Digital product import and domestic value-added ratio of export: Theory and demonstration

Abstract

Based on the matching of the database of China Industry Business Performance and China Customs Trade from 2000 to 2013, this paper constructs the digital product import index, and adopts the method of panel data modeling to systematically investigate the impact and mechanism of digital product import on the domestic value-added rate of Chinese enterprises' export from the theoretical and empirical aspects. The research finds that the import of digital products significantly promotes the improvement of the domestic value-added rate of enterprises' export, and the core conclusion is still valid after considering the endogeneity of variables, changing the measurement index and estimation method. The mechanism test finds that the import of digital products improves the domestic value-added rate of enterprises' export through two channels: cost markup and relative price. In addition, the heterogeneity test finds that the import of digital products has a stronger effect on the improvement of the domestic value-added rate of enterprises' export in non-export enterprises, pure general trading enterprises, foreign-funded enterprises, labor-intensive enterprises and enterprises in the eastern region.

1. Introduction

In the context of Global Value Chain (GVC), all economic entities participate in the international division of labor based on their own advantages, and obtain and share economic benefits according to the Domestic Value-added Ratio (DVAR) provided to the production of products. Therefore, relying solely on the import of foreign intermediate products to complete the last process of product production, China's manufacturing not only makes little profit from exports, but also is subject to others for a long time. Once the upstream supply chain of foreign countries fluctuates and breaks, China's manufacturing will not be able to timely alleviate the conduction pressure caused by the fluctuations and breaks of the upstream supply chain [1]. To this end, improving the domestic value-added rate and profit level of Chinese enterprises' exports has become an important proposition to promote China's high-quality development, and it is also the focus of current academic research.

At the same time, digital trade has become an important driving force for international economic and trade development in the post-pandemic era. In 2020, China's “14th Five-Year Plan” proposes to speed up the process of digital development and promote the further improvement of digital level. On November 7, 2021, the Fourth China-Singapore Cooperation Innovation Forum on Trade in Services with the theme of “Digital Empowerment, creating a new pattern of high-level opening-up of Trade in Services” was held in Shanghai. The forum held heated discussions on how to develop digital trade, how to improve the capacity and level of digital trade governance, how to establish and improve the digital trade governance system, and how to promote the high-level development of service trade with digital empowerment. According to the United Nations Conference on Trade and Development (UNCTAD), the scale of global digital delivery services will reach 3167.587 billion US dollars in 2020, accounting for about 63.55 % of the volume of trade in services. Between 2005 and 2020, the global average annual growth rate of digital delivery services is 6.68 %, higher than the average annual growth rate of 4.21 % for global delivery services during the same period. Global trade in digital goods, as measured by tangible ICT products, reached US $2460.435 billion in imports in 2019, accounting for about 13.36 % of the total import volume of trade in goods. From 2000 to 2019, the average annual growth rate of the import scale of digital goods trade was 4.69 %, and the relatively high import scale of digital trade has become an important engine driving the economic development of various countries. In the global value chain division mode, as an important yardstick to measure the real benefits of a country's enterprises participating in international trade, the development of digital trade will inevitably bring new impacts on it [2]. At present, no literature has conducted a systematic quantitative analysis of the relationship between digital product import and enterprises' export DVAR, and only some literature focuses on import factors such as the quality of imported intermediates [3], the liberalization of imported intermediates trade [4], the types of imported products [5] and import competition [6], ignoring the impact of the increase in the scale of digital product import on enterprises' export DVAR, which has important practical value in the current situation. In view of this, this paper, based on the enterprises' panel data from 2000 to 2013, examines the internal correlation between digital product import and enterprises' export DVAR, and provides theoretical support and practical ideas for the internal mechanism of digital product import promoting the improvement of enterprises' value chain status.

This paper may enrich and expand existing research in the following aspects: first, few literatures have examined the influencing factors of Chinese enterprises' export DVAR from the perspective of digital product import. Based on the micro data of enterprises participating in digital trade, this paper examines the impact of digital product import on the export DVAR of enterprises, and finds that the import of digital products significantly promotes the improvement of export DVAR of enterprises, enriching relevant literatures. It makes up for the existing research on the influencing factors of Chinese enterprises' export DVAR, and has certain guiding value for China's strategic policy choice of promoting high-level opening up. Secondly, this paper deeply explores the channel paths through which the import of digital products affects the export of DVAR, so as to explain the essential relationship between the two and make corresponding supplements to existing research.

The rest of the paper is organized as follows: Section 2 reviews the relevant literature and proposes the research hypothesis; Section 3 builds the empirical model, measures the key variables and introduces the data sources; Section 4 reports the main empirical results; Section 5 conducts the influence channel test; Section 6 is the discussion; and Section 7 is the conclusion and further policy implications.

2. Literature review and research hypothesis

2.1. Review of relevant literature

The existing literature shows that digital trade brings technology cross-border spillover by promoting the flow of R&D elements, which is an important way for enterprises to obtain positive externalities of knowledge spillover and improve their technological level [[7], [8], [9], [10]]. Developing the import trade of digital products, such as the import of industrial robots, with the help of the “learning and demonstration effect” brought by digitization, is conducive to improving the productivity and R&D efficiency of enterprises and improving their innovation ability [11]. Some scholars have paid attention to the impact of digital product import on the development of enterprises. Liu Jiaqi and Sun Puyang [12] defined digital products for the first time and found that the import of digital products helps improve the innovation ability of enterprises. Subsequent scholars have confirmed that the import of digital products can improve the technical complexity of enterprises' export [13], promote the export upgrading of emerging service trade [14], improve the cost markup rate of enterprises [15], and improve the quality of enterprises' export products [16].

At the same time, the existing research on export DVAR mainly focuses on the measurement and influencing factors. In terms of measurement, Wang et al. [17] and Koopman et al. [18] used the input-output model to measure the export DVAR at the national or industry level, and Upward et al. [19], Zhang et al. [20], Kee and Tang [21] measured the export DVAR at the micro-enterprise level based on the matching of the database of China Industry Business Performance and China Customs Trade. The examination of factors influencing export DVAR primarily centers on foreign direct investment [20], the quality of imported intermediate goods (Zhu Zhujun et al., 2018), trade liberalization [22], outward direct investment [23], robot application [24], digital economy [25], and digital technology competition [26]. However, there is currently limited literature available regarding the impact of digital product imports on enterprise export DVAR. Therefore, this article aims to elucidate the relationship between these two aspects.

2.2. Import of digital products and export of enterprises DVAR

Referring to the research ideas of Kee and Tang [[21], [27]], the theoretical expression of export DVAR is obtained as follows Eqn (1):

$$DVA{R}_{ft}=1-{\chi }_M\frac{1}{mk{p}_{ft}}\frac{1}{1+{\left(P_t^I/P_t^D\right)}^{\sigma -1}}$$ (1)

The above Eqn (1) shows that the export DVAR of an enterprise is determined by the markup ($mk{p}_{ft}$) and the relative price of imported intermediates and domestic intermediates ($p_t^I/p_t^D$). The first derivation of $mk{p}_{ft}$ and $p_t^I/p_t^D$ in Eqn (1) is obtained, and the following is obtained Eqn (2):

$$\frac{\partial DVA{R}_{ft}}{\partial mk{p}_{ft}}=x_M\frac{1}{mk{p}_{ft}^2}\frac{1}{1+{\left(p_t^I/p_t^D\right)}^{k-1}}>0\text{and}\frac{\partial DVA{R}_{ft}}{\partial \left(p_t^I/p_t^D\right)}=\left(k-1\right)x_M\frac{1}{mk{p}_{ft}}\frac{{\left(p_t^I/p_t^D\right)}^{k-2}}{{\left[1+{\left(p_t^I/p_t^D\right)}^{k-1}\right]}^2}>0$$ (2)

As can be seen from Eqn (2), both the cost markup rate ($mk{p}_{ft}$) and the relative price of imported intermediates ($p_t^I/p_t^D$) are positively correlated with the export DVAR of enterprises, that is, the larger the cost plus rate ($mk{p}_{ft}$) and the relative price of imported intermediates ($p_t^I/p_t^D$), the larger the export DVAR of enterprises. On the contrary, the export DVAR is smaller. The economic logic of this conclusion is that if the enterprise's cost markup rate increases, it means that the enterprise expands the ratio of total output to total input, and the enterprise's profitability increases, thus increasing the domestic value-added rate of the enterprise's export. At the same time, if the relative price of imported intermediate products and domestic intermediate products increases, according to the principle of cost minimization, enterprises will substitute imported intermediate products for production, thus increasing the domestic value-added rate of enterprises' exports.

Next, this paper will further explore how the import of digital products affects the export DVAR of enterprises through the channels of cost markup rate ($mk{p}_{ft}$) and the relative price of imported intermediates and domestic intermediates ($p_t^I/p_t^D$).

• (1)Cost markup rate ($mk{p}_{ft}>0$) channel.

The next step is to explore how the import of digital products affects the export DVAR of enterprises through the cost markup rate ($mk{p}_{ft}$).

Through the import of industrial robots and chips and other digital products can directly enhance the type and quantity of technology-intensive elements within the enterprise, but also put forward higher requirements for the matching labor skills and technology, and promote the rise of high-end jobs such as artificial intelligence engineers and technicians, industrial robot system operators and the elimination of low-skill and low-skill labor. This kind of making up for the key factors and eliminating the backward ones is conducive to the optimal allocation of resources, and then promotes the improvement of enterprise productivity. The higher the productivity of the enterprise, the lower the input cost per unit of output, the more conducive to the enterprise to establish a higher product price. Ottaviano and Melitz [28], Bellone et al. [29] both believe that there is a positive correlation between enterprise productivity and its cost markup rate. At the same time, by importing digital products, enterprises can break the limitations of traditional factor input and increase the input of information and intelligent digital production factors. The diversification of production factor input enables enterprises to expand the production of product types and quantities, which effectively weakens the negative impact of changes in international market demand and prices on enterprises, and promotes them to obtain more stable and abundant profits, and the improvement of profits encourages enterprises to have more funds to invest in R&D and innovation. In addition, the import of foreign competitive advanced products will bring competitive pressure to domestic enterprises that produce homogeneous products, and then produce substitution risks, so faced with import competition, domestic enterprises will increase research and development investment to improve old products and develop new products. Generally speaking, the development and innovation activities of enterprises can better meet the diversified needs of consumers in many aspects such as the appearance, function and quality of products, thus increasing the sales of products and occupying more markets. Since consumers are not sensitive to the price of new products, enterprises tend to obtain higher profits by setting higher prices to make up for the cost of innovation. Cassiman and Vanormelingen [30] found through empirical test that product R&D and innovation can promote the improvement of enterprises' cost markup rate. Liu Qiren and Huang Jianzhong [31] found that enterprises' R&D and innovation activities promoted the increase of cost markup rate. Yuan Liu (2023) conducted an empirical test using the trade database of Chinese industrial enterprises and Customs from 2000 to 2013 and found that the import of digital products significantly promoted the increase of enterprises' cost markup rate through the two function channels of “improving productivity” and “increasing R&D input".

According to the conclusion of the above mechanism analysis $\frac{\partial DVA{R}_{ft}}{\partial mk{p}_{ft}}>0$, combined with the theoretical analysis $\frac{\partial mk{p}_{ft}}{\partial Dig{e}_{ft}}>0$, the further conclusion can be obtained $\frac{\partial DVA{R}_{ft}}{\partial Dig{e}_{ft}}=\frac{\partial DVA{R}_{ft}}{\partial mk{p}_{ft}}\frac{\partial mk{p}_{ft}}{\partial Dig{e}_{ft}}>0$, which means that the import of digital products improves the export DVAR of enterprises through the cost plus rate channel.

• (2)Relative price ($p_t^I/p_t^D$) channels.

With the import of digital products, enterprises will break through the restrictions of traditional element investment and add information and intelligent digital investment. The diversification of production factor inputs encourages enterprises to increase the types of products produced on the basis of the original product types, which means that the supply of products in the domestic market will increase, which will reduce the price of intermediate products in the domestic market ($p_t^D$), and then increase the relative price of imported intermediate goods and domestic intermediate inputs ($p_t^I/p_t^D$). At the same time, the diversification of imported products reduces the cost of enterprises [[32], [33], [34]], including the search, information and matching costs for digital product importers to obtain intermediate inputs in the domestic market. The decline of these costs further reduces the supply price of domestic intermediate inputs, thereby increasing the relative price of imported intermediate goods and domestic intermediate inputs ($p_t^I/p_t^D$). It can be seen that by importing digital products, the relative price of imported intermediate goods and domestic intermediate inputs $\left(p_t^I/p_t^D\right)$ can be increased, that is $\frac{\partial \left(p_t^I/p_t^D\right)}{\partial Dig{e}_{ft}}$. According to the conclusion of the above mechanism analysis $\frac{\partial DVA{R}_{ft}}{\partial (p_t^I/p_t^D)}>0$, combined with the theoretical analysis $\frac{\partial \left(p_t^I/p_t^D\right)}{\partial Dig{e}_{ft}}$, the conclusion can be further obtained $\frac{\partial DVA{R}_{ft}}{\partial Dig{e}_{ft}}=\frac{\partial DVA{R}_{ft}}{\partial \left(p_t^I/p_t^D\right)}\frac{\partial \left(p_t^I/p_t^D\right)}{\partial Dig{e}_{ft}}>0$, which means that the import of digital products improves the domestic value-added rate of enterprises' exports through the relative price of intermediate goods. Based on the above analysis, this paper proposes the following hypotheses.

Hypothesis 1

Digital product import promotes the improvement of export DVAR.

Hypothesis 2

Digital product import improves export DVAR through cost markup and intermediate relative price channels.

3. Empirical model, index measurement and data processing

3.1. Setting of empirical model

This study focuses on the impact of import of digital products on export DVAR of enterprises. The following measurement model (Eqn (3)) is set with reference to relevant literature:

$$DVA{R}_{fit}={\alpha }_o+{\alpha }_{1}Dig{e}_{ft}+{\alpha }_2{X}_{fit}+{\delta }_t+{\delta }_i+{\delta }_r+{\epsilon }_{fit}$$ (3)

Where the subscript $f$ indicates the enterprise, $i$ indicates the industry, $t$ indicates the year, $r$ indicates the region. DVAR_fit represents the domestic value-added rate of enterprises' exports and $Dig{e}_{ft}$ represents the import index of enterprises' digital products. $X_{fit}$ represents the control variables at the enterprise level, including: Enterprise scale (Scale), which is expressed by the sales volume of the enterprise, and logarithmic processing is performed for specific use; Enterprise scale (Age),which is expressed as the difference between the year in which the enterprise was established and the year in which the enterprise was established. State-owned enterprise virtual variable (State), if the enterprise belongs to the class of state-owned enterprises, the value is 1, otherwise the value is 0; Foreign capital enterprise dummy variable (Fdi): If the enterprise belongs to the class of foreign capital enterprises, the value is 1, otherwise the value is 0; Financing constraint (Loan) is expressed by the ratio of accounts receivable to fixed assets, and logarithms are used in specific use; Total factor productivity (Tfp): The LP method is used to measure, and the logarithm is taken when it is used. In addition, considering that the omission of explanatory variables in the empirical model may lead to bias in the regression results, in order to effectively reduce the bias, this paper also controls non-observational fixed effects such as year fixed effect ${\delta }_t$, industry fixed effect ${\delta }_i$ and region fixed effect ${\delta }_r$, ${\epsilon }_{fit}$ is random disturbance term.

3.2. Calculation of core indicators

• (1)Measurement of export DVAR index

Refer to the research methods Upward et al. [19], Zhang Jie et al. [20], Kee and Tang [21] and Liu Xinheng [35] to calculate the rate of export domestic added value of enterprises. In the specific calculation, the matching data of China Industrial Enterprise Database and China Customs Trade Database are adopted, and the formula is as follows Eqn (4):

$$DVA{R}_{fit}^N=\left\{\begin{array}{l}1-\frac{i{m}_{fit}^{adj\_o\_bec}+zb{p}_{fit}^{adj}+i{m}_{fit}^f}{y_{fit}},N=1\\ 1-\frac{i{m}_{fit}^{adj\_p}+zb{p}_{fit}^{adj}+i{m}_{fit}^f}{y_{fit}},N=2\\ {\varphi }_1(1-\frac{i{m}_{fit}^{adj\_o\_bec}+zb{p}_{fit}^{adj}+i{m}_{fit}^f}{y_{fit}})+{\varphi }_2(1-\frac{i{m}_{fit}^{adj\_p}+zb{p}_{fit}^{adj}+i{m}_{fit}^f}{y_{fit}}),N=3\end{array}\right)$$ (4)

Where, $N=1$ represents general trade enterprises, $N=2$ represents processing trade enterprises, and $N=3$ represents mixed trade enterprises. ${\varphi }_1$ represents the proportion of general trade exports in total exports of mixed trade enterprises, ${\varphi }_2$ represents the proportion of processing trade exports in total exports of mixed trade enterprises, $i{m}_{fit}^{adj\_o\_bec}$ represents the actual amount of intermediate goods imported by general trading enterprises, $i{m}_{fit}^{adj\_p}$ represents the actual amount of intermediate goods imported by processing trade enterprises, $zb{p}_{fit}^{adj}$ represents the portion of capital goods transferred to export products by way of depreciation, $i{m}_{fit}^f$ represents the use of domestic raw materials containing elements of foreign products, $y_{fit}$ represents the output of the enterprise, specifically expressed by the total output value of the enterprise, data from the China industrial enterprise database.

• (2)Calculation of imports of digital products

According to the research ideas proposed by Liu Jiaqi and Sun Puyang [12], digital products are divided into two categories, namely tangible and intangible products. 25 keywords belonging to tangible digital products are extracted,1 and trade agents are excluded. Imported digital products are identified through 25 keywords according to the Chinese product tax code in the Chinese customs database. Then, with reference to the subheading notes in the Notes on Import and Export Tariff Commodities and Items (2020 Edition), the commodities that cover the above keywords but are not digital products will be excluded. Finally, according to the ICT product categories released by OECD in 2020, the digital products identified above will be combined with the digital economic development level of the import source countries. Construct the expression of digital product imports at the enterprise level as follows Eqn (5):

$$Dig{e}_{ft}=\sum _d\frac{M_{fdt}^{dige}}{M_{ft}^{total}}\times NR{I}_{dt}$$ (5)

Where $Dig{e}_{ft}$ represents the import of digital products by the enterprise $f$ in the year $t$, and the natural logarithm is taken when used; $M_{fdt}^{dige}$ represents the amount of digital products imported by the enterprise $f$ from the country or region $d$ in the year $t$, $M_{ft}^{total}$ represents the total import amount of the enterprise $f$ in the year $t$. Drawing on the research ideas of Dang Lin et al. (2021), this study uses the Network Readiness Index ($NR{I}_{dt}$) to measure the development level of digital economy in country or region $d$, with data from Global Information Technology Report. Definitions of main variables, data sources, and descriptive statistical analysis are listed in Table 1 and Table 2.

Table 1.

Main variable definitions and data source descriptions.

Variable	Name of Variable	Meaning of Variable
DVAR	Export Domestic Value-added Rate	Measured by matching the Industrial Enterprise Database and the Customs Database, +1 natural logarithm
Dige	Import of Digital Products	Measured by the Customs Database and the Global Information Technology Report, +1 natural logarithm
Scale	Enterprises' Size	Measured by the Industrial Enterprise Database, Enterprise Sales +1 natural logarithm
Age	Enterprises' Age	Measured by the Industrial Enterprise Database, the year minus the year of establishment +1 natural logarithm
State	State-owned or not	Measured by the Industrial Enterprise Database, “1″ state-owned enterprise, “0″ non-state-owned enterprise
Fdi	Foreign-funded or not	Measured by the Industrial Enterprise Database, “1″ foreign-funded enterprise, “0″ local enterprise Financing Constraints
Loan	Financing Constraints	Measured by the Industrial Enterprise Database, the ratio of accounts receivable to fixed assets +1 natural logarithm
Tfp	Enterprises' Productivity	Measured by the Industrial Enterprise Database, LP method +1 natural logarithm

Table 2.

Descriptive statistical analysis of main variables.

Variable	Sample size	Mean	Standard deviation	Minimum	Maximum
DVAR	135140	0.6056	0.1264	−4.5937	0.6931
Dige	135140	10.2518	2.9035	0	22.0901
Scale	135140	10.6363	1.3049	8.5172	18.8718
Age	135140	2.1710	0.6005	0.6931	4.7449
State	135140	0.0307	0.1725	0	1
Fdi	135140	0.3015	0.4589	0	1
Loan	135140	0.1606	0.1290	−2.2370	0.6927
Tfp	135140	2.0424	0.1499	−3.1285	2.6313

3.3. Data use and explanation

The data used in this study are mainly from China Industrial Enterprise Database and China Customs Trade database, and the research period is 2000–2013. We first aggregate the monthly China Customs trade data into annual data, and then aggregate the annual China Customs trade data at the product level to the enterprise level, and then match the annual China Customs trade database at the enterprise level with the database of Chinese industrial enterprises according to the identity information of enterprises. Based on the matched database, the export DVAR of Guangzhou enterprises, the import of digital products and the control variables at the enterprise level are measured. The database of China's industrial enterprises at the enterprise level comes from the National Bureau of Statistics, and it is processed according to the methods of Cai and Liu [36], Brandt [37] and Nie Huihua et al. [38]. The product-level China Customs Trade database is from the General Administration of Customs of China.

4. Empirical test and analysis

4.1. Baseline regression results

The benchmark regression results of the impact of digital product imports on export DVAR of enterprises are shown in Table 3. In order to investigate the robustness of the results, columns [[1], [2], [3], [4], [5], [6]] in Table 3 are the baseline regression results obtained by adding control variables step by step. Column [1] in Table 3 is the conclusion that only considers the influence of the import index of digital products. It can be found that its estimated coefficient is negative and positive, and it passes the significance level of 1 %, which preliminarily indicates that the import of digital products significantly promotes the increase of export DVAR of enterprises. Columns [[2], [3], [4], [5], [6]] are the regression results after gradually adding control variables. It can be found that all the estimated coefficients of digital product import variables in each column are still significantly positive, indicating that the significant positive relationship between digital product import and export DVAR of enterprises is robust. In addition, by observing the estimated coefficients of control variables, it can be found that they are consistent with most research conclusions on the export DVAR of enterprises. The enterprise's scale is significantly negative, which means that the larger the enterprise is, the lower its DVAR is. The enterprise's age coefficient is significantly positive, which means that the earlier the enterprise is founded, the higher its DVAR is. The possible explanation is that the longer the enterprise's existence time, the more mature the enterprise's management level and production technology, and then transform the enterprise's competitive advantage in the export market. The virtual variable coefficient of state-owned enterprises is significantly positive, indicating that state-owned enterprises have a significant promoting effect on the export of DVAR. The virtual variable coefficient of foreign-funded enterprises is significantly negative, indicating that foreign-funded enterprises have a significant inhibitory effect on the export DVAR of enterprises. The possible explanation is that foreign-funded enterprises are more engaged in processing trade [39], and mainly rely on the import of foreign intermediate input factors for assembly and re-export, thus reducing the export DVAR of enterprises. Financing constraints are significantly negative, indicating that financing constraints will restrict the improvement of export DVAR. The total factor productivity is significantly positive, indicating that enterprises with higher productivity have higher export DVAR, because enterprises with higher productivity have higher cost-plus rate, and the increase of cost plus rate can significantly increase the export DVAR of enterprises.

Table 3.

Basic results.

Variable	[1]	[2]	[3]	[4]	[5]	[6]	[7]
Dige	0.0195*** (0.0010)	0.0190*** (0.0010)	0.0193*** (0.0010)	0.0190*** (0.0010)	0.0186*** (0.0010)	0.0186*** (0.0011)	0.0184*** (0.0010)
Scale		−0.0114*** (0.0017)	−0.0192*** (0.0017)	−0.0198*** (0.0017)	−0.0182*** (0.0017)	−0.0178*** (0.0017)	−0.0819*** (0.0026)
Age			0.08756*** (0.0043)	0.0812*** (0.0044)	0.0743*** (0.0044)	0.0742*** (0.0044)	0.0728*** (0.0043)
State				0.1243*** (0.0139)	0.1016*** (0.0140)	0.1008*** (0.0140)	0.1068*** (0.0138)
Fdi					−0.0739*** (0.0048)	−0.0736*** (0.0048)	−0.0773*** (0.0047)
Loan						−0.0491*** (0.0183)	−0.0790*** (0.0181)
Tfp							0.7542*** (0.0232)
Cons	−0.3239*** (0.0040)	−0.1941*** (0.0193)	−0.2970*** (0.0199)	−0.2806*** (0.0200)	−0.2468*** (0.0200)	−0.2430*** (0.0201)	−1.0903*** (0.0328)
Year fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Industry fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Region fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	135270	135270	135240	135240	135240	135225	135140
Adj. R-sq	0.1473	0.1484	0.1583	0.1602	0.1659	0.1661	0.1903

Note:***, **, * indicate significant at 1 %, 5 % and 10 % levels, respectively. Standard error in parentheses.

4.2. Endogeneity test

The instrumental variable method is used to overcome the endogeneity problem caused by reverse causality between the import of digital products and the cost markup rate. Drawing on the research ideas of Yu Huan et al. [13], the expression of instrumental variable import tariff of digital products is as follows Eqn (6):

$${\tau }_{ft}=\sum _{p\in f}\left(\frac{\overline{valu{e}_{fp}}}{\sum _{p\in f}\overline{valu{e}_{fp}}}\right){\tau }_{pt}$$ (6)

Where ${\tau }_{ft}$ represents the import duties on digital products of enterprise $f$ in the year $t$, $\overline{valu{e}_{fp}}$ represents the average amount of digital products imported by enterprise $f$, ${\tau }_{pt}$ represents the tariff rate of digital product $p$ in the year $t$, data from the WTO Tariff Download Facility. The regression results are shown in column [1] of Table 4. It can be found that the estimated coefficient of the variable of import digital product is positive and passes the significance level of 1 %, indicating that the import of digital product improves the export DVAR of enterprise, and the baseline regression result is robust and reliable. The value of the Wald rk F statistic is also significant at the 1 % level, and also strongly rejects the null hypothesis that “instrumental variables are weakly identified”, indicating that the instrumental variables selected in this paper are reasonable. In addition, considering that the lag of variables may lead to endogeneity problems, column [2] of Table 4 shows the regression result of digital product imports (explanatory variables) lagging for one period. It can be found that the result is still significantly positive, indicating that the lag of explanatory variables for one period still significantly improves the export DVAR of enterprises, and the baseline regression result is robust. At the same time, the KP-LM (Kleibergen Paap rk LM statistical) statistic $p$ passed the significance level of 1 %, strongly rejecting the null hypothesis of “insufficient identification of instrumental variables”. The Wald rk F statistic $p$ is also significant at the 1 % level, and also strongly rejects the null hypothesis that “instrumental variables are weakly identified”, indicating that the instrumental variable selected in this paper is reasonable. In addition, considering that the lag of variables may lead to endogeneity problems, column [2] of Table 4 shows the regression result of digital product import (explanatory variable) lagging for one period. It can be found that the result is still significantly positive, indicating that the lag of explanatory variable for one period still significantly improves the export DVAR of enterprise, and the baseline regression result is robust.

Table 4.

Endogeneity test results.

Variable	[1]	[2]
	2SLS estimation	Explanatory variables in lag
Dige	0.2466*** (0.0434)	0.3133*** (0.4769)
Control variables	Yes	Yes
Year fixed effect	Yes	Yes
Industry fixed effect	Yes	Yes
Region fixed effect	Yes	Yes
KP-LM	698.255***
Wald rk F	786.564***
N	122348	98459

Note:***, **, * indicate significant at 1 %, 5 % and 10 % levels, respectively. Standard error in parentheses.

4.3. Robustness test

• (1)Other measures of export DVAR of enterprise

The benchmark regression in this paper is based on the estimated result of enterprise export DVAR calculated on the assumption that domestic intermediate input factors contain 5 % share of foreign products. In order to investigate the stability of the results, enterprise export DVAR is re-calculated here. The first is the firm export DVAR (DVAR10), which is measured by the 10 % share of foreign products in the domestic factors put into use. The second method is to remove the impact of capital goods conversion and re-calculate the firm export DVAR (DVAR11). The third way is to use the original calculation of enterprise export DVAR, do not take logarithm, use the original calculation value to estimate. The above re-calculated DVAR for export of enterprises is brought back into the benchmark regression model [1], and the test results are listed in Table 5 below. It can be found that the estimated coefficients of columns [[1], [2], [3]] of import of digital products in Table 5 are still significantly positive, indicating that the significant promotion effect of import of digital products on export DVAR of enterprises is robust. It does not change depending on how the explained variable is measured.

• (2)Other measures of digital product import of enterprise

Table 5.

Robustness test results.

Variable	[1]	[2]	[3]	[4]	[5]
	Other measures of firms exporting DVAR			Other measures of firms import of digital goods
	DVAR10	DVAR11	DVAR	Digevalue	Digeshare
Dige	0.0188*** (0.0011)	0.0193*** (0.0011)	0.0133*** (0.0006)	0.0021*** (0.0005)	0.0305*** (0.0009)
Control variables	Yes	Yes	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes	Yes	Yes
Industry fixed effect	Yes	Yes	Yes	Yes	Yes
Region fixed effect	Yes	Yes	Yes	Yes	Yes
Cons	−0.1352*** (0.0197)	−0.1672*** (0.0205)	−0.8636*** (0.0108)	−0.0592*** (0.0093)	−0.0329*** (0.0193)
N	127122	127092	127303	127336	127143
Adj. R-sq	0.1805	0.1739	0.2058	0.2100	0.2048

Note:***, **, * indicate significant at 1 %, 5 % and 10 % levels, respectively. Standard error in parentheses.

In this part, the amount of digital products imported by enterprise (Dige_value) and the share of digital products imported by enterprise (the ratio of the amount of tangible digital products imported by enterprise to the total amount of imports by enterprise) (Dige_share) are used to replace the measurement methods of digital products of enterprise considering the development level of digital economy of the import source countries in the benchmark regression. The test results are shown in column [4] and column [5] of Table 5. Through the test results, it can be found that the sign and significance of the estimated coefficient of the import variable of digital products have not changed greatly, indicating that the impact of the import of digital products on the export DVAR of enterprise does not vary with the measurement methods of the explanatory variables, and the benchmark regression results are relatively robust.

4.4. Heterogeneity analysis

• (1)Heterogeneity of whether export or not

From the regression results of columns [1,2] in Table 6, it can be found that the import of digital products significantly promoted the increase of export DVAR of both exporting and non-exporting enterprises. However, comparing the absolute value of the estimated coefficient, it is found that for non-exporting enterprises, the effect of digital product import on enhancing export DVAR is stronger than that of exporting enterprises. The possible reason is that the productivity of exporting enterprises is higher than that of non-exporting enterprises, and the effect of promoting the productivity improvement through the import of digital products is much lower than that of non-exporting enterprises. Therefore, the effect of digital product import of non-exporting enterprises on enhancing the export DVAR of enterprises is stronger than that of exporting enterprises.

• (2)Heterogeneity of trade modes

Table 6.

Subsamples regression Ⅰ

• (5)
  Heterogeneity of region

Variable	[1]	[2]	[3]	[4]	[5]
	Export	Non-export	Pure processing	Pure ordinary	Mixed
Dige	0.0177*** (0.0011)	0.0214*** (0.0033)	0.0123*** (0.0017)	0.0155*** (0.0013)	0.0131*** (0.0010)
Control variables	Yes	Yes	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes	Yes	Yes
Industry fixed effect	Yes	Yes	Yes	Yes	Yes
Region fixed effect	Yes	Yes	Yes	Yes	Yes
Cons	−0.1212*** (0.0208)	−0.2605*** (0.0591)	−1.7202*** (0.0567)	−0.2259*** (0.0388)	−0.7667*** (0.0363)
N	94206	32862	57822	25496	51734
Adj. R-sq	0.1889	0.1322	0.1346	0.1778	0.1383

Note:***, **, * indicate significant at 1 %, 5 % and 10 % levels, respectively. Standard error in parentheses.

Columns [[3], [4], [5]] of Table 6 show the estimated results of three types of enterprises: pure processing trade, pure ordinary trade and mixed trade. The results show that the import of digital products significantly improves the export DVAR of three types of enterprises. However, comparing the absolute value of the estimation coefficient, it is found that the import of digital products has the greatest impact on pure ordinary trade enterprises, followed by mixed trade enterprises, and pure processing trade enterprises. The possible reason is that the production mode of pure processing trade enterprises “two ends outside” determines that they are not sensitive to the market, and the purpose of import is to export, so the participation of pure processing trade enterprises in the local market is low. At the same time, pure processing trade enterprises are more in order to earn processing fees, which also determines that it is difficult for them to obtain the productivity improvement effect through the import of digital products, thus increasing the cost markup rate and ultimately leading to the increase of export DVAR. Therefore, the import of digital products has the least impact on export DVAR of pure processing trade enterprises through the cost markup rate effect.

• (3)Heterogeneity of ownership

Columns [1,2] of Table 7 show the estimated results of foreign and local enterprises, which show that the import of digital products significantly improves the export DVAR of both foreign and local enterprises. However, comparing the estimated coefficient values, it is found that for foreign-funded enterprises, the effect of digital product import on enhancing export DVAR is stronger than that of local enterprises. The possible reason is that compared with local enterprises, the “foreign” nature of foreign-funded enterprises themselves leads to a smaller gap with foreign advanced technologies. Therefore, through the import of digital products, it is easier to use and absorb the technology and experience contained in foreign advanced digital products, which is more conducive to the competitive advantage of enterprises in the export market.

• (4)Heterogeneity of factor intensity

Table 7.

Subsamples regression Ⅱ

Variable	[1]	[2]	[3]	[4]	[5]	[6]	[7]
	Foreign-funded	Local	Labor	Capital	Technology	East	Midwest
Dige	0.0189*** (0.0014)	0.0174*** (0.0013)	0.0313*** (0.0032)	0.0178***(0.0016)	0.0133***(0.0016)	0.0197***(0.0010)	0.0187***(0.0037)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Industry fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Region fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Cons	−1.0997***(0.0482)	−1.1474***(0.0452)	−1.1653***(0.1255)	−0.7995***(0.0507)	−1.2837***(0.0533)	−1.0908***(0.0341)	−0.7912***(0.1527)
N	68008	57095	32915	49764	55796	102839	22965
Adj. R-sq	0.1859	0.2123	0.2218	0.1981	0.1707	0.1864	0.2793

Note:***, **, * indicate significant at 1 %, 5 % and 10 % levels, respectively. Standard error in parentheses.

Drawing on the research ideas of Dai Xiang and Jin Bei [40], the manufacturing industry is divided into three types: labor intensive, capital intensive and technology intensive according to the feature of factor intensity.2 The estimated results are shown in Table 5 below. It can be seen from the results of columns [[3], [4], [5]] in Table 7 that import of digital products significantly improves export DVAR of labor-intensive, capital-intensive and technology-intensive enterprises. However, by comparing the values of the estimation coefficients of import variables of digital products among the three, it is found that the estimation coefficients of variable of import digital products of labor-intensive enterprises is the largest, followed by capital-intensive enterprises. The technology-intensive enterprises are the smallest, indicating that the import of digital products has the greatest effect on the export DVAR of labor-technology-intensive enterprises.

Columns [6,7] of Table 7 show the estimated results of enterprises in the eastern region and enterprises in the central and western regions. The results show that the promotion effect of digital product import on DVAR export of enterprises in the eastern region is higher than that of enterprises in the central and western region. The possible reason is that the development level of digital economy in the eastern region is relatively high and the digital foundation for the development of digital trade is relatively strong, which can better release the transformation effect of technology and productivity brought by digital trade, and thus the promotion effect of export DVAR is higher than that of enterprises in the central and western regions.

5. How does digital product import affect export DVAR: A test of influencing channels

5.1. Construction of the intermediary effect model

According to the baseline estimation results, the import of digital products has a significant promoting effect on the export DVAR of enterprises. However, what internal mechanism is used by the import of digital products to promote the increase of export DVAR of enterprises? The analysis of the internal mechanism will help to deeply understand the internal relationship between export DVAR of enterprises and import of digital products. This paper constructs the intermediary effect model to analyze how the import of digital products affects the DVAR of enterprises. Based on the above analysis of the influence mechanism, the enterprises’ cost markup rate ($mk{p}_{fit}$) and the relative price of imported intermediates and domestic intermediates ($p_t^I/p_t^D$) are selected as the intermediary variables, and the following intermediary effect models are constructed by referring to the practice of existing literatures to test the mechanism of the effect of digital product import on enterprises' export DVAR as follows Eqn (7), Eqn (8), Eqn (9), and Eqn (10):

$$DVA{R}_{fit}={\alpha }_0+{\alpha }_{1}Dig{e}_{ft}+{\alpha }_2{X}_{fit}+{\delta }_t+{\delta }_i+{\delta }_r+{\epsilon }_{fit}$$ (7)

$$mk{p}_{fit}={\beta }_0+{\beta }_{1}Dig{e}_{ft}+{\beta }_2{X}_{fit}+{\delta }_t+{\delta }_i+{\delta }_r+{\epsilon }_{fit}$$ (8)

$$shar{e}_{fit}=c_0+c_{1}Dig{e}_{ft}+c_2{X}_{fit}+{\delta }_t+{\delta }_i+{\delta }_r+{\epsilon }_{fit}$$ (9)

$$DVA{R}_{fit}=d_0+d_{1}Dig{e}_{ft}+d_{2}mk{p}_{fit}+d_{3}shar{e}_{fit}+d_4{X}_{fit}+{\delta }_t+{\delta }_i+{\delta }_r+{\epsilon }_{fit}$$ (10)

Where $mk{p}_{fit}$ represents the enterprises' cost markup rate. Using the research idea of Domowitz et al. [41] for reference, the accounting method is used for calculation, and the natural logarithm is taken for regression. $shar{e}_{fit}$ is expressed by the ratio of domestic intermediate inputs to foreign intermediate inputs, in which the difference between enterprise intermediate inputs and imported intermediate inputs is used to calculate the domestic intermediate inputs, and the natural logarithm is still taken in the regression.

5.2. Regression results of the intermediary effect model

Table 8 reports the test results of the mechanism of import of digital products on export DVAR. Column [2] reports the estimated results of enterprises' cost markup rate as the explained variable. It is found that the estimated coefficient of digital product import is significantly positive, indicating that digital product import significantly promotes the increase of enterprises' cost markup rate. Column [3] reports the estimation results based on the ratio of domestic intermediate inputs to foreign intermediate inputs as the explained variable. It is found that the estimated coefficient of digital product import is significantly positive, indicating that digital product import significantly increases the ratio of domestic intermediate inputs to foreign intermediate inputs, that is, the relative prices of imported intermediate goods and domestic intermediate goods. Column [4] reports the impact of the intermediary variable enterprises' cost markup rate on export DVAR. It is found that the estimated coefficient of the intermediary variable enterprises' cost markup rate is significantly positive, indicating that enterprises' cost markup rate significantly improves export DVAR. Column [5] reports the effect of the ratio of domestic intermediate inputs to foreign intermediate inputs on export DVAR of enterprises. It is found that the estimated coefficient of the ratio of domestic intermediate inputs to foreign intermediate inputs of intermediary variables is significantly positive, indicating that the ratio of domestic intermediate inputs to foreign intermediate inputs significantly improves export DVAR of enterprises. It shows that the relative prices of imported intermediates and domestic intermediates can help increase the export DVAR of enterprises. It is worth noting that after adding the firm's cost markup rate and the ratio of domestic and foreign intermediate inputs to column [4] and column [5], the absolute value of the estimated coefficient of the core explanatory variable digital product import decreased significantly compared with the benchmark regression results in column [1]. At the same time, column [6] is the result after adding the intermediary variables of the enterprises' cost markup rate and the ratio of domestic intermediate inputs to foreign intermediate inputs. Compared with the benchmark regression result of column [1], the absolute value of the core explanatory variable of the estimated coefficient of digital product import has further decreased. Again, it shows that the import of digital products improves the export DVAR of enterprises through two possible channels: increasing the cost markup rate and relative price of enterprises. The above analysis shows that there are two significant channels of “cost markup” and “relative price".

Table 8.

Regression of mediation model.

Variable	[1]	[2]	[3]	[4]	[5]	[6]
	DVAR	Markup	share	DVAR	DVAR	DVAR
Dige	0.0185*** (0.0011)	0.0014*** (0.0004)	0.1926*** (0.0046)	0.0184*** (0.0010)	0.0052*** (0.0007)	0.0045*** (0.0007)
Markup				0.1228*** (0.0143)		0.2397*** (0.0111)
share					0.1117*** (0.0008)	0.1125*** (0.0008)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes	Yes	Yes	Yes
Industry fixed effect	Yes	Yes	Yes	Yes	Yes	Yes
Region fixed effect	Yes	Yes	Yes	Yes	Yes	Yes
Cons	−0.1224*** (0.0192)	0.2908*** (0.0092)	2.5810*** (0.1561)	−0.9092*** (0.0390)	−1.4225*** (0.0239)	−0.1228** (0.0179)
N	127143	127304	127304	127143	127143	127143
Adj. R-sq	0.1834	0.1610	0.2249	0.1971	0.3026	0.3192

Note:***, **, * indicate significant at 1 %, 5 % and 10 % levels, respectively. Standard error in parentheses.

6. Discussion

Making full use of the technical characteristics of digital products is conducive to further improving the profitability of Chinese enterprises, as well as their influence and trade gains in the international market. Few studies have explored the impact of digital product import on the export DVAR of Chinese enterprises. Based on the matching of the database of China Industry Business Performance and China Customs Trade from 2000 to 2013, this paper constructs the digital product import index, and systematically examines the impact and mechanism of digital product import on the export DVAR of Chinese enterprises from two aspects of theory and empirical research. We find that the import of digital products has a significant promotion effect on the export DVAR of Chinese enterprises. After considering the endogeneity, using different estimation methods and different measurement indexes, the core conclusion still holds. Our conclusion is basically consistent with previous studies. (2023)[25] found that the digital economy significantly improved the export DVAR of Chinese enterprises. Qi Jianhong and Cai Zhenkun [24] examined the impact of robot application on export DVAR, and found that robot application helps to improve the export DVAR of Chinese enterprises. The existing research lacks the exploration of the relationship between digital product import and enterprise export DVAR. This paper examines the impact of digital product import on the export DVAR of Chinese enterprises from two aspects of theory and empirical research, making up for the existing exploration of the influencing factors of Chinese enterprises' export DVAR.

Considering whether the enterprise exports or not, the promotion effect of digital product import on the export DVAR of non-exporting enterprises is higher than that of exporting enterprises. In terms of trade mode, the promotion effect of digital product import on the export DVAR of pure general trade enterprises is the strongest, followed by mixed trade enterprises, and the smallest of pure processing trade enterprises. In terms of factor intensity, the promotion effect of digital product import on the export DVAR of labor-intensive enterprises is the strongest, followed by capital-intensive enterprises, and the weakest of technology-intensive enterprises. In terms of ownership, the promotion effect of digital product import on the export DVAR of foreign-funded enterprises is stronger than that of local enterprises. In terms of the economic development level of the region where the enterprise is located, the promotion effect of digital product import on the export DVAR of enterprises in the eastern region is stronger than that of enterprises in the central and western regions.

The mechanism analysis shows that the import of digital products mainly improves the export DVAR of enterprises through the two channels of “cost markup” and “relative price.” In terms of the investigation of channel mechanism, this paper follows the previous practice of first examining the influencing factors of export DVAR, and then includes the factor of digital product import on this basis to explore the channel path of digital product import affecting the export DVAR of enterprises, which limits the channel path of digital product imports affecting enterprises' export DVAR, and makes it impossible to comprehensively evaluate the impact path of digital product imports on enterprises’ export DVAR, which is the limitation of this study. In the future, the channel path of digital products import that affects the export DVAR of enterprises can be directly investigated from the perspective of digital products import, so as to comprehensively evaluate the impact of digital products import on the export DVAR of enterprises.

7. Conclusion

Based on the database of China Industry Business Performance and China Customs Trade from 2000 to 2013, this paper constructs the import index of digital products, and systematically investigates the impact and mechanism of digital product import on China's export DVAR from the theoretical and empirical aspects. The results show that the import of digital products significantly improves the export DVAR of enterprises. The import of digital products mainly improves the export DVAR of enterprises through the “cost markup” and “relative price”. The research of this paper confirms that the import of digital products can improve the export DVAR of enterprises, which is of great significance for the economic effect of China's “digital trade” policy in the new era. First, enterprises should be encouraged to increase the import of digital products, build the import platform of digital products, develop digital trade, and improve digital technology. Second, the cost of digital trade is generally lower than that of traditional trade, so it is necessary to vigorously promote digital trade, create a good business environment for digital trade, encourage enterprises to transform digitally, and improve the level of intelligent trade and digital management. Third, the import of digital products has a higher promotion effect on the export of DVAR of non-export enterprises, pure general trade enterprises, labor-intensive enterprises, foreign-funded enterprises and enterprises in the eastern region. Therefore, it is necessary to fully understand the heterogeneous impact of digital product import on enterprises, take differentiated countermeasures, focus the focus of digital import trade on the above enterprises and industries, and maximize the driving effect of digital import trade. Fourth, the international procurement experience of pure general trade enterprises and labor-intensive enterprises should be paid attention to, and support them to import digital products rich in digital technology, so as to give full play to the driving effect of digital import trade. Fifth, the areas with a lower level of digital economic development will have more restrictions on the display of digital product technology. Therefore, the government should help the areas with a lower level of digital economic development to improve the construction of digital infrastructure, increase the Internet penetration rate, introduce high-end digital talents, and create platform enterprises with strong competitiveness.

Ethics requirement

Not applicable.

Data availability statement

Data will be made available upon request to the corresponding author.

Funding declaration

This study was funded by Guangdong Social Science Planning 2022 annual discipline co-construction project (No. GD22XYJ25), the 14th Five-Year Plan for the Development of Philosophy and Social Sciences in Guangzhou in 2022 (No. 2022GZGJ27), the Featured Innovation Project of Guangdong Provincial Education Department in 2022 (No. 2022WTSCX139), and the Social Science Planning Project of Foshan City in 2024 (No. 2024-GJ221).

CRediT authorship contribution statement

Liu Yuan: Writing – review & editing, Methodology, Funding acquisition, Data curation. Shaoming Chen: Writing – review & editing, Validation, Supervision, Resources. Yufang Wang: Supervision.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.