{H}_{oo}/H_{io}{H}_{oj}$$

(6)

where $s_{ij}$ denotes the total supply of goods and services in province i (j), which is equal to GDP minus net exports; and $d_j$ denotes the total demand for goods and services in province j (i), which is equal to GDP minus net outflows. ${\displaystyle \sum s_i}={\displaystyle \sum d_j}$ denotes the sum of supply or equal demand in each region; $Q_{ij}$ denotes the trade friction coefficient of products and services between the two regions. $H_{ij}$ denotes the amount of goods sent from province i to province j, $H_{oo}$ denotes the total amount of goods sent from all provinces (equal to the total amount of arrivals), $H_{io}$ denotes the total amount of goods sent from province i, and $H_{oj}$ denotes the total amount of goods arrived from province j. However, due to the low proportion of rail freight volume to total freight volume in China, the average value of the proportion in this sample from 2006–2017 is about 8.7–14.7%, for which this paper draws on existing studies to multiply the inter-regional freight volume by the ratio coefficient of total freight volume to rail transport volume, in order to present a more comprehensive picture of the realistic characteristics of freight exchange volume.

3.2.2. Independent Variable

Digital economy (DT). Combining the existing literature and connotation understanding, this paper constructs the index system of the digital economy from three dimensions: carrier facilities, application forms and development support. Carrier facilities include fiber optic cable density, per capita cell phone capacity and per capita Internet access ports. Application forms include Internet penetration rate, cell phone penetration rate and per capita telecommunication service volume. Development support includes the ratio of information technology employment, the ratio of information technology fixed asset investment, and the ratio of information technology industry output. The indicators are downscaled and de-priced through ratio measurement, and the entropy method is applied to measure them.

3.2.3. Mediation Variables

(1)

Trade cost effect (TC). This paper used the Head–Ries trade cost index [40], which inferred the trade cost formed in all trade included based on the proportion of inter-regional trade. Specifically, the equation is ${\overline{\tau }}_{ij}=\sqrt{{\tau }_{ij}{\tau }_{ji}/{\tau }_{ii}{\tau }_{jj}}={\left(IR{T}_{ij}\cdot IR{T}_{ji}/IR{T}_{ii}\cdot IR{T}_{jj}\right)}^{1/2\theta }$. Where ${\tau }_{ij}$ and ${\tau }_{ji}$ are the trade cost between province i and province j, ${\tau }_{ii}$ and ${\tau }_{jj}$ are the inter-provincial trade costs of provinces i and j, respectively. $\theta$ represents the trade cost elasticity coefficient, which is set to 4 in this paper.

(2)

Market demand effect (MD). This paper used the ratio of per capita consumer spending in this province to other provinces to measure the relative market demand size in order to verify the mediating effect of market demand [41].

(3)

Resource allocation effect (RA). We analyzed capital allocation and labor allocation from two perspectives and measured the capital resource mismatch indices (Misk) and labor resource mismatch indices (Misl) [42]. The method uses a variable coefficient model based on the LSDV method to regress the C-D production function with constant returns to scale to obtain the capital-output elasticity of each region and then calculate the deviation between the actual use factors and the effective allocation factors and obtain the mismatch indices of capital and labor based on the deviation. Since there are two kinds of resource mismatch: under-allocation and over-allocation, the absolute value is taken for the resource mismatch index. In this paper, the output is the real GDP of each region with 2006 as the base period, the capital is measured by the stock of fixed assets in each province using the perpetual inventory method [43], and the labor is the number of urban units employed in each province.

(4)

Technological innovation efficiency (TI). This paper uses the commonly used measure of innovation efficiency to verify the path of the role played by innovation capacity in the digital economy to promote domestic trade. It is estimated using the DEA-Mamlquist index, with input variables being R&D inputs and full-time equivalents, and outputs being new product revenues and patent applications granted.

3.2.4. Control Variables

Considering other endogenous variables affecting inter-regional trade, this paper selects regional economic development level (rgdp), which reflects regional demand capacity for products and services while examining economic scale, using GDP per capita deflated by 2006 as the base period and expressed in logarithm. The industrial structure (indust), which reflects the comparative advantage of each region and is an important factor affecting regional trade flows, is expressed as the ratio of the value added of the tertiary sector to the value added of the secondary sector. The level of local protection (soej), which is used to examine the level of local protection in arriving provinces, i.e., the government restricts the inflow of products from other provinces to protect the supply of local enterprises, is measured by the ratio of the output value of state-owned enterprises to the GDP of each province. The level of international trade (nx), which takes into account that the level of international trade may have a substitution effect on the domestic trade cycle, is controlled using the net export to GDP ratio of each region. The level of external openness (fdi) is measured using the logarithm of the actual use of foreign investment balances. Geographical distance (dist), as the key factor in the gravity model that can impede trade flows, is expressed using the distance calculated by the latitude and longitude of each region. Contiguity (adj), the contiguity of regions facilitates the development of bilateral trade, setting the geographic contiguity of sending and arriving provinces to 1 and 0 otherwise.

3.3. Data Sources

Considering the availability of data and the consistency of the time horizon, this paper examines 29 Chinese provinces (excluding Tibet, Hong Kong, Macau and Taiwan and Hainan, whose provincial rail freight volumes are basically very small). The data relate to bilateral asymmetric trade between regions, and a total of 812 study subjects were obtained. The freight exchange volume of the national railway between administration regions is obtained from the China Traffic Yearbook. The values of other variables were obtained from the China Statistical Yearbook and the National Bureau of Statistics.

4. Empirical Results and Discussion

4.1. Characteristic Portrayal

In order to explore the spatial evolution characteristics of the digital economy and inter-regional trade, this paper draws the spatio-temporal maps in 2006 and 2017 based on ARCGIS 10.2 software, which is shown in Figure 2. It can be found that in 2006 and 2017, the regions with higher levels of digital economy development are concentrated in Jiangsu Province, Beijing and Guangdong Province, while the western region is lower overall. The graph on the left side shows that in 2006, in terms of trade outflows, the top three provinces were Jiangsu Province, Guangdong Province and Shandong Province, which together accounted for about 28% of the total outflows and were concentrated in the eastern coastal region, while the three provinces with the lowest percentage were concentrated in the northwest region, whose sum was less than 2%. In terms of trade inflows, the top three provinces are Guangdong Province, Jiangsu Province and Henan Province, together accounting for 27% of the total outflows, distributed in the eastern coastal and inland regions, while the smallest share is still in the northwest, whose sum is still less than 2%. Interestingly, in the 2017 graph on the right, Jiangsu, Shanghai and Shandong provinces remain the provinces with the highest trade outflows, accounting for 30% of total outflows, while Henan, Guangdong and Jiangsu provinces are the provinces with the highest trade inflows, accounting for about 25%. However, the three provinces with the least share of outflows and outflows are still very small, but inflows have increased. An overall comparative analysis of inter-regional trade flows in 2006 and 2017 shows that a significant increase occurred in 2017. The yellow curve and light green curve in the left panel are converted to orange in the right panel, as well as the orange line in the left panel is sparser while the right panel has largely spread across the whole. Meanwhile, the purple curve on the left is less and concentrated in the eastern region, while the purple curve on the right has shown the overall distribution characteristics. Finally, the blue curve in the left panel is the highest level, while in the right panel, a network layout has been formed in the southeast, and the maximum value is about five times higher than in 2006.

The comprehensive analysis finds that the developed provinces have strong inter-regional trade circulation momentum, while the northwest region has limited inter-regional trade circulation capacity. The possible explanations are with the advantages of open policy and geographical location, the eastern coastal region has stronger production capacity and absorbs domestic and foreign production factor inputs. It uses the western region as an important raw material guarantee and the central region to provide basic raw materials, prompting inputs from other provinces and thus promoting trade inflows. Further, the eastern coastal region exports its products and services to other regions as intermediate or final products, thus promoting trade outflow. In turn, the trade cycle dynamics are enhanced. However, the overall lower level of industrial development in the western region, the weaker ability to connect with the eastern region in the industrial chain and supply chain, as well as the insufficient technological endogenous power of its own development, make the capacity of trade outflow and inflow limited. In addition, from the perspective of net outflow and net inflow, the central trade exchanges have gained momentum. Anhui Province, as a province in the central region, is close to the net outflow of developed eastern provinces, showing a strong trade outflow; at the same time, from the net inflow, the eastern coastal regions such as Guangdong Province, Zhejiang Province and Fujian Province transferred to the central regions such as Henan Province, Shaanxi Province and Shanxi Province from 2006 to 2017, which to a certain extent indicates the enhanced momentum of trade trends among the central regions.

4.2. Benchmark Testing

According to the research idea and model setting of this study, the regression results of the empirical tests are given in

Table 1. Baseline regression, robustness and endogeneity analysis regression results.

	OLS		FE		Lag Effect		IV-2SLS
	Outflow	Inflow	Outflow	Inflow	Outflow	Inflow	Outflow	Inflow
DT	1.922 ***	1.387 ***	1.279 ***	0.995 ***			2.870 ***	2.397 *
	[0.0851]	[0.0897]	[0.3493]	[0.3519]			[1.1101]	[1.2425]
L.DT					1.849 ***	1.123 ***
					[0.3978]	[0.4007]
rgdp			1.696 ***	1.580 ***	1.413 ***	1.682 ***	2.667 ***	2.910 ***
			[0.2492]	[0.2511]	[0.2897]	[0.2918]	[0.2754]	[0.3044]
indust			0.146	0.211 **	0.0488	0.154	0.379 ***	0.411 ***
			[0.0927]	[0.0934]	[0.1043]	[0.1051]	[0.0654]	[0.0699]
soej			−0.299 ***	−0.257 ***	−0.331 ***	−0.246 ***	−0.498 ***	−0.468 ***
			[0.0792]	[0.0798]	[0.0895]	[0.0902]	[0.0199]	[0.0211]
fdi			0.073 ***	−0.015	0.090 ***	−0.0172	0.142 ***	0.0619
			[0.0270]	[0.0272]	[0.0298]	[0.0300]	[0.0344]	[0.0373]
nv			−0.569 ***	−0.405 **	−0.723 **	−0.307	−1.585 ***	−2.209 ***
			[0.1942]	[0.1957]	[0.2281]	[0.2298]	[0.0819]	[0.0904]
dist			−1.390 ***	−1.337 ***	−1.394 ***	−1.340 ***	−1.055 ***	−1.136 ***
			[0.0312]	[0.0314]	[0.0330]	[0.0332]	[0.0344]	[0.0353]
adj			0.450 ***	0.510 ***	0.459 ***	0.519 ***	6.948 ***	7.541 ***
			[0.0426]	[0.0430]	[0.0451]	[0.0454]	[0.6344]	[0.6933]
cons	4.144 ***	4.277 ***	12.82 ***	12.91 ***	12.97 ***	12.94 ***	6.948 ***	7.541 ***
	[0.0281]	[0.0288]	[0.3567]	[0.3594]	[0.3937]	[0.3966]	[0.6344]	[0.6933]
R2	0.0413	0.0215	0.6604	0.6552	0.6507	0.6456	0.5433	0.4944
Kleibergen-Paap rk LM statistic							503.514 (0.000)	502.514 (0.000)
Kleibergen-Paap rk Wald F statistic							767.437 (16.38)	767.437 (16.38)
Sending	No	No	Yes	Yes	Yes	Yes	Yes	Yes
Arriving	No	No	Yes	Yes	Yes	Yes	Yes	Yes
Time	No	No	Yes	Yes	Yes	Yes	Yes	Yes
Observations	9744	9744	9744	9744	8932	8932	9744	9744

Note: ***, **, * denote significance at the 1%, 5% and 10% levels, respectively, and within [ ] are standard errors.

. The ordinary OLS estimation results show the baseline effects of the digital economy on inter-regional trade outflows and inflows. The fixed effects model FE incorporates a series of control variables as well as setting dummy variables for sending province, arriving province and time for high-dimensional fixed effects estimation. The results find that the digital economy significantly contributes to inter-regional trade outflows and inflows. In addition, the coefficient of the impact of the digital economy on trade outflow is always larger than that of inflow, which may be due to the following reasons: on the one hand, China’s trade outflow is larger than trade inflow, the net trade outflow regions have stronger product competition and larger scale markets, and other regions have strong market demand for them, and the digital economy has a greater impact effect on trade outflow through inter-temporal and inter-regional advantages. On the other hand, the less developed regions set higher trade barriers to protect the market competitiveness of local products and services, resulting in a smaller impact of the digital economy on trade inflows.

From the control variables, the coefficient of the level of economic development (rgdp) is significantly positive, indicating that the level of local economic development significantly promotes local trade with foreign trade, effectively promoting the inter-regional trade cycle. The positive coefficient of industrial structure (indust) indicates that the upgrading level of industrial structure in the sending province promotes local trade outflows and inflows; however, the coefficient on trade outflows is not significant in the fixed effects model. The possible reason is that the higher level of industry in the sending province is more likely to promote trade outflows, while the higher level of tertiary industry in the arriving province contributes to increased trade inflows. The coefficient on the level of local protection (soej) is significantly negative, indicating that higher levels of local protection in other provinces inhibit inter-regional trade. The possible explanation is that state-owned enterprises in the arrival province provide more local products and services that satisfy most of the demand, thus reducing trade outflows from the sending province as well as trade inflows to the arrival province. The coefficient of external openness level (fdi) is significantly positive only for trade outflows, indicating that FDI inflows contribute to the increase of productive capital in the sending province and thus to the regional production level, while technological innovation spillovers enhance regional productivity and promote local trade outflows, while the impact on trade inflows to the local area from the outside is hardly significant. The coefficient of international trade level (nx) is significantly negative, indicating that regional net export trade substitutes inter-regional trade and suppresses domestic trade outflows and inflows. Geographical distance (dist) has a significant negative effect on domestic trade, as well as regional proximity (adj) favors inter-regional trade, consistent with the presupposition of this study.

4.3. Robustness and Endogeneity Analysis

4.3.1. Independent Variables Suitability

Considering that the independent variables are comprehensive indicators and their weight distribution affects the final measurement results, this paper measures the indicators according to the secondary classification. The impact of carrier facilities (CA), application forms (AF) and development support (DS) on inter-regional trade is tested. The results are shown in

Table 2. Impact of digital economy disaggregated indicators on domestic trade cycle.

	Outflow	Lag	Inflow	Lag		Outflow	Inflow		Outflow	Inflow
CA	0.172	0.659 ***	0.238	0.648 ***	AF	1.186 ***	1.240 ***	DS	1.043 ***	0.722 ***
	[0.1666]	[0.2025]	[0.1677]	[0.2038]		[0.4005]	[0.4029]		[0.1818]	[0.1831]
Sending	Yes	Yes	Yes	Yes		Yes	Yes		Yes	Yes
Arriving	Yes	Yes	Yes	Yes		Yes	Yes		Yes	Yes
Time	Yes	Yes	Yes	Yes		Yes	Yes		Yes	Yes
R2	0.6571	0.6481	0.6529	0.6438		0.6574	0.6532		0.6582	0.6534
Observations	9744	9744	9744	9744		9744	9744		9744	9744

Note: *** denotes significance at the 1% level, and within [ ] are standard errors.

.

The coefficient of the effect of carrier facilities of the digital economy on inter-regional trade is found to be positive but insignificant. The possible reason is that there is a lag in the impact of digital infrastructure on economic activities, and this paper takes its lagged term and finds that carrier facilities can promote inter-regional trade outflows and inflows at the 1% significance level. The application form and development support promote inter-regional trade circulation at 1% level of significance. These two indicators are mainly based on the technology and platform applications of the digital economy, which to a certain extent can illustrate the robustness of the explanatory variables in this paper. It is also found that the application form of the digital economy has a slightly greater impact on the domestic trade cycle than other forms, indicating that the application of network technology in the digital economy can promote inter-regional trade faster.

4.3.2. Lagged Effect of Independent Variables

In order to clarify the two-way influence relationship between the digital economy and inter-regional trade, this paper estimated the inter-regional trade by lagging the digital economy by one period (L.DT), and the results are shown in Table 1 for the lag effect. It is found that L.DT significantly promotes inter-regional trade outflow and inflow, and the promotion effect is greater than the effect generated by the digital economy in the current period, and the possible explanation is that there is a lagging effect of infrastructure construction in the digital economy.

4.3.3. Endogeneity Test

Considering that the setting of comprehensive indicators of digital economy development may also include other indicators with possible correlation with inter-regional trade, in order to further enhance the reliability of the analysis, this paper uses the penetration rate of the telephone in 1984 in each region as an instrumental variable for the level of digital economy [44]. The digital economy has continuity, and regions with historically high phone penetration rates will promote the development of the digital economy as the technology level improves, and the usage rate of phones in earlier years has declined significantly with the progress of the times, and the impact on inter-regional trade is exclusive. In order to make this variable satisfy the dynamic characteristics, this paper uses the interaction term between the telephone penetration rate of each region in 1984 and the social fixed asset investment in the information transmission computer services and software industry in the previous year as the instrumental variable of the digital economy [45] and applies the 2SLS method to estimate the model, and the estimation results are shown in Table 1 for the two-stage instrumental variable (IV-2SLS) model, yielding effects of instrumental variables on inter-regional trade outflows and inflows that are consistent with the benchmark results, as well as LM statistics and F statistics that pass the feasibility test.

4.4. Spatial Effect Test

Figure 3 shows the local Moran indices of inter-regional trade outflows and inflows in 2006 and 2017, both significant at the 1% level of significance, with positive and largely between 0.4 and 0.6. The correlation is at a high level, indicating that inter-regional trade exhibits a “high-low” agglomeration. In 2006, it can be found that inter-regional trade outflow and inflow agglomeration are relatively dispersed, and high-low agglomeration areas are relatively evenly distributed. While, in 2017, the overall level of agglomeration increased, shifted to the upper right, and became more concentrated, indicating that the level of inter-regional trade in China has been increasing, evolving from a scattered “high-low” to a concentrated “higher-higher” agglomeration.

Table 3. Spatial effect tests for different spatial matrices.

	W1		W2		W3
	Outflow	Inflow	Outflow	Inflow	Outflow	Inflow
DT	0.821 ***	0.349 *	1.042 ***	0.816 ***	1.224 ***	0.960 ***
	[0.1955]	[0.1909]	[0.1929]	[0.1911]	[0.2026]	[0.2048]
rgdp	1.539 ***	1.263 ***	1.602 ***	1.366 ***	1.581 ***	1.441 ***
	[0.1366]	[0.1335]	[0.1377]	[0.1365]	[0.1450]	[0.1465]
indust	0.184 ***	0.325 ***	0.231 ***	0.338 ***	0.151 ***	0.250 ***
	[0.0509]	[0.0497]	[0.0513]	[0.0509]	[0.0552]	[0.0558]
soej	−0.220 ***	−0.165 ***	−0.180 ***	−0.140 ***	−0.252 ***	−0.234 ***
	[0.0434]	[0.0423]	[0.0437]	[0.0433]	[0.0459]	[0.0463]
fdi	0.0701 ***	−0.0055	0.0779 ***	0.0093	0.0718 ***	−0.0251
	[0.0148]	[0.0144]	[0.0151]	[0.0150]	[0.0157]	[0.0159]
nx	−0.442 ***	−0.322 **	−0.234 **	−0.187 *	−0.482 ***	−0.389 ***
	[0.1066]	[0.1040]	[0.1078]	[0.1067]	[0.1132]	[0.1143]
dist	−1.107 ***	−1.016 ***			−1.359 ***	−1.299 ***
	[0.0580]	[0.0553]			[0.0909]	[0.0915]
adj			1.612 ***	1.606 ***	0.513 ***	0.579 ***
			[0.0897]	[0.0874]	[0.1250]	[0.1258]
cons	8.223 ***	7.631 ***	−0.635	−1.754 ***	11.67 ***	11.64 ***
	[0.5575]	[0.5355]	[0.4613]	[0.4555]	[0.7591]	[0.7642]
W*DT	1.360 ***	2.312 ***	3.291 ***	8.788 ***	−0.135	1.004 **
	[0.3389]	[0.3308]	[0.9833]	[0.9749]	[0.3692]	[0.3738]
ρ	0.398 ***	0.456 ***	0.593 ***	0.633 ***	0.215 ***	0.179 ***
	[0.0112]	[0.0106]	[0.0156]	[0.0150]	[0.0139]	[0.0143]
R2	0.6535	0.6405	0.5814	0.5646	0.6522	0.6447
Sending	Yes	Yes	Yes	Yes	Yes	Yes
Arriving	Yes	Yes	Yes	Yes	Yes	Yes
Time	Yes	Yes	Yes	Yes	Yes	Yes
Observations	9744	9744	9744	9744	9744	9744

Note: ***, **, * denote significance at the 1%, 5% and 10% levels, respectively, and within [ ] are standard errors.

reports the spatial effects of the digital economy on inter-regional trade. It is found that the digital economy promotes inter-regional trade circulation at all three spatial matrices at the 1% significance level, and $\rho$-values are significant at the 1% significance level, further indicating the existence of spatial relevance of inter-regional trade. Consistent with the benchmark model, the coefficient of the impact of the digital economy on inter-regional trade outflow is always greater than that of trade inflow. W*DT indicates the impact of the local digital economy level on trade outflow and inflow of neighboring places, and it is found that the improvement of the local digital economy level significantly promotes inter-regional trade circulation in neighboring places and surrounding areas, and there is a positive spatial spillover effect. This proves that the local digital economy relies on digital technology to enable production factors and information technology to spread and share across space and time, generating a “spatial spillover effect” to promote local trade outflows and trade inflows from other places. At the same time, the digital economy can effectively integrate inter-regional resources and enhance regional production capacity, thus promoting the inter-regional trade cycle in space.

5. Further Analysis

5.1. Channel Mechanisms Testing

According to the theoretical hypothesis, the digital economy mainly influences the inter-regional trade cycle by reducing trade costs, promoting market demand, enhancing resource allocation efficiency and promoting technological innovation efficiency. In this paper, H1–H4 is tested by a high-dimensional fixed-effects model, and the results are shown in

Table 4. Channel mechanism test of digital economy affecting inter-regional trade.

	TC	Outflow	Inflow	MD	Outflow	Inflow	TI	Outflow	Inflow
DT	−2.738 ***	0.355	0.0622	0.238 **	1.194 ***	0.894 **	3.238 ***	1.032 **	0.903 **
	[0.6493]	[0.2723]	[0.2740]	[0.0950]	[0.3478]	[0.3497]	[0.1175]	[0.3626]	[0.3655]
M		−0.337 ***	−0.341 ***		0.358 ***	0.426 ***		0.076 **	0.0284
		[0.0043]	[0.0043]		[0.0372]	[0.0374]		[0.0302]	[0.0305]
R2	0.3750	0.7940	0.7914	0.8687	0.6635	0.6597	0.7211	0.6606	0.6552
N	9744	9744	9744	9744	9744	9744	9744	9744	9744
	Misk	Outflow	Inflow	Misl		Outflow	Inflow
DT	−0.754 ***	1.522 ***	1.062 ***	0.556 ***	0.359 ***	1.538 ***	1.228 ***
	[0.0324]	[0.3588]	[0.3617]	[0.0277]	[0.0479]	[0.3562]	[0.3590]
DT-sq					−0.338 ***
					[0.0585]
M		0.322 ***	0.088			−0.466 ***	−0.419 ***
		[0.1097]	[0.1106]			[0.1281]	[0.1291]
R2	0.6995	0.6606	0.6551	0.9576	0.8425	0.6608	0.6555
N	9744	9744	9744	9744	9744	9744	9744

Note: *** and ** denote significance at the 1% and 5% levels, respectively, and within [ ] are standard errors.

.

According to Table 4, (i) the regression coefficient of DT on trade cost (TC) is negative at the 1% significance level, which can effectively reduce trade costs. At the same time, TC shows significant suppression of both trade outflows and inflows, explaining 70–80% of the mediating effect, which proves that the digital economy can significantly reduce trade costs and thus promote inter-regional trade, which is consistent with the research H1 of this paper. (ii) The regression coefficient of DT on market demand effect (MD) is positive at a 1% significance level and can significantly increase market demand, and MD can significantly increase local trade outflows and inflows, which can explain 6–9% of the mediating effect, which indicates that digital economy can promote domestic trade cycle by increasing market demand, which is consistent with H2 of this paper. (iii) The regression coefficient of DT on technological innovation efficiency (TI) is positive at a 1% significance level, which can significantly enhance the efficiency of technological innovation, and TI can promote inter-regional trade outflow, while the effect on trade inflow is insignificant. The possible explanation is that the level of technological innovation in China has been increasing, but there is still a large gap compared to developed countries, especially the less developed regions that have a weak technological innovation capacity and a strong demand for technological innovation, knowledge exchange and highly skilled personnel. With the increasing level of development of the digital economy, it can expand the scope of technology diffusion and exchange, share and promote inter-regional trade outflow; however, the effect of technological innovation efficiency improvement on trade inflow is very small probably because most regions are net inflow of technology, and the existence of time lag in technology transformation makes the current effect not obvious.

(iv) DT can reduce the capital resource mismatch indices (Misk) at a 1% significance level, i.e., improve the efficiency of capital resource allocation, but Misk promotes inter-regional trade outflow, as well as the effect on trade inflow is not significant. The possible explanation is that China’s regional capital resource allocation is continuously optimized, and the digital economy relies on emerging technologies such as big data and cloud computing to effectively improve capital mismatch, which allows capital to be transferred to undercapitalized production sectors and continuously weakens the sectoral integration boundaries thereby enhancing the capital allocation effect. However, China’s capital allocation efficiency is still low and financing is difficult and expensive, making the lack of investment still insufficient, there is also excessive investment, waste serious “hot and cold coexistence” structure. The overall positive effect of capital allocation efficiency on inter-regional trade is greater than the negative effect of over-investment, thus promoting trade outflows. DT can promote the labor resource mismatch indices (Misl) and thus inter-regional trade cycle at a 1% significance level. The inclusion of the squared term DT-sq of the digital economy in the regression equation shows a non-linear change of promotion followed by suppression, indicating that as the level of the digital economy increases further it can effectively mitigate labor resource mismatch. Possible explanations are with the increasing level of development of the digital economy, there is an obvious mismatch of labor resources in China, and the dual economy of urban and rural areas as well as the household registration restrictions of laborers seriously hinder the flow of labor factors. In the early stage of digital economy development, regions actively attracted low-skilled and high-skilled labor to high-paid regions, which aggravated the mismatch of labor resources. However, with the continuous improvement of digital infrastructure, application forms and digital technology, the development of the digital economy can break the spatial and temporal constraints and release more sectoral jobs, which can lead labor from over-allocated regions to under-allocated regions and thus optimize labor resource allocation. Further, the regression coefficients of Misl on inter-regional trade outflows and inflows are significant at the 1% significance level, indicating that the digital economy can promote domestic trade circulation by improving labor resource mismatch, which is consistent with the research hypothesis of this paper.

5.2. Heterogeneity Test

5.2.1. Geographical Location Heterogeneity

Considering that regional heterogeneity is mainly reflected in the heterogeneity caused by geographical distribution and economic development differences, this paper includes the trade cost threshold variable to investigate the regional heterogeneity of the digital economy on inter-regional trade. The threshold regression model (7) is set as:

$$\ln IR{T}_{ijt}={\alpha }_0+{\alpha }_{1}D{T}_{it}I(T{C}_{it}\le \gamma )+{\alpha }_{2}D{T}_{it}I(T{C}_{it}>\gamma )+{\alpha }_3{\displaystyle \sum X_{it}}+{\theta }_i+{\theta }_j+{\mu }_t+{\epsilon }_{ijt}$$

(7)

where I(*) represents the indicator function and TC is the threshold variable. $\gamma$ represents the threshold. The analysis is based on coastal, inland and border areas as well as eastern, central and western.

According to

Table 5. Regression results of regional heterogeneity.

	Coastal	Inland	Border	East	Central	West
Outflow	0.909 ***	4.279 ***	1.171 ***	1.038 ***	1.291 ***	4.624 ***
	[0.2987]	[0.3701]	[0.4069]	[0.2893]	[0.4764]	[0.3637]
	−2.229 ***	1.431 ***	−6.612 ***	−2.275 ***	−2.368 ***	0.993 ***
	[0.3659]	[0.4141]	[0.6281]	[0.3466]	[0.4927]	[0.3876]
Threshold	8.706	5.801	8.968	8.668	4.217	4.301
F-statistic	97.27	272.9	118.07	97.27	151.67	227.34
p-value	0.000	0.000	0.000	0.000	0.000	0.000
	Coastal	Inland	Border	East	Central	West
Inflow	−0.178	4.046 ***	0.839 **	−0.007	0.024	4.535 ***
	[0.3037]	[0.3607]	[0.3785]	[0.2992]	[0.4640]	[0.3470]
	−3.271 ***	1.889 ***	−9.725 ***	−2.707 ***	−7.234 ***	0.515
	[0.3836]	[0.3965]	[0.5684]	[0.3584]	[0.6473]	[0.3643]
Threshold	9.258	4.267	8.819	8.668	7.485	4.213
F-statistic	61.59	282.43	214.92	71.37	116.68	328.53
p-value	0.000	0.000	0.000	0.000	0.000	0.000
N	3024	4704	2016	3360	2688	3696

Note: Coastal areas include Liaoning Province, Hebei Province, Tianjin City, Shandong Province, Jiangsu Province, Shanghai City, Zhejiang Province, Fujian Province and Guangdong Province and other 9 areas; inland areas include Beijing, Henan Province, Hunan Province, Hubei Province, Sichuan Province, Guizhou Province, Shanxi Province, Shaanxi Province, Chongqing Municipality, Anhui Province, Jiangxi Province, Gansu Province, Qinghai Province, Ningxia Hui Autonomous Region and other 14 areas; Border areas include Xinjiang Uygur Autonomous Region, Yunnan Province, Inner Mongolia Autonomous Region, Jilin Province, Heilongjiang Province, Guangxi Zhuang Autonomous Region and other six regions. The difference between the coastal region and the eastern region is that the eastern region also includes Beijing. *** and ** denote significance at the 1% and 5% levels, respectively, and within [ ] are standard errors.

, it turns out that the digital economy shows significant differences in inter-regional trade in different regions, and there is a threshold effect of trade costs with large differences. Clearly, for inland and western regions, the regression coefficients of the digital economy on outflows and inflows (4.279, 4.624) are much larger than those for other regions (around 1), indicating that the digital economy has a more prominent and potential role in promoting inter-regional trade in less developed or non-border regions. The trade cost thresholds are relatively low (5.801, 4.301), and the digital economy still has a facilitating effect on inter-regional trade when the thresholds are exceeded. Although the regression coefficients are smaller but close to the overall level, indicating that the impact of the digital economy on inter-regional trade in the inland and western regions is not bounded by trade costs and can significantly promote outflows and inflows.

In contrast, for coastal, border, eastern and central regions, the impact of the digital economy on inter-regional trade is constrained by a significant threshold of trade costs. DT significantly promotes IRT when it is below the threshold and significantly inhibits it when it is above. The mean value of the total sample of trade costs in this study is 3.729, and the corresponding values for the 50%, 75%, 90%, 95% and 99% quartiles are 3.328, 4.512, 5.886, 6.881 and 10.498, respectively. for these regions, the digital economy inhibits outflows and inflows only when trade costs reach high levels above the 95% quartile, which laterally confirms the core conclusion that the digital economy can significantly boost inter-regional trade. The cost threshold of 4.217 for trade outflows in the central region is lower than the cost threshold of 7.485 for inflows, probably because the eastern region is mainly net trade outflows and the western region is mainly net trade inflows, while the trade outflows in the central region are between the eastern and western regions and geographically they are at the important hub of bearing the east and west, so the trade inflows are less constrained by the costs, while the trade outflows are constrained when the trade costs are at the 75% quantile level or higher above the 75th percentile level.

5.2.2. Industry-Intensive Heterogeneity

There are large differences in factor endowments between regions in China, and the development of the digital economy has a differentiated impact on inter-regional trade in different factor-intensive regions. The manufacturing industry is classified according to the factor intensity criterion and divided into four categories according to the two digits of the National Economic Classification and Code 2011: labor-intensive, capital-intensive, technology-intensive and resource-intensive [46]. Then, the location quotient method is applied to measure the scale advantage of each industry, and the average value of the location quotient of each industry from 2006–2017 is used to define The industry-intensive category of the region, the specific equation is $I_s=\frac{1}{n}{\displaystyle \sum _{s=1}^n\frac{L_{ijt}/L_{it}}{L_{jt}/L_t}}$, where $L_{ijt}$ is the average annual employment in industry j in region i in year t, $L_{it}$ is the average annual employment in all industries in region i in year t, $L_{jt}$ is the national average annual employment in industry j in year t and $L_t$ is the national average annual employment in year t and s is the number of intensive industries. Data from the China Industrial Economic Statistical Yearbook. The results are shown in

Table 6. Regression results for intensive industry heterogeneity.

	Outflow				Inflow
	Labor	Capital	Technology	Resources	Labor	Capital	Technology	Resources
DT	2.178 ***	2.292 ***	1.502 ***	0.997 ***	2.385 ***	1.628 ***	0.054	0.967 ***
	[0.6007]	[0.3979]	[0.4297]	[0.3655]	[0.6125]	[0.4108]	[0.4333]	[0.3640]
R2	0.6822	0.6362	0.6454	0.6574	0.6807	0.6388	0.6456	0.6459
N	3360	5376	3360	6384	3360	5376	3360	6384

Note: Labor-intensive: processing of food from agricultural products; manufacture of foods; manufacture of liquor; beverages and refined tea; manufacture of textile; manufacture of textile, wearing apparel and accessories; manufacture of leather, fur, feather and related products and footwear; manufacture of paper and paper products; printing and reproduction of recording media; manufacture of articles for culture, education, arts and crafts sport and entertainment activities. Capital-intensive: manufacture of tobacco; processing of timber, manufacture of wood, bamboo, rattan, palm and straw products; manufacture of chemical fibres; smelting and pressing of ferrous metals; smelting and pressing of non-ferrous metals. Technology-intensive: manufacture of raw chemical materials and chemical products; manufacture of medicines; manufacture of general purpose machinery; manufacture of special purpose machinery; manufacture of transport equipments; manufacture of electrical machinery and apparatus; manufacture of computers, communication and other electronic equipment; manufacture of measuring instruments and machinery other manufacture. Resource-intensive: processing of petroleum, coal and other fuels; manufacture of rubber and plastics products; manufacture of non-metallic mineral products; manufacture of metal products. *** denotes significance at the 1% level, and within [ ] are standard errors.

.

The first four columns are estimated for trade outflows and the study finds that DT significantly boosts trade outflows from labor, capital, technology and resource regions at a 1% significance level. The last four columns show that DT boosts trade inflows to labor, capital and resource regions at a 1% significance level, while it has almost no effect on trade inflows to technology-intensive regions. Specifically, (i) DT has a larger effect on trade outflows and inflows from labor-intensive regions (2.178, 2.385), which is also consistent with the realistic development of China. China is the core export market for labor-intensive manufacturing products and also the world’s larger market for imports of labor-intensive manufacturing products, which has a broad market space for labor-intensive goods and services, and digital economy enhancement helps to reduce the trade costs of goods and services, enhance market competitiveness and thus promote inter-regional trade cycle. (ii) the effect of DT on trade outflow from capital-intensive regions is greater than the effect on trade inflow, and the possible explanation is that capital-intensive regions are an important basis for the transformation of China’s industrial economy from crude to intensive, and the in-depth development of digital economy promotes trade outflow from capital-intensive regions to other regions through the smooth dissemination of data, information and other resources. However, because capital-intensive industries require strong capital support and the ability to focus on technological innovation, making the digital economy prompted by the impact of trade inflows effect is smaller. (iii) the effect of DT on trade outflows from technology-intensive regions is significant at the 1% significance level, while the effect on trade inflows is small and insignificant, which is consistent with the findings of H4. A possible explanation is the higher level of economic development in the technology-intensive regions of the country, which has a strong regional agglomeration effect. The digital economy has contributed to a weaker propagation effect from less technology-intensive regions to higher regions, as well as a higher degree of dependence of China’s technology industry on foreign technology, and the insignificant impact of inter-regional trade inflows generated in technology-intensive regions. (iv) the effect of DT on outflows and inflows generated in resource-intensive regions is significant at the 1% significance level, with relatively small impact effects (0.997, 0.967). The possible explanation is that resource-intensive regions are mainly mineral resources and trade flows are regulated by low-carbon and green economy policies. China’s resource-intensive industries have significant resource endowment advantages, a relatively homogeneous export commodity structure as well as low value added, and the boosting effect of the digital economy may be smaller compared to other industries.

5.3. Practical Application

This study has some cases in practical application. First, the application of digital technology has led to the application of translation platforms, reducing search costs and transaction costs due to language barriers; the digital economy empowers the logistics industry, and big data accelerates the matching of logistics information, significantly reducing the transportation costs of inter-regional trade and thus promoting inter-regional trade. Secondly, digital platforms such as Alibaba and eBay have massive trading of goods and services, increasing the supply of goods from manufacturers and promoting a cross-regional exchange of resources and technologies, which in turn promotes inter-regional trade. Again, the improvement of digital infrastructure, such as broadband access and fiber optic cable inputs, provides the possibility for less-developed regions to obtain favorable goods and services, significantly promoting inter-regional trade in less-developed regions. Finally, technological applications such as videoconferencing provide favorable channels for trade in services such as health care and education.

5.4. Limitations and Future Directions

Although this study draws a series of impact mechanisms and regional heterogeneity of the digital economy affecting inter-regional trade, there are still some limitations. First, the time selection of this study is limited, for the changing trend in recent years could not be included, which makes the conclusion of the analysis limited. Second, existing studies have not formed a unified standard for the measurement of the digital economy, and there may be differences in the selection of indicators for the digital economy, leaving room for further testing. Finally, the measurement of inter-regional trade is only developed in conjunction with Chinese data and may be less generalizable to other countries. In addition, there is room for further improvement in this study in the future. On the one hand, this study examines the channel mechanism of the digital economy affecting inter-regional trade based on the transportation channel, the demand side and the supply side and the explored mechanism may be imperfect, and the mechanism can be explored from other perspectives in the future. On the other hand, the empirical analysis of the digital economy on inter-regional trade in this study is conducted at the macro level, and how it can be applied to the more detailed city level is a direction for further investigation in the future. Finally, this study applies to the government’s formulation of macroeconomic policies regarding the advancement of the digital economy and the promotion of consumption in the domestic market. As well as enterprises applying digital technology to enhance productivity and market scale advantage to provide a reference.

6. Conclusions and Recommendations

The digital economy not only provides a strong impetus for high-quality economic development but also provides a new path to promote inter-regional trade and boost domestic demand. Based on China’s 2006–2017 panel data, this paper empirically analyzes and researches the spatial effect, channel mechanism and regional heterogeneity of the digital economy on inter-regional trade using the spatial Durbin model and the mediation effect model. The research results show that:

First, The digital economy clearly promotes both outflows and inflows of inter-regional trade, with positive spatial spillover effects. In less developed regions or non-border regions, the promotion of inter-regional trade by the digital economy is more prominent and has greater potential.

Vigorously promote the development of the digital economy. The government should increase the new infrastructure for the digital economy, take the “East counts and West counts” as the traction, and purposefully and moderately ahead of the layout of data centers, arithmetic networks, 5G and other new-generation digital infrastructures, so as to provide carriers for the participation of underdeveloped regions in a wider range of market transactions. Improve the level of digital technology inputs, build a trade exchange platform and thus promote the precise connection between production and marketing, cultivate the formation of a modern and leading supply chain through the government’s help in providing enterprises with favorable conditions for talent, technology and publicity and promote the specialized development of inter-regional trade. Increase efforts to promote the digital transformation of enterprises, actively drive the integration of enterprises with digital technology at the organizational and technological levels, accelerate the digital empowerment of the entire trade chain, enhance the level of trade digitization and provide protection for inter-regional trade.

Second, the digital economy, mainly in the transportation channel and on the demand side, promotes inter-regional trade by reducing trade costs and stimulating market demand.

Enterprises continue to strengthen the digital economy’s empowerment of traditional industries such as finance and logistics to reduce transaction costs and transportation costs of inter-regional trade. Relying on digital technology to establish digital management platforms and development platforms, make full use of all kinds of resources to unleash market demand for inter-regional trade. Create an online and offline integration model to drive the circulation of services trade, cultivate and grow new business forms and modes such as telecommuting and online diagnosis and treatment and promote the empowerment of digital technology for traditional services trade, so as to facilitate the realization of a higher level of market demand and broader market potential.

Third, on the supply side, the role of the digital economy in promoting inter-regional trade through resource allocation and technological innovation effects needs to be enhanced.

The role of the digital economy in resource allocation should be further explored to enhance productivity and increase product supply and promote inter-regional trade. Promote the comprehensive integration of the digital economy and traditional financial institutions, enhance the adaptability between financial supply and real financial demand and improve the efficiency of capital allocation. Actively explore the social security system for labor mobility in the context of the digital economy, promote the mobility of high-skilled talents to less developed regions and at the same time, cultivate laborers with digital skills in various aspects to improve the allocation of labor resources. In addition, the government should continue to promote the construction of national-level innovation platforms, accelerate the construction of industrial innovation bases in various regions, continuously enhance the innovative power of the digital economy, give full play to the spillover effect of knowledge and technology, further release the dividends of the digital economy, and synergistically promote the unimpeded flow of the domestic market.

Fourth, the digital economy promotes labor-intensive, capital-intensive and resource-intensive inter-regional trade, with labor-intensive regions having the largest impact effect; however, for technology-intensive regions, the digital economy has a significant impact on domestic trade outflows and a limited impact on trade inflows.

Promote the deep integration of the digital economy and regional advantageous industries, give full play to the advantages of regional resource endowment, promote the realization of a coordinated division of labor in different industry-intensive regions, deepen cooperation and cross-regional trade and promote inter-regional trade. Developing and expanding digital industry clusters as an important hand, the government has introduced supportive policies to enhance the development capacity of digital industry clusters in digital technology, application platforms, solutions, etc., to stimulate the spillover effect of the digital economy on technological innovation of enterprises, and to guide the transformation of technology-intensive industries from an international cycle to a domestic cycle.