Digital Service Trade and Labor Income Share—Empirical Research on 48 Countries

Abstract

Fast-paced digitization is affecting all aspects of life, including the way we trade. This study empirically analyzes the influence and mechanism of digital service trade on labor income share using 48 countries’ panel data from 2005 to 2019. The research results show that digital service trade has a positive effect on labor income share. Specifically, the impact of digital service trade on labor income share includes productivity effect, structural effect, and distribution effect. The results also show that the export of digital service trade is more conducive to increasing domestic labor income share than the import. The positive impact of digital service trade on labor income share in OECD countries is more significant than in non-OECD countries. At the same time, the digital service trade’s income distribution effect has heterogeneity due to differentiated national digital service trade barriers’ level. This paper discusses theoretically and empirically the impact of digital service trade on the share of labor income for the first time, and puts forward policy suggestions on formulating more open and win-win digital service trade policies, improving labor income, and promoting social equity and inclusive global trade.

1. Introduction

In the fourth industrial revolution, with digitalization as its main feature, the use of digital technology greatly reduced the cost and time of the trade, making the mode, structure, and pattern of global trade continue to change and giving birth to “digital trade”. Weber (2010) [1] first proposed the concept of digital trade; it is defined as products or services transmitted in electronic form. Digital service trade (from now on DST) refers to the trade in products and services delivered through the cross-border transmission of networks, which does not include digital trade in goods. It is not only included in digital trade, but also service trade [2].

From 2005 to 2019, the global DST generally maintained a high growth rate [3]. Since 2019, COVID-19 has disrupted the production and supply chain, and global trade has been hit hard. The world economy is still affected by the epidemic and the complex and volatile international situation. Many poverty reduction achievements have been lost. However, in 2020, the proportion of global trade in digitally delivered services in global trade jumped from 51.8 to 63.5%, with a year-on-year growth of 11.7%. DST has dramatically accelerated the development of international business. The result of digital trade has spawned a global industrial chain and value chain of digital products [4], which can open up new channels for vulnerable trade groups to enter the international market. It may have a profound impact on the global division of labor and the construction of competitive advantage [5]. The development of DST is reshaping the trade ecosystem. It is very important to clarify the impact of DST on labor income share, which has important theoretical and practical significance.

In this study, we attempted to answer the following questions: Can the growing DST have a positive impact on labor income share? If there is a positive impact, what is the path mechanism? What kind of DST barriers will be more conducive to the positive income distribution effect of DST?

The structure of the rest of the paper is as follows. Section 2 presents the literature in the related fields. Section 3 presents the theoretical analysis and research hypothesis. Section 4 describes the model construction process, indicator measurement methods, and sample data sources. Section 5 provides an empirical analysis. Section 6 provides and discusses the mechanism test. Section 7 summarizes the research conclusions and further provides practical policy recommendations.

2. Literature Review

2.1. Research on the Change in Labor Income Share

Whether the labor income share is in a reasonable range is not only related to the vital interests of micro individuals, but also an important indicator reflecting the quality of macroeconomic development. E. Daudey and C. Garcia Penalosa [6] pointed out that “the increase of income gap between factors will significantly worsen the interpersonal income distribution”. Capital share is positively related to income inequality (Bengtsson E. and Waldenstrm D. 2018) [7]. The primary distribution determines the basic pattern of final income distribution [8]. Due to the high marginal propensity of workers to consume, the labor income decline leads to a reduction in the consumption rate of the whole society, making economic growth more dependent on investment and exports. It will also increase the social security pressure on the government, which is not conducive to harmonious labor relations, and may even affect political stability. Reducing inequality is also an essential part of the UN Sustainable Development Goals. However, after the 1980s, the decline of the labor share and the widening income gap worldwide triggered extensive research in the academic community. The main reasons include the process of globalization, industrial structure, trade and globalization, FDI, capital account openness, export expansion, imperfect competition in the market and product market, technological progress, changes in the prices of production factors, etc. Rognlie (2015) showed that the increase in the value of housing was one of the reasons for the decline in labor income share in the overall economy [9]. The research of G. Gutiérrez and Piton S. (2020) challenged the general view that the percentage of global labor income declined, and pointed out that the there are differences in the accounting systems of countries when calculating the percentage of labor income, and the impact of housing and self-employed workers should be considered in the study of labor income share, otherwise it is easy to draw erroneous conclusions [10].

However, it is worth noting that since 2008, labor income share has risen both in developed economies and developing economies [11], and the declining trend of labor income share has almost turned after the economic crisis worldwide. From 2000 to 2020, China’s labor income share declined from 59.2% in 2000 to 54.9% in 2011, and rose to 58.6% in 2020. However, the existing literature mainly focuses on the reasons for the decline of labor income share, and few studies have explored the recovery of labor income share in recent years. Some studies focused on the phenomenon that China’s labor income share fell first and then rose [12], and attributed its changes to: industrial structure transformation [13] and technological progress bias [14]. The transnational research focusing on the recovery of labor income share worldwide attributed the reason to the higher capital-output ratio of developing countries compared with developed countries.

2.2. Research on Digital Service Trade

In 2012, the U.S. Bureau of Economic Analysis (USBEA) first defined the concept of DST as cross-border service trade in which information and communication technologies participate and play an important role [15], specifically including copyright and license fees, financial and insurance services, professional and technical services of communication services, etc. The concept of DST has been expanded by various organizations and countries, including the OECD [16], IMF, and UNCTAD [17]. In December 2020, China’s Digital Trade Development White Paper 2020, released by the Chinese Academy of Information and Communications, defined DST as closely integrated with international trade through digital technology [18]. According to the Ministry of Commerce of the People’s Republic of China, DST includes the digitalization of traditional service industries, as well as new economic models or business forms that emerge after technological iteration.

UNCTAD has put forward a set of statistical schemes for DST. Based on the classification of major products, it has screened out the service trade that can be delivered through cross-border network transmission, distinguished digital service trade and non-digital service trade, and thus provided the basis for analysis and measurement of DST of countries.

Domestic research on DST focuses on the connotation of DST [2], development status [19], rules and policies of DST [20], DST barriers [21], its impact on export complexity [22], its impact on carbon emissions [23], the DST network [24], etc. In research on digital trade include preferential trade agreements, DST rules and other topics, it is pointed out that establishing authoritative, scientific, and influential digital trade rules in the world is a significant issue that needs to be considered and solved first to promote the development of digital trade and improve the international trading system. Regarding the research on the income distribution effect of digital trade, Xie Jingting (2022) [25] theoretically analyzed the possible impact of digital trade on labor income share, but lacked mechanism and empirical research. Research found that, as an essential part of digital trade, DST is conducive to alleviating the income gap.

2.3. Marginal Contribution of This Paper

The marginal contribution of this paper mainly includes the following aspects: (1) Based on the existing research, this paper discusses the possible impact of DST on labor income share through theoretical and empirical analysis for the first time, and analyzes its mechanism. (2) This paper analyzes and verifies that DST has a significant positive impact on labor income share, and summarizes its impact mechanisms as having a productivity effect, structural effect, and distribution pattern remodeling effect. (3) The results also show that the export of DST is more conducive to the increase of domestic labor income share than the import. The positive impact of DST on labor income share in OECD countries is more significant than that in non-OECD countries; at the same time, DST’s income distribution effect has heterogeneity due to the differentiated level of national DST barriers. In addition, the existing research mainly focuses on the interpretation of the declining trend of labor income share. This paper seeks to better analyze the economic phenomenon of the rising labor income share in recent years, supplement and improve the research on the impact of DST in the field of the income distribution, and supplement the research on the influencing factors of the labor income share.

3. Theoretical Analysis and Research Hypothesis

3.1. The Essential Attribute of DST and Its Similarities and Differences with Traditional Trade

Digital trade is the expansion of traditional trade in the era of the digital economy. The division of labor and scale economy is the internal motivations of both traditional trade and digital trade [26]. The increment of DST is also the sum of the expansion margin and the intensification margin. According to the heterogeneous enterprise trade model of Metliz (2004) [27], trade costs and related factors will affect enterprise export expansion (intensive margin), import and export market dynamics (expansion margin), and then affect the total trade volume. However, the difference between DST and traditional trade is an important reason why it can still grow against the trend in the current epidemic and complex international situation. Digital technology has brought about significant changes in the original communication and transmission methods. The difference between digital trade and traditional trade determines that the impact of digital trade on factor income may have a completely different causal relationship and internal mechanism from conventional trade. It is reflected in the following.

3.1.1. Reduce Trade Costs

Trade costs include all the costs of delivering goods to end users, mainly including: transportation costs (freight costs and time costs), policy barriers (tariff and non-tariff barriers), information costs, contract execution costs, costs related to the use of different currencies, legal and regulatory fees, and local distribution costs (wholesale and retail) [28]. Due to the virtualization of the transaction mode and the delivery mode transmitted through the platform, DST has overcome the logistics cost and time cost of freight transportation between the two parties in traditional trade, and significantly accelerated the completion time of the transaction. At the same time, platform transactions in DST also reduce related trade costs, such as information costs and contract execution costs.

Unlike traditional trade, DST has the inherent characteristics of virtualization, platform, intensification, inclusiveness, personalization, and ecology. Virtualization is embodied in its trade targets, namely digital service products and services, which contain digital knowledge and information, and have the characteristics of virtual elements. The virtualization of delivery means that DST mainly conducts transactions and transmission through virtualized network platforms. At the same time, the commerce of digital service trade mostly relies on the network platform, which makes the supply and demand of both sides of the transaction more efficient. The labor, capital, and technology can be intensively invested on the network platform. All of the above have greatly reduced trade costs. Maurice Obstfeld and Kenneth Rogoff (2000) [29] believe that all major problems in international macroeconomics depend on trade costs. Trade costs are closely related to economic policies (Anderson and Wincoop 2002) [30]. At the same time, they also believe that trade costs have a significant impact on welfare, and the value of policy-related costs often exceeds 10% of national income. The reduction of trade cost will be conducive to the increase of factor share.

3.1.2. Promote Technological Innovation and Improve Labor Productivity

Compared with traditional service trade, DST relies on the support of digital technology. Digital technology has made great changes in the original communication and transmission methods, and trade has also undergone digital and intelligent upgrading. Digital trade emphasizes the crucial role of digital technology in order, production or transmission, enabling services to be transmitted through the internet and e-commerce trading platforms.

The intensive nature of DST can more intensively input labor, capital, and technology through the network platform, enabling the supply and demand of both parties to the transaction to be more efficiently matched. The externalities of the internationalization of DST encourage countries participating in digital trade to obtain technological spillover effects [31]. Enterprises participating in the export of DST will increase their innovation efforts to meet the personalized needs of the demand side. Meanwhile, the import of DST will intensify the competition in the industry, create a reverse force effect on domestic enterprises, and promote the innovation of domestic DST. Therefore, it can be considered that in digital trade, information such as relatively dispersed trade and consumer preferences on the demand side in traditional trade can be gathered through the network platform. This intensification provides more possibilities for digital trade to meet differentiated needs, and for private customization of personalized products and services [32]. It improves the reaction speed and efficiency of the production side to the demand side. This has constantly promoted the innovation of products and services and promoted the improvement of labor productivity. Both the import and export of DST can promote the creation of domestic digital technology and the improvement of labor productivity.

3.1.3. Promote the Upgrading of the Industrial Structure

In the context of the digital economy, the labor resources of various countries can complete the service orders required by other countries online, such as online consulting, telemedicine, language training, etc., that, to some extent, break through the transaction boundary of services. This promotes the prosperity of the knowledge economy. Knowledge and services are more accessible to make money than before. Digital technology has brought about the digitalization of consumer content, which has enabled more in-depth integration of financial, education, medical, design, consulting, and other services, and has spawned more new digital consumer service products, thus further increasing the proportion of DST in international service trade.

The development of digital trade has created a new demand for the skills of the labor force, which requires the labor force to better adapt to the technical skills needed by digital trade. At the same time, cross-border digital transactions of services have also intensified industry competition, which is more dependent on the continuous improvement of the training and skill levels of enterprise employees in the industry. Digital trade can promote the continuous improvement of the level of human capital [31]. The level of human capital is an essential factor determining the R&D behavior of enterprises. The progress of human capital will promote the development of knowledge-intensive industries and high-tech industries, and promote the continuous upgrading of industrial structure. At the same time, digital service trade, as a new growth point of the global economy, has launched a fierce competition among countries in order to occupy this highland. It has also promoted the increase of R&D investment intensity. DST may promote the upgrading of the industrial structure by improving human capital and increasing R&D investment.

3.1.4. Rebuild Competitive Advantage

International trade originates from the urgent desire of countries to improve their current welfare level and obtain trade benefits. In the pattern of traditional trade and global value chain, the low-end lock-in of developed countries makes it difficult for developing countries to improve the marginal output of labor. Schmitz (2004)’s [33] research shows that after joining GVC, most developing countries are locked in low-value-added production links. This capture-type relationship makes it difficult for developing countries to complete higher functional upgrading and chain upgrading. The income distribution in the value chain is highly uneven [34]. Zhang Shaojun and Hou Huifang [35] also found that the more embedded in the global value chains, the worse the terms of trade of developing countries will be. Traditional trade tends to “lock” countries that are in a weak position in the value chain and cannot climb up to the higher value-added link, which reduces the demand for high-tech labor, makes it challenging to optimize human capital, and also leads to a decline in labor income share.

However, existing studies state digital trade is the third stage of global trade development [36]. At the same time, its decentralized and borderless development has reshaped the main body of international trade and the international division of labor. By making good use of digital technology and providing better quality and differentiated products and services for the demand side, the late developing countries or enterprises can seize the competitive advantage, and break the pyramid-shaped governance structure and distribution mode dominated by the original GVC. They will be able to participate in higher value-added links, which significantly affects income distribution patterns.

3.2. Digital Service Trade and Labor Income Share

3.2.1. Productivity Effect

According to the above analysis, both the import and export of DST can promote the innovation of domestic technology and the improvement of labor productivity. Karabarbounis and Neiman (2013) [37], and Acemoglu and Restrepo (2018) [38] found that due to the promotion of technological progress, the demand for labor in production has gradually declined. Many machinery and equipment have been put into production activities to replace human resources. Capital-biased technological progress brings innovation and productivity improvement, which will reduce labor income share. Previous studies have also pointed out that digital trade and traditional trade have the same behavior essence. Specialized production, division of labor, and scale economy are the internal motivations of conventional trade and digital trade, which can improve productivity. Therefore, DST may reduce labor income share by increasing labor productivity, which is summarized as the productivity effect in this paper.

3.2.2. Structural Effect

In addition to the productivity effect, this paper holds that DST will impact labor income share through structural effect and distribution pattern remodeling effect.

From the perspective of structural effect, the development of DST promotes the increase of the proportion of the service industry in a country. According to the above analysis, DST will promote the upgrading of industrial structure by improving human capital and increasing R&D investment. Yao Zhanqi [31] also found that the development of digital trade can promote the upgrading of industrial structure. However, due to various labor income shares of different industries, industrial restructuring will lead to changes in labor income shares. The industrial structure is the crucial factor affecting labor income share [39], and the upgrading of industrial structure will have a positive impact on labor income share. Therefore, DST may have a positive impact on labor income share through the structural effect of industrial structure upgrading.

3.2.3. Reshaping Distribution Pattern Effect

In the traditional trade theory, the terms of trade is an essential reference to measure a country’s foreign trade profits, which reflects the comparison between the strength and weakness of the bargaining power of both sides, thus determining the distribution pattern of trade profits. According to the definition of OECD (2020) [16], terms of trade refer to the ratio of export and import price indexes. Taussing (1927) [40] pointed out that the net physical terms of trade can directly reflect a country’s profitability in the global market. It is generally believed that the improvement of terms of trade is positively related to the increase in trade benefits [36].

With the development of digital trade, the number of consumers of digital products has dramatically increased, which has reduced the average use cost of digital products and made consumers more inclined to purchase digital products and services. The above analysis also shows that compared with traditional trade, DST has lower transaction costs, which will be conducive to giving play to scale economy effect, reshaping competitive advantages, and improving the international market share of its products and services. Existing studies show that an essential factor affecting labor income share includes the competition state of the market and product market [41]. Therefore, the improvement of comparative advantage and terms of trade will allow a country to obtain a specific monopoly position in the market and product market, enable it to have more say in the economic rent distribution in business, further promote the increase of trade gains and the improvement of welfare, thus promoting the growth of the country’s labor income share. This paper summarizes it as the reshaping effect of the distribution pattern.

To sum up, the impact of DST on labor income share ultimately depends on the net effect of the negative productivity effect, the positive structural effect, and the reshaping effect of the distribution pattern. Based on the above inference, this paper puts forward the following hypotheses.

Hypothesis 1a (H1a).

Digital service trade has a positive effect on labor income share.

Hypothesis 1b (H1b).

Digital service trade has a negative impact on labor income share.

Hypothesis 2 (H2).

Digital service trade reduces labor income share through productivity effect.

Hypothesis 3 (H3).

Digital service trade improves labor income share through structural effect.

Hypothesis 4 (H4).

The export of digital service trade improves labor income share through the reshaping effect of the distribution pattern.

4. Research Design

4.1. Sample Selection and Data Source

This paper selected the national and regional DST data released by the United Nations Conference on Trade and Development (UNCTAD) database to match with the World Bank World Development Indicators and Penn Table PWT10.0 (Pen World Table 10.0) data. Countries with serious data deficiencies were eliminated. Due to data availability, the data are the annual observations of 48 major countries and regions in the world over the 15 years 2005–2019, which are used as research samples for empirical tests.

4.2. Models and Variables

4.2.1. Dependent Variable

Labor income share is the proportion of labor factor income in added value. Since the SNA guidelines followed by most countries except the United States include “all units engaged in market production independent of the owner” in the enterprise sector, this leads to different treatment of housing and self-employment. In order to ensure the feasibility of a transnational comparison of labor income share, this paper refers to G. Gutiérrez and Piton S. (2020) [10] to exclude real estate activities from the calculation of labor income share. Since it is very challenging to estimate the wages of self-employed workers (Elsby, Hobijn and Sahin 2013 [42]), we mitigated this impact by controlling the proportion of self-employed workers in the labor force in the control variables. Regarding Wu Xiaoyi, (2020) [43], in the robustness analysis, this paper selected the labor income share data of The Penn World Tables (PWT). The dataset was preferable because it uses the most plausible adjustment approach for each country and year, it expands the coverage of self-employed income adjusted labor income shares by using proxy variables for countries whose mixed income data are not available (S. Paul 2020) [44].

In the current measurement of labor income share, the macro labor income share data provided by the PWT Table is relatively stable, with a slight fluctuation range [43].

4.2.2. Core Explanatory Variable

The definition and measurement of DST adopted in this paper was the definition, scope, and measurement method of UNCTAD. We used indicators of digitally delivered service trade published in UNCTAD database to measure a country’s digital service trade level; digital11, digital12, and digital13 are, respectively, the total DST, the proportion of the total DST in domestic service trade, and the ratio of the total DST in the world. The data were selected from the UNCTAD database, including the annual observations of 48 major countries and regions in the world in the 15 years 2005–2019.

4.2.3. Intermediary Variables

• Labor productivity: this paper uses labor productivity as an intermediary variable to verify the labor productivity effect. Specifically, we used the labor productivity data of countries in the PWT10.0 Payne Table.
• Industrial structure: referring to Gan Chunhui et al. [45], this paper adopted the ratio of the output value of the tertiary industry to the output value of the secondary industry as the measurement of industrial structure. Specifically, the theoretical analysis of this paper is that DST has multiple intermediary effects of upgrading industrial structure and then increasing labor income share through the promotion of human capital and the promotion of R&D investment. This paper refers to Zhu Shujin [46]. It uses the World Development Index (WDI) dataset of the World Bank to express the level of human capital by the percentage of higher education enrollment rate in the total enrollment population. The R&D investment level of a country is expressed by the percentage of R&D investment in GDP.
• Terms of trade: according to the theoretical analysis in this paper, DST reshapes the trade distribution pattern through the reshaping effect of the distribution pattern, affecting labor income share. This paper refers to the definition of OECD (2020) [16], and selects the ratio of a country’s export and import value index to represent its terms of trade level as the intermediary variable of the impact of DST on labor income share.

4.2.4. Other Control Variables

According to existing research, capital deepening (Rognlie, 2015) [9], economic development level, industrial structure [37], technological progress [35,36], outsourcing [47], FDI, foreign trade [48,49], self-employed workers (G. Gutiérrez and Piton S. 2020) [10] will all have a particular impact on labor income share. Considering the robustness of measurement results, to analyze the effects of DST on labor income share, and referring to the indicators selected by Zhu Shujin [46] and other existing literature, the final control variables in this paper include the following variables:

• Tax: net production tax;
• Fixgdp: capital deepening degree;
• Rgdpcap: real GDP per capita;
• Structure: industrial structure;
• Tec: technical innovation level;
• Fdigdp: proportion of foreign investment in GDP;
• Tradegdp: the proportion of trade in GDP;
• Selfemploy: the proportion of self-employed personnel in the total labor force.

Table 1. Descriptive statistics of variables.

Variable	Obs	Mean	Std. Dev	Min	Max
lnlabshare	720	0.586463	0.1628288	−1.136276	−0.3231072
lndigital1	720	9.581921	1.640501	4.102759	12.83659
lndigital2	720	3.642725	0.4331602	1.918634	4.515753
lndigital3	720	−0.3214078	1.617604	−5.804321	2.704915
rgdpcap	720	34891.51	19138.14	3224.09	96812.5
lntec	720	−0.8285807	0.3258221	−2.0498	−0.183335
taxgdp	720	18.39807	6.836428	7.83788	62.5028
fdigdp	720	9.973271	34.26041	−57.5323	449.083
lnfixgdp	720	−1.469341	0.2072539	−2.24639	−0.62378
lnstructure	720	0.382617	0.2111652	−0.109204	1.15787
tradegdp	720	112.2912	81.19878	24.3902	442.62
lnhc	720	1.13949	0.1432298	0.618882	1.47054
lnrd	720	−4.576146	0.9648483	−7.79619	−3.07326
tradecondition	720	1.012705	0.2100585	0.564779	1.8976
lnproductivity	720	3.666924	0.6874357	1.23256	4.717159
selfemploy	720	22.77785	15.92892	6.09	84.37

reports the descriptive statistical results of each variable.

4.2.5. Model Settings

In order to test the research hypothesis, referring to Sheng Bin and Hao Birong [50], this paper constructed the benchmark model as follows:

$$\ln {labshare}_{i,t}=\alpha 0+\alpha 1\ln {\mathrm{digital}}_{i,t}+\alpha 2{Control}_{i,t}+{Year}_{i,t}+{Country}_{i,t}+{\epsilon }_{i,t}$$

(1)

In the above model, lnlabshare is labor income share, which is the explanatory variable of the econometric model, and lndigital is the level of DST, which is the core explanatory variable of concern in this paper. i and t represent country and year, respectively, and control represents a series of control variables; year and country are year and country fixed effects, respectively. The related variables are logarithmized in the model. In order to avoid “pseudo regression”, we conducted a correlation test on the data before the empirical analysis process. What we are interested in is the coefficient of DST in the model, which describes how a country’s labor income share will change as the level of DST increases.

5. Empirical Analysis

5.1. Benchmark Estimation

This paper established the estimation coefficient of the econometric model according to the econometric norms and data. To avoid the problems of cross-section correlation, heteroscedasticity and collinearity and ensure the stability of the data, before the empirical analysis, we first conducted the variance expansion factor test (VIF) on the variables, and found that the variance expansion factor of each variable was 6.56, indicating that there was no serious multiple collinearities among variables. Panel data passed the LLC and IPS panel unit root test, which indicated that the data were stable. The fixed effect model was selected through the Hausman test. To reduce the influence of heteroscedasticity and cross-section correlation on the regression results, we used the two-way fixed effect model to alleviate the problem, and controlled the relevant time-varying and time-invariant variables according to the country fixed effect and annual fixed effect. Shi Bingzhan [51] believed that holding the country effect could effectively solve the endogenous bias caused by missing variables.

Table 2. DST and labor income share.

Dependent Variable: Lnlabshare	(1)	(2)	(3)	(4)	(5)	(6)
lndigital1	0.0137 ***	0.0045 **	0.0110 ***			0.0095 ***
	(0.0015)	(0.0023)	(0.0020)			(0.0020)
lndigitla12				0.0228 ***
				(0.0052)
lndigital3					0.0072 ***
					(0.0015)
lnstructure			0.1268 ***	0.1211 ***	0.1232 ***	0.1229 ***
			(0.0213)	(0.0214)	(0.0214)	(0.0213)
lntec			0.0586 ***	0.0564 ***	0.0560 ***	0.0586 ***
			(0.0085)	(0.0085)	(0.0085)	(0.0085)
fixgdp			0.0018 ***	0.0019 ***	0.0018 ***	0.0018 ***
			0.0002	0.0002	0.0002	0.0002
tradegdp			−0.0006 ***	−0.0005 ***	−0.0005 ***	−0.0006 ***
			(0.0001)	(0.0001)	(0.0001)	(0.0001)
rgdpcap			−0.0000 ***	−0.0000 ***	−0.0000 ***	−0.0000 ***
			(0.0000)	(0.0000)	(0.0000)	(0.0000)
Taxgdp			0.0003	0.0003	0.0003	0.0003
			0.0002	0.0002	0.0002	0.0002
selfemploy			0.0010 ***	0.0005	0.0010 ***	0.0010 ***
			(0.0004)	(0.0004)	(0.0004)	(0.0004)
Control	no	no	yes	yes	yes	yes
country/ year	no	yes	yes	yes	yes	yes
N	720	720	720	720	720	720
adj. R-sq	0.1061	0.9198	0.9491	0.9481	0.9483	0.9485

Note: the t value is in parentheses; *** p < 0.01, ** p < 0.05.

shows the estimated results of the benchmark regression of the impact of DST on labor income share.

The dependent variable of columns 1, 2, and 3 in Table 2 is lndigital1, the total amount of DST. Columns 1 to 3 are the regression results of non-control, control variables, control variables, and fixed effects of country and time. The coefficients of DST (digital1) are significantly positive at the 1% level, which means that DST has a significant positive impact on labor income share. Hypothesis H1a is validated.

According to the results in the third column, the regression coefficient of control variables such as industrial structure, technological innovation, capital deepening, foreign trade, and economic development level on labor income share is significantly positive, consistent with the existing research results.

5.2. Robustness Check

5.2.1. The Alternative Variable of the Dependent Variable

To further verify the reliability of the regression results, in Table 2, we replaced the independent variables in columns 4 and 5 with the proportion of DST in the country’s service trade: digital2, and the ratio of DST in the world: digital3. The results in the fourth and fifth columns show that the coefficients of DST are 0.0226 and 0.0060, respectively, which are both significantly positive at the 1% level. It can be seen that after the replacement of DST measurement indicators, the impact of DST on labor income share is still significantly positive, which means that with the increase of DST, a country’s share of labor income has significantly increased. In addition, we replaced the explained variable in this paper with the labor income share measured in the PWT table, and the empirical results were still stable. The above variable replacement did not change the basic conclusions of this paper.

5.2.2. Endogenous Test

So far, we have controlled the unobserved country-specific characteristics, year characteristics and the common impact on all countries that may be related to labor income share through year-fixed effect and country-fixed effect. However, strictly speaking, there may be other explanations for the relationship between a country’s level of DST and its share of domestic labor income. The direct and effective way to solve endogenous problems is to select appropriate tool variables. We followed the practice of Milner and Scheffel (2017) [52], taking the 1-year lag of the core independent variable as the tool variable, and constructed a dynamic panel model. As shown in

Table 3. Estimation results of 2SLS and GMM.

Panel A: 2SLS Regression
Dependent variable: Lnlabshare	(1)	(2)	(3)
lndigital1	0.0138 ***
	(0.0024)
lndigital2		0.0255 ***
		(0.0074)
lndigital3			0.0106 ***
			(0.0019)
Control	yes	yes	yes
Country/Year	yes	yes	yes
N	644	644	644
adj. R-sq	0.3809	0.3724	0.3703
Panel B: Differential GMM Dynamic Panel Regression
Dependent variable: Lnlabshare	(1)	(2)	(3)
L.lnlabsh	0.5079 ***	0.5238 ***	0.5074 ***
	(0.0385)	(0.0383)	(0.0385)
lndigital1	0.0044 ***
	(0.0016)
lndigital2		0.0021 ***
		(0.0051)
lndigital3			0.0059 ***
			(0.0012)
N	624	624	624
AR2	0.189	0.438	0.154
Hansen	0.626	0.817	0.520

Note: the t value is in parentheses; *** p < 0.01.

Panel A, three kinds of measurement indicators of DST exports were adopted, respectively. The results of two-stage least squares (2SLS) regression using the 1-year lag of endogenous variables as the tool variable were consistent with the conclusions from the benchmark regression in Table 2. This shows that whether the instrumental variable method was used or not to estimate the impact of the level of DST on labor income share, a consistent conclusion can be drawn.

Considering the dynamic consistency and endogeneity of the development level of DST, in order to ensure the robustness of the regression results, this paper also used the differential GMM method for regression. Panel B shows the differential GMM regression results under three different measures of the level of DST. AR2 in each model is bigger than 0.1, indicating that there is no second-order autocorrelation. The Hansen statistics are not significant, indicating that the tool variables selected in the model are effective, and the coefficients of core explanatory variables are stable and significant at the 1% level. The above regression results show that the benchmark regression results above are basically reliable and stable in quality.

5.3. Heterogeneity Analysis

5.3.1. Digital Service Trade Barrier

With the development of global digital trade, negotiating rules related to digital trade has become more critical. However, due to the differences in the development level of DST among countries, there are differences in the strength of digital trade barriers and protection systems. Developed countries relatively advocate a highly open international trade environment, while developing countries may tend to formulate more appropriate protection to ensure the development space of their industries [2]. In order to investigate the impact of DST on labor income share under different business environments of DST, this paper analyzed the heterogeneity of the sample with the level of digital service trade barriers as the standard.

The quantitative indicators of DST barriers mainly include the DST restriction index Dstri in the OECD database, which can be used as the proxy variable of DST barriers. However, due to the period of this indicator (starting in 2014), and its measurement method (relying on expert empowerment) being difficult to supplement and improve artificially, this paper introduced the measurement method of DST barriers based on the practice of Francois [53]. Specifically, Singapore was taken as the benchmark country for free trade, then we measured the difference between the proportion of its DST in GDP and that of other countries. This difference formed a ratio with the proportion of Singapore’s digital services trade in GDP, which was used to measure digital services trade barriers. The larger the number, the higher the DST barrier, and the more restrictive policies are, and vice versa.

This paper divided the sample into the highest 25%, the lowest 25%, and the sample between 25 and 75% for quantile regression. The regression results are shown in

Table 4. Impact of DST on labor income share under different quantile digital trade barrier levels.

Dependent Variable: Lnlabshare	(1) Sample of Countries with Digital Barriers to Trade below 25% Quantile	(2) Sample of Countries with Digital Trade Barriers in the 25–75% Quantile	(3) Quantile Sample of Digital Trade Barrier above 75%
Lndigital1	−0.0056	0.0132 ***	0.0060 ***
	(0.0078)	(0.0035)	(0.0021)
Control country/year	yes	yes	yes
N	180	360	180
adj. R-sq	0.9499	0.9513	0.9579

Note: the t value is in parentheses; *** p < 0.01.

. The sample regression results are not significant for the group at the 25% quantile level. In the sample group with digital trade barriers above 75%, DST has a significant positive effect on labor income share, and its coefficient is lower than that of the sample group with a 25–75% level. For the sample countries with DST barriers between the 25 and 75% quantiles, the impact of DST on labor income share is significantly positive, and the coefficient is the biggest.

The possible reasons are that in the sample countries with low levels of DST barriers, their DST development is still in the primary stage, resulting in backward policy formulation and imperfect systems related to DST, and their DST volume is still at a low level, so their scale cannot have an impact on the macro labor income share. However, under the high DST barriers, the more restrictive policies are, the less conducive to the development of DST. Some studies have also pointed out that digital trade barriers have a significant negative impact on DST [54]. Therefore, compared with moderate DST barriers, a higher level of DST barriers is not conducive to the development of DST, and will hinder it from playing its role in increasing labor income share.

5.3.2. DST import and DST Export

Some studies hold that foreign trade has significantly promoted labor income share [55]. Some studies have also concluded that foreign trade has a significant inhibitory effect on labor income share [56]. Jiang Lei and Zhang Yuan [57] divided foreign trade into export and import, and concluded that the two had opposite impacts on labor income share. It can be seen that import and export trade may have different effects on the impact of labor income share due to their different effects on the factors affecting labor income share. Therefore, it is necessary to study the similarities and differences between DST imports and DST exports in the process of reviewing the impact of DST on labor income share.

We replaced the explained variables in the benchmark equation with the DST import and export indicators, respectively, and obtained the following regression results.

As shown in

Table 5. Import, export of DST and labor income share.

Dependent Variable: Lnlabshare	Export			Import
	(1) DST Amount	(2) Proportion of DST in Domestic Service Trade	(3) Proportion in the World	(1) DST Amount	(2) Proportion of DST in Domestic Service Trade	(3) Proportion in the World
Lndigital1	0.0741 ***			0.0057 **
	(0.0086)			(0.0021)
lndigitla12		0.0574 ***			0.0183 **
		(0.0102)			(0.0077)
Lndigital3			0.0782 ***			0.0071 ***
			(0.0090)			(0.0022)
Control	yes	yes	yes	yes	yes	yes
country/year	yes	yes	yes	yes	yes	yes
N	720	720	720	720	720	720
adj. R-sq	0.9523	0.9491	0.9523	0.9470	0.9470	0.9474

Note: the t value is in parentheses; *** p < 0.01, ** p < 0.05.

, the explanatory variables in columns 1 to 3 are three measures of digital service trade exports, and the dependent variables in columns 4 to 6 are three measures of DST imports. It can be seen that with different measurement indicators, the impact of DST exports and DST imports on labor income share is significantly positive, all at the 1% level. This is different from the research results of the effects of imports and exports on labor income share in traditional trade, which means that with the development of DST, both imports and exports of DST will have a positive impact on a country’s share of labor income. At the same time, the coefficient of the export level of DST is far greater than the import coefficient of DST; that is, the export of DST is more conducive to the increase of domestic labor income share than the import.

5.3.3. OECD and non-OECD Countries

The Organization for Economic Cooperation and Development (OECD) divides the main restrictions affecting the development of DST into five categories: infrastructure and connectivity, electronic transactions, payment systems, intellectual property rights, and other barriers affecting DST, and provides a set of DST restriction index measures [58]. The sample of this paper included 30 OECD countries and 18 non-OECD countries. To investigate the difference between OECD and non-OECD countries in labor income share affected by DST, this paper grouped the samples for regression, and the results are as follows.

According to the results in

Table 6. OECD and non-OECD Countries.

Dependent Variable: Lnlabshare	(1) OECD	(2) Non-OECD
lndigital1	0.0146 ***	0.0060 *
	(0.0026)	(0.0032)
Control country/year	yes	yes
N	450	270
adj. R-sq	0.9357	0.9490

Note: the t value is in parentheses; *** p < 0.01, * p < 0.1.

, the coefficient of digital service trade in the sample of OECD countries is greater than that of non-OECD countries; that is, the positive impact of DST on labor income share in OECD countries is more significant. The possible reason is that most OECD countries advocate for digital trade liberalization development, and advocate reducing DST barriers worldwide. Therefore, their DST barriers may be set more reasonably. Their DST development is relatively ahead of other countries’, and the DST import and DST export volumes are also higher, which makes the positive impact of the development of DST on labor income share much bigger.

5.3.4. Heterogeneity Impact of Various Digital Service Trade Segments

The statistical program of digital service trade proposed by UNCTAD [17] is to screen the service trade that can be delivered through cross-border transmission through the network based on the classification of primary products, mainly including the following nine categories: communication services, computer services (including computer software) sales and marketing services (excluding trade and leasing services), information services, insurance and financial services, management administration and backstage services, licensing services, engineering and related scientific research and development services, and education and training services. To better explore the impact of DST on labor income share, we replaced the dependent variables in the benchmark regression with the statistical data of the above nine categories of digital service trade, the results are shown in

Table 7. Segments of DST.

Dependent Variable: Lnlabshare	(1) Communication Services	(2) Computer Services	(3) Sales and Marketing	(4) Information Services	(5) Insurance and Finance	(6) Management	(7) Licensing	(8) Engineering and RD	(9) Education and Training
Lndigital	0.0141 ***	0.0145 ***	0.0027	0.0032 ***	0.0269 ***	−0.0047 ***	0.0093 ***	0.0011 ***	0.0025 **
	(0.0031)	(0.0038)	(0.0024)	(0.0013)	(0.0056)	(0.0012)	0.0020	0.0013	0.0010
Control	yes	yes	yes	yes	yes	yes	yes	yes	yes
country/year	yes	yes	yes	yes	yes	yes	yes	yes	yes
N	720	720	720	720	720	720	720	720	720
adj. R-sq	0.9478	0.9272	0.9518	0.9368	0.9469	0.9471	0.9498	0.9405	0.9268

Note: the t value is in parentheses; *** p < 0.01, ** p < 0.05.

. It can be seen that most of the subsectors in DST have a significant positive impact on labor income share. Still, the management administration and backstage services industry have a significant negative effect on labor income share. In other industries, insurance and finance, computer services and communication services have a greater positive impact. It can be seen that DST has a positive effect on the share of labor income in total, and there is a certain degree of heterogeneity in the subdivided industries.

6. Mechanism Test

Based on the theoretical analysis, we believe that DST affects labor income share through productivity effect, structural effect and distribution pattern remodeling effect. To test the above mechanism, we constructed the following regression model with reference to Matray (2021) [59]:

$${\mathrm{Med}}_{i,t}=\beta 0+\beta 1\ln {\mathrm{digital}}_{i,t}+\beta 2{Control}_{i,t}+{Year}_{i,t}+{Country}_{i,t}+{\epsilon }_{i,t}$$

(2)

$$\ln {labshare}_{i,t}=\lambda 0+\lambda 1\ln {\mathrm{d}igital}_{i,t}+\lambda 2\ln {\mathrm{d}igital}_{i,t}*{M\mathrm{ed}}_{i,t}+\lambda 3{Control}_{i,t}+{Year}_{i,t}+{Country}_{i,t}+{\epsilon }_{i,t}$$

(3)

where Med refers to the intermediary variable. Equations (2) and (3) were used to test how DST affects the labor income share through the productivity effect, structural effect, and the distribution pattern remodeling effect.

6.1. Productivity Effect

Referring to existing literature, this paper selected the labor productivity index in PWT10.0 to represent a country’s labor productivity, and used Models 2 and 3 to verify the mechanism. The following table shows the regression results when the core explanatory variables are, respectively, the total volume of DST, the export of DST, and the import level of DST.

It can be seen from

Table 8. DST affects labor income share through productivity effect.

Dependent Variable	(1)		(2)		(3)
	lnproductivity	lnlabshare	lnproductivity	lnlabshare	lnproductivity	lnlabshare
Total: lndigital1	0.0396 ***	0.0277 ***
	(0.0091)	(0.0091)
Export:lndigital11			0.26 ***	0.1102 ***
			(0.0354)	(0.0155)
Import: lndigital21					0.0228 ***	0.0198 ***
					(0.0087)	(0.0099)
lndigital × lnproductivity		−0.0022 **		−0.0054 *		−0.0032 **
		(0.0011)		(0.0030)		(0.0025)
Control country/year	yes	yes	yes	yes	yes	yes
N	720	720	720	720	720	720
adj. R-sq	0.9927	0.9522	0.9915	0.9541	0.9925	0.9479

Note: the t value is in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1.

that, first, we used Model 2 to test the impact of digital service trade on labor productivity. Second, we introduced the multiplier term of DST and labor productivity, and used Model 3 to test the productivity effect. The regression results show that whether the explanatory variable is the total volume of DST or the export level and import level of DST, DST has a positive impact on labor productivity, and is significant at the 1% level. Among them, DST exports have a greater positive effect on labor productivity than imports. After introducing the multiplier term, the regression results show that the multiplier coefficient is negative; that is, the improvement of labor productivity will reduce the positive impact of DST on labor income share. Assumption H2 is verified.

6.2. Structural Effect

In the theoretical analysis, this paper proposes that digital service trade may promote the upgrading of industrial structure by enhancing human capital and promoting R&D investment. Referring to the standard indicators in the existing literature, this paper used the human capital index in PWT10.0 to represent a country’s human capital level, and used the percentage of R&D investment in GDP as a proxy variable of a country’s R&D investment level. As shown in

Table 9. Digital service trade affects labor income share through structural effect.

Dependent Variable	(1)	(2)	(3)	(4)
	Lnhc	Lnrd	Lnstructure	Lnlabshare
lndigital1	0.0080 ***	0.0719 ***		0.0079 ***
	(0.0027)	(0.0213)		(0.0019)
lnhc			0.098 **
			(0.0574)
lnrd			0.0226 ***
			(0.0076)
lndigitall1 × lnstructure				0.0143 ***
				(0.0021)
Control country/year	yes	yes	yes	yes
N	720	720	720	720
adj. R-sq	0.9757	0.9657	0.9681	0.9499

Note: the t value is in parentheses; *** p < 0.01, ** p < 0.05.

, first of all, we examined the impact of DST on human capital and R&D investment. As can be seen from columns 1–2, the DST coefficient is 0.0080, 0.0719, which is significant at 1% level, indicating that DST can promote the improvement of human capital level and R&D investment. Secondly, the third column tests the impact of human capital level and R&D investment on industrial structure upgrading. The regression result is significantly positive, indicating that the increase of human capital and R&D investment can promote industrial structure upgrading. Finally, the fourth column examines the intermediary mechanism of industrial structure upgrading in the impact of DST on labor income share. It can be seen that the interaction coefficient between DST and industrial structure is significantly positive, which means that there is a transmission channel of DST export → industrial structure upgrading → labor income share between DST export and labor income share in industrial structure. Assumption H3 is verified.

6.3. The Distribution Pattern Reshaping Effect

According to the previous theoretical analysis, digital service trade may reduce transaction costs through digital technology advantages, thereby giving play to export scale effect, improving the international market share of its products and services, changing the competitive status of products and markets, and improving a country’s terms of trade. The improvement of the terms of trade will enable a country to obtain a particular monopoly position in the market and product market, so that it has more say in the economic rent distribution in commerce, resulting in the remodeling of the distribution pattern, thus promoting the increase of the country’s labor income share.

This paper further tested the mechanism of the distribution pattern reshaping effect. Referring to the definition of OECD (2020) [16], this paper selected the ratio of a country’s export and import value index to represent its terms of trade level. The regression results are shown in

Table 10. Mechanism of improving terms of trade.

Dependent Variable	(1)	(2)	(3)
	Tradecondition	Tradecondition	Lnlabshare
Total:lndigital1	0.0149
	(0.0136)
Export:Lndigital11		0.1427 ***	0.0862 ***
		(0.0621)	(0.0088)
Import:Lndigital21		0.0074
		(0.0137)
Lndigital1 × lntradecondition			0.0098 **
			(0.0014)
Control country/year	yes	yes	yes
N	720	720	720
adj. R-sq	0.7103	0.7123	0.9471

Note: the t value is in parentheses; *** p < 0.01, ** p < 0.05.

. It can be seen from columns 1–3 that the regression coefficient of total DST and DST imports on terms of trade is insignificant, while DST exports have a significant impact on terms of trade, indicating that DST exports can significantly improve a country’s terms of trade. Specifically, for every 1% increase in DST exports, a country’s terms of trade can improve by 14.27%. It can be seen from the third column that the coefficient of the multiplier term of DST exports and terms of trade is also significantly positive, indicating that the mechanism of DST to improve a country’s share of labor income through improving terms of trade is established. Assumption H4 is verified.

7. Conclusions

This paper empirically analyzes the influence and mechanism of digital service trade (DST) on labor income share by using 48 countries (regions)’ panel data.

The results indicate that: (1) A country’s DST significantly increased its domestic labor income share, and vigorously developing DST will help optimize income distribution. (2) The export of DST is more conducive to the increase of domestic labor income share than the import. The positive impact of DST on labor income share in OECD countries is more significant than that in non-OECD countries. The research results of this paper also show that setting a reasonable DST barrier level will help to exert the distribution pattern reshaping effect and enjoy a new round of digital dividends. (3) DST mainly affects labor income share through productivity effect, structural effect, and distribution pattern remodeling effect. From the perspective of productivity effect, whether via the total volume of DST or the level of exports and imports using DST, DST has a positive impact on labor productivity, and the improvement of labor productivity will reduce the role of DST in improving labor income. From the perspective of structural effect, DST can improve labor income share through industrial structure upgrading. From the standpoint of the reshaping effect of the distribution pattern, the rapid growth of DST contributes to the improvement of the terms of trade, which enables a country to gain a particular comparative advantage in the global market, so that it has more say in the economic rent distribution in business, thus promoting the increase of the country’s labor income share. The research in this paper expands the scope of research on DST and fills the gap of research on DST on labor income share.

According to the research conclusions, we propose the following policy recommendations. Countries should speed up the cultivation of DST competitiveness, accelerate the negotiation process of global digital trade rules, and seize the opportunity to build a new round of digital global value chains. Countries should strengthen international cooperation, further expand the openness of the service industry, accelerate the development of domestic DST, actively participate in bilateral and multilateral negotiations on DST, and build a more inclusive, inclusive and sustainable DST rule system.

China attaches great importance to the development of digital trade, and the scale of China’s DST continues to increase, but its proportion is still lower than the world average. There is still much room for improvement. At the same time, China’s current DST barriers place it in the more than 75 quantile group, which shows that compared with other countries in the sample, China currently has higher DST barriers, which will not be conducive to the development of DST and its economic effects. The negotiation process of global digital trade rules is accelerating, and the construction of the digital global value chain is accelerating. China should deepen international cooperation, further expand the opening of relevant service markets, unswervingly promote higher level opening to the outside world, fully consider the development law of DST, promote the development of emerging service industries, and promote the digital transformation of traditional service industries. It should also pay attention to cultivating the human capital needed for the development of the digital service industry and DST, strengthen the training of digital talents, increase investment in research and development in the field of digital services, encourage enterprises to actively participate in the division of global DST, form advantages of specialization and scale, further develop cross-border e-commerce platforms, and strengthen the construction of logistics and infrastructure supporting DST.

Limited by data acquisition, the main limitation of this paper is that the research conclusions are based on national macro data. The suggestion for future research is that the impact of DST on labor income share can be analyzed based on the micro data of transnational enterprises. Leung and Tang (2023) [60] provide a dynamic equilibrium model and show that when the wage is characterized by a labor contract (rather than market-clearing in the spot labor market in each period), the compensation will display sluggishness. The reduced form dynamics will be changed. Therefore, in future research, on the premise that the data can be obtained, the impact of this factor on the robustness of the results should also be considered and solved.