The rapid global expansion of the digital economy has made it a critical driver for contemporary economic progress. A growing body of literature has examined the role of the digital economy, digital trade, and digital services trade (DST) in shaping economic development outcomes. For instance, Cheng and Huang (2022), Ren et al. (2022), Wang et al. (2022), Zhang et al. (2023), Zainullin et al. (2024), and Wang and Wang (2025) explored the contribution of digital economy to economic growth and green economic transformation. Similarly, Chen et al. (2025), Danish et al. (2023), Wang and Rani (2025), and Bian and Zhang (2025) investigated how digital trade contributes to economic expansion and supports sustainable development. The nexus between the digital economy and income inequality was examined by Wang and Shen (2024); Mulenga and Mayondi (2022) and Yeerken and Feng (2024) analyzed the effects of digital services trade on economic performance and inclusive growth. Empirical findings suggest that DST can help reduce income inequality (Zhu et al., 2022), enhance learning and competitiveness (Goldberg et al., 2009), and lower transaction and trade costs (Buckley, 2009; Humphrey & Schmitz, 2002). Recent evidence also indicates that DST significantly promotes GDP growth in emerging economies (Yeerken & Feng, 2024; Mulenga & Mayondi, 2022), while in OECD countries its impact is often moderated by environmental and regulatory factors (Gao et al., 2024).

Despite these advances, empirical research into how the trade in digital services contributes to economic growth remains relatively limited, particularly for developing countries. This gap indicates the need for systematic and comprehensive analysis of the impact of DST on economic performance in such countries. The present study aims to provide robust empirical evidence of how digital service trade contributes to the economic growth of 87 developing economies.

The contributions of this research can be summarized as follows:

First, this study contributes to the existing literature by empirically investigating the impact of DST on economic growth in developing countries - a topic that has received limited scholarly attention to date. Developing countries constitute a substantial and dynamic segment of the global economy, characterized by accelerating digitalization, expanding internet penetration and increasing participation in cross-border digital exchanges. At the same time, these economies are facing structural challenges, such as inadequate digital infrastructure, institutional inefficiencies and regulatory fragmentation, which constrain the full realization of DST’s growth-enhancing potential. By integrating a comprehensive measure of DST into the theoretical framework of cross-border trade, this study offers novel insights into how the expansion of digital services trade influences economic performance in developing contexts.

Second, the study demonstrates empirically that DST is a significant driver of economic growth in developing countries. It operates through key mechanisms such as generating employment and promoting technological innovation. The analysis reveals that the magnitude of the growth-enhancing effects of DST is conditioned by a number of contextual factors, including the quality of the digital infrastructure, the effectiveness of regulation and the level of human capital. These factors act as critical moderators. This multidimensional framework emphasizes the complex pathways through which the DST influences economic performance. It also shows the importance of the complementary institutional and technical capacities in maximizing the developmental impact of the DST.

Third, by integrating DST within advanced analytical frameworks, this study provides comprehensive and generalizable empirical evidence that deepens the understanding of the nexus between digital trade and economic growth, thereby enriching the literature on digital trade and international economics. It contributes to theoretical development by linking DST to several strands of economic theory. In particular, Trade Theory and Structural Change Theory explain the role of trade expansion and sectoral transformation in driving growth; Endogenous Growth Theory emphasizes the mediating role of employment through human capital accumulation and knowledge spillovers; and Schumpeterian Innovation Theory highlights the contribution of technological innovation to economic development through the mechanism of creative destruction.

Lastly, previous studies have mostly used traditional econometric methods, which can be subject to methodological issues such as multicollinearity, endogeneity and overfitting of models, all of which can undermine the accuracy and reliability of empirical findings, especially when analyzing large datasets. To more effectively capture complex interactions between variables, this study uses advanced machine learning methods, including the Double-Selection LASSO linear regression (DSLR), Partialing-Out LASSO Linear Regression (POLR) and Cross-fit Partialing-Out LASSO Linear Regression (Cross-fit POLR). These approaches improve model accuracy, reduce overfitting and effectively handle high-dimensional data, thus providing a more rigorous and reliable empirical analysis of the factors that influence economic growth.

The study is divided into six main sections. The first section introduces the topic, followed by a literature review in the second section. The third section describes the theoretical framework and model development. The fourth section contains data and explains the empirical estimation strategy. The fifth section presents the findings of the study, and the sixth section draws conclusions and discusses research implications.

The concept of digital trade is widely discussed in the economic literature, covering various aspects such as sharing economy, digital economy, digital transformation and interaction economy. Although the digitalization of the economy is generally recognized as inevitable, it is not fully reflected in national accounting systems (Katz & Koutroumpis, 2013; OECD, 2024). This is largely because the national accounts continue to focus on tangible products, blurring the distinction between trade in public services and trade in digital services, despite the rapid expansion of the latter in recent decades (IMF, 2023). In today’s world, the internet has significantly transformed global economic activity. It has eliminated the importance of geographical distance and strengthened connections between people and businesses at an unprecedented speed and scale. While the economy of the 20th century was primarily driven by the flow of physical goods, today’s economy is increasingly centered around the exchange of intangible goods, specifically the flow of digital services (Neil & Yeung, 2019).

A growing body of empirical literature has examined the economic implications of trade in digital services, though most studies have focused on developed economies and left developing countries relatively unexplored. For example, Zhang et al. (2021) investigated the impact of digital services trade on economic growth across 30 Chinese provinces during the period 2015–2019. Their findings indicate that technologically advanced digital infrastructure, sectoral digital industry development and regional factor productivity proxied by R&D expenditure all have a positive and significant impact on economic growth. Similarly, Wang and Choi (2018) examined the relationship between the digital economy and economic growth in the BRICS countries using data spanning 2000–2016. They concluded that the digital economy positively influenced economic growth in these emerging economies and pointed out that the BRICS nations needed to invest substantially in digital trade infrastructure and Internet facilities in order to maximize the benefits of digitalization. In parallel, Simon and Pingfang (2021) assessed the impact of the digital economy on international trade and economic growth across 53 African countries between 2000 and 2018. Their findings suggest that the digital economy has a positive and significant impact on both trade and growth. This leads the authors to recommend investing heavily in digital technologies as a way to accelerate Africa’s economic development.

Further empirical contributions reinforce these findings. Bakry et al. (2023) and Niebel (2018) examined the relationship between the digital economy, ICT development and economic growth in Jordan and other emerging economies. Both studies found that the digital economy and adoption of ICTs have a positive impact on economic growth, emphasizing the importance of policies that improve digital infrastructure and promote digital literacy. In addition to this, Maune (2019) explored the role of digital services trade in African economies and found that although trade in digital services promotes economic growth, it also reduces the volume of exports of goods. This suggests a complex reallocation effect on trade structures. Additional evidence from India shows that digital services trade has a significant impact on both economic growth and the current account balance. Digital services imports, however, tend to hinder growth in developed countries (Thomas, 2018). Extending the scope of this literature, Pan et al. (2022) revealed that the digital economy contributed positively and significantly to total factor productivity in China. However, their analysis emphasizes that the extent of this effect depends on the level of digital infrastructure development. This indicates that infrastructure remains a crucial prerequisite for realizing the economic benefits of digitalization.

Taken together, these studies establish a consistent empirical foundation suggesting that digital services trade and the broader digital economy positively influence economic growth. However, much of this evidence is derived from developed economies or from a limited set of emerging markets, leaving the dynamics in developing economies underexplored. The existing literature reveals important heterogeneities across countries and regions, particularly regarding the role of infrastructure, the structure of goods versus services trade and the differential impacts between advanced and developing economies. This gap underscores the need for more comprehensive, cross-country analyses of developing economies. Addressing this gap, this study examines the impact of trade in digital services on GDP growth in 87 developing countries, thereby contributing to a relatively under-developed area of literature.

Digital trade can be conceptually grounded in the New Trade Theory (NTT), which provides a framework for understanding international trade in the context of technological disruption and rapidly evolving global data flows (Dirk & Michael, 2012). Unlike traditional trade theories that emphasize comparative advantage, NTT highlights the role of digital trade in helping market participants to operate under monopoly competition, paying particular attention to scale economies, product differentiation and innovation-led trade. The fragmentation of international trade, characterized by the cross-border dispersal of production and service activities, reduces trade costs and improves resource allocation efficiency (Helpman & Krugman, 1985). There is a key difference between traditional trade in services and digital trade, which uses digital technologies to provide services with minimal dependence on physical distance. DST not only expands trade but also facilitates technological diffusion, enhancing firm productivity and fostering innovation. From a growth perspective, total factor productivity (TFP) within the augmented Solow growth model is expected to increase alongside the expansion of the digital economy and digital infrastructure (Zhang et al., 2021; Thomas, 2018; Pan et al., 2022). The model’s incorporation of human capital and technology provides a theoretical foundation for linking digital transformation to long-term growth (Mankiw et al, 1992).

Furthermore, DST impacts GDP both directly and indirectly. Directly, it contributes to economic output by expanding the scope and efficiency of service delivery. Indirectly, DST mediates growth through employment (EMP) and technological innovation (TI). Increased digital trade generates labor demand in both skilled and supporting sectors, raising household incomes and consumption. Simultaneously, participation in DST encourages companies to adopt advanced technologies and continuously innovate, enhancing productivity and supporting long-term economic growth. By integrating NTT with the augmented Solow growth framework, this study conceptualizes DST as a critical driver for GDP growth in developing countries, operating through multiple channels that combine trade expansion, human capital utilization, and technological advancement. The definition of the growth model can be found in Solow (1957) and Mankiw et al. (1992).

$Y_{it}=A_{it}\cdot K_{it}^{\alpha }\cdot H_{it}^{\beta }\cdot L_{it}^{\theta }$ (1)

where Yt is GDP over time (t) in response to physical capital (K), human capital (H), labor (L), and technology (A) over time t and country i. It is necessary to note that L and H are not the same. While human capital refers to the skills acquired through education, training, and experience, labor implies the abilities that people naturally possess (Mankiw et al, 1992).

In contemporary economies, DST enhances productivity by improving information flows, reducing transaction costs and enabling innovation (Niebel, 2018). So DST can be modeled as a determinant of TFP, making A_t a function of DST:

$A_{it}=A_0{e}^{{\delta }_{1}LnDS{T}_{\text{it}}}$ (2)

By substituting A_it from Eq. (2) into Eq. (1) and taking the natural logarithm of both sides, the growth model can be expressed as a function of DST as follows:

$\mathrm{Ln}{Y}_{it}=\mathrm{Ln}{A}_0+{\delta }_1\mathrm{Ln}DS{T}_{it}+\mathrm{Ln}{K}_{it}^{\alpha }+\mathrm{Ln}{H}_{it}^{\beta }+\mathrm{Ln}{L}_{it}^{\theta }$ (3)

Next, by incorporating control variables (X_it) as well as country (μ_i) and time (π_t) fixed effects, the baseline model can be formulated as follows:

$\mathrm{LnGDP}{P}_{it}={\delta }_0+{\delta }_1{\mathrm{LnDST}}_{\mathrm{it}}+{\delta }_k\sum _{k=2}^n\mathrm{Ln}{X}_{it}+{\mu }_i+{\pi }_t+{\epsilon }_{it}$ (4)

where LnGDP_it = LnY, δ₀ = LnA₀ is the constant term, δ₁ is the slope coefficient for DST, and δ_k shows the impact of all control variables of Eq. (4), respectively.

Finally, by taking the partial derivative of LnGDP with respect to LnDST, the marginal effect of DST on economic growth can be derived as follows:

$\underset{\text{Change in unit}}{\underbrace{\frac{\partial GD{P}_{it}}{\partial DS{T}_{it}}={\delta }_1\frac{GD{P}_{it}}{DS{T}_{it}}}}\text{or}\underset{\text{Change in \%}}{\underbrace{\frac{\partial GD{P}_{it}/GD{P}_{it}}{\partial DS{T}_{it}/DS{T}_{it}}={\delta }_1}}$ (5)

Building upon the theoretical foundations discussed above, this study formulates the following research hypotheses:

H₁: Digital Service Trade (DST) has a significant positive impact on economic growth in developing countries.

H₂: Employment and technological innovation mediate the relationship between DST and economic growth.

H₃: Digital infrastructure, human capital, and regulatory quality moderate the relationship between DST and economic growth.

This study employs panel data spanning the period 2005–2023 for 87 developing economies. All variables are obtained from reliable international sources, specifically the World Development Indicators (WDI) and UNCTAD databases. For analytical clarity, the selected variables are categorized as follows:

Explained variables: In this paper, per capita GDP is employed as the primary dependent variable, as it is widely recognized as a reliable indicator of economic performance. This approach is consistent with prior studies, including Yeerken and Feng (2024) and Mulenga and Mayondi (2022), which similarly utilize per capita GDP to capture variations in economic outcomes.

Core explanatory variable: Digital services trade (DST) encompassing both digital services exports and imports refers to the delivery of services through digital channels. The United Nations Conference on Trade and Development (UNCTAD) employs a primary product classification system to identify services that can be transmitted online across borders, thereby distinguishing between digital and non-digital service transactions. Using statistical data and computational methods, UNCTAD has further estimated the overall volume of digital services trade across countries. Building on the frameworks established by Yeerken and Feng (2024), and Mulenga and Mayondi (2022), this study adopts DST delivery indicators as the primary explanatory variable to evaluate the extent of digital services trade development in a country.

Control variables: Drawing on prior research, including Yeerken and Feng (2024), and Mulenga and Mayondi (2022), this study incorporates a set of control variables, namely goods imports (GIM), goods exports (GEX), trade openness (TO), structural change (SCH), and population (POP). These variables are widely recognized as significant determinants of GDP growth dynamics and are therefore included as essential control factors in the present analysis.

Mediating variables: In this study, the level of employment (EMP) and technological innovation (TI) are used as the main mediating variables, with data obtained from the World Development Indicators (WDI). Yeerken and Feng in their 2024 paper also use these factors as mediators. The impact of DST on GDP can be mediated through EMP and TI. By fostering the expansion of the digital industry, it generates new job opportunities, which increase household income and stimulate aggregate demand. This, in turn, supports economic growth. It also facilitates cross-border knowledge transfer and encourages firms to adopt advanced technologies, which increases productivity and competitiveness. These channels illustrate how the EMP and TI work together to help DST translate into higher GDP.

Moderating Variables: The Digital Infrastructure Index (DII) is constructed by combining three key indicators of digital connectivity and technological capability: internet users (% of the population), mobile broadband subscriptions (per 100 people), and secure internet servers (per million people). These elements together determine a country’s ability to support and promote digital trade. Human capital of the right quality, as measured by the average number of years of schooling among adults aged 15 and older, is essential for improving digital skills, fostering innovation and effectively using digital technologies. Regulatory quality (REQ), represented by the Regulatory Quality Estimate from the Worldwide Governance Indicators, reflects the capacity of governments to design and implement sound policies that foster digital trade and economic growth.

Table 1 gives a detailed overview of the selected variables; Fig. 1 depicts the annual trends of GDP and DST across 87 developing economies from 2005 to 2023. The figure indicates that GDP experienced a decline during certain periods (2020), whereas LnDST peaked in 2021, likely reflecting the accelerated reliance on digital services during the COVID-19 pandemic.

Fig. 1. 

Annual rise of LnGDP and LnDST for 87 developing countries from 2005 to 2023.

Table 2 presents the descriptive statistics and correlation coefficients for the variables included in this study. As reported in Panel A, SCH records the highest mean value, whereas DST exhibits the lowest. In terms of dispersion, GEX shows the highest standard deviation, while TO demonstrates the lowest, reflecting notable differences in variability across the variables. Panel B reports the correlation coefficients, where DST and GDP display a modest positive correlation of 4.3%. Among the control variables, SCH exhibits the strongest positive association with GDP, whereas POP shows the strongest negative correlation. These preliminary observations indicate that structural change may exert a more substantial influence on GDP growth, and the relatively modest correlation between DST and GDP suggests that the effects of DST may operate indirectly through mediating mechanisms such as EMP and TI or be conditioned by moderating factors such as DII, HC, and REQ.

Before applying the advanced machine learning techniques, we first assessed several preliminary statistical tests, including cross-sectional dependence (CSD) among the variables, stationarity tests, cointegration tests, and machine learning regularization. Table 3 presents cross-sectional dependency (CSD) and panel unit root tests, focusing on various economic variables. The CSD tests, as indicated by the BP_LM and PS_LM statistics, reveal a strong dependency among the variables, with significant values across the board, particularly for GDP (4.9E+4) and DST (1.5E+4), suggesting a high degree of correlation among the cross-sectional units in the dataset. The Pesaran-CD test also supports this dependency.

In the unit root analysis, the CADF and CIPS tests are used to assess stationarity. Most variables, including GDP and DST, GIM, GEX and TO, display non-stationarity at the level, with negative test statistics that do not exceed critical values. However, upon first differencing, all variables become stationary. This analysis suggests that while the variables exhibit cross-sectional dependence, they require differencing to achieve stationarity, a prerequisite for reliable econometric modeling.

In Table 4, the Westerlund cointegration test is employed to examine the presence of long-run equilibrium relationships between the dependent and independent variables. The results of the G_t, P_t, and P_a statistics confirm a significant long-term association among the variables, whereas the G_a statistic does not provide sufficient evidence to support the existence of a long-run effect of DST on GDP growth.

Table 5 presents the variable selection procedure and lists all selected variables based on the Standard LASSO, Adaptive LASSO and ElasticNet LASSO estimators. The optimal alpha values indicate the level of regularization used by each model. The Standard LASSO uses the weakest regularization, the Adaptive LASSO applies moderate regularization, and the Elastic-Net LASSO employs the strongest regularization. Despite the differences in alpha, all models yield similar predictive performance, with Mean Squared Errors (MSE) around 0.057. However, Adaptive LASSO emerges as the best model, striking a balance between effective regularization and minimal prediction error, evidenced by its lowest MSE and more refined variable selection through adaptive weighting.

Figs 4, 5, 6 illustrate the coefficient path plots and cross-validation plots for the Standard, Adaptive, and ElasticNet LASSOs, respectively. The coefficients path plots display the variability of each variable’s coefficient in response to changes in the regularization factors, showing how coefficients approach zero as the optimal lambda increases. The cross-validation plots highlight model performance across different values of the regularization factors, with the lowest MSE (green line) and the best lambda (red dashed line), clearly marked. Similarly, other dashed lines represent the individual cross-validation folds.

Fig. 4. 

The coefficient path plots (left) and the Cross-validation plot (right) based on the Standard LASSO algorithm.

Fig. 5. 

The coefficient path plots (left) and the Cross-validation plot (right) based on the Adaptive LASSO algorithm.

Fig. 6. 

The coefficient path plots (left) and the Cross-validation plot (right) based on the ElasticNet LASSO algorithm.

After conducting preliminary panel data analysis and applying LASSO regularization techniques to variable selection, this study uses inferential LASSO-based machine learning methods — namely DSLR, POLR, and Cross-fit POLR — to estimate the long-run impact of DST on GDP. The results are reported in Table 6. Columns (1), (3), and (5) present the impact of DST on GDP without including control variables, whereas Columns (2), (4), and (6) display the results after accounting for control variables. Among these approaches, the Cross-fit POLR model stand outs, as it is built on a robust LASSO framework and shows superior performance in capturing temporal dynamics and minimizing model selection bias. The results reveal that all selected variables exert a statistically significant influence on GDP in developing economies. Specifically, holding other factors constant, a 1% increase in DST leads to an estimated 0.034% increase in GDP.

The positive and significant effect of DST on GDP growth seems both theoretically sound and practically justified. According to the economic concept of comparative advantage, countries benefit from expertise in efficient production; the growth of DST is considered a form of efficient production (Zhou & Dahal, 2024; Tetteh, 2024). Similarly, the theory of endogenous growth emphasizes that the development of technology and knowledge leads to economic growth (Helpman & Krugman, 1985; Etro, 2023). Besides, the diffusion of technology helps to boost productivity, allowing local firms to operate with higher industrial added value (Nucci et al., 2023; Foster & He, 2022).

Moreover, the impacts of GIM and GEX on GDP are also positive, indicating that higher trade volumes contribute to GDP growth. POP presents a negative coefficient of -0.586, meaning diminishing returns to GDP growth with larger populations, which may complicate resource management. The SCH has a coefficient of 0.159, indicating that structural change plays a significant positive role in this relationship at the 1% significance level.

Table 9 presents the results of the heterogeneity analysis based on income levels, digital infrastructure, and regulatory quality, offering insights into how the impact of DST on GDP varies across different country groups. Specifically, Columns (1), (3) and (4) show that low-income (LI), upper-middle-income (UMI), and high-income (HI) countries exhibit positive and statistically significant coefficients of 0.017, 0.017, and 0.095, respectively. These findings indicate that higher levels of DST are positively associated with GDP growth in these economies. In contrast, the results in Column (2) reveal that lower-middle-income (LMI) countries experience a negative but statistically insignificant relationship between DST and GDP.

Furthermore, as reported in Column (5), DST exerts a positive and statistically significant effect on GDP growth in countries with high digital infrastructure levels (HDIL). However, the effects of DST on GDP in countries with low digital infrastructure levels (LDIL), as well as in those with high or low regulatory quality, remain positive but statistically insignificant. These findings highlight the critical role of digital infrastructure in enhancing the economic benefits of DST adoption across different development contexts.

The empirical findings of this study provide robust evidence that DST exerts a positive and significant impact on GDP growth across 87 developing economies, corroborating both theoretical and practical expectations (Gao et al., 2024; Xiong & Luo, 2023). Using advanced ML techniques, particularly the Cross-fit POLR model, we find that a 1% increase in DST is associated with a 0.034% increase in GDP, highlighting the economic relevance of DST as a driver of economic performance. This result supports H₁ and aligns with previous empirical studies, including those by Yeerken & Feng (2024), Mulenga & Mayondi (2022), Gao et al. (2024) and Xiong & Luo (2023). Additionally, the positive effect of DST is consistent with classical economic theory, which emphasizes gains from comparative advantage and production efficiency (Zhou & Dahal, 2024; Tetteh, 2024). It is also in line with endogenous growth theory, which highlights the importance of technological advancement and knowledge diffusion in fostering long-run economic expansion (Helpman & Krugman, 1985; Etro, 2023; Nucci et al., 2023; Foster & He, 2022).

The mechanism analysis further underscores the mediating roles of employment and technological innovation in translating DST into economic growth. Consistent with the empirical evidence of Yeerken & Feng (2024), Jin & Lu (2024), and Ndubuisi et al. (2021), increased participation in digital trade creates additional labor demand and employment opportunities, raising household income and consumer spending. At the same time, engagement with digital services encourages companies to adopt advanced technology and pursue continuous innovation. This enhances productivity and supports sustainable economic growth (Wen et al., 2023). These findings confirm H₂ and align with the theoretical framework that identifies labor-market dynamics and technological development as critical transmission channels through which the benefits of DST materialize (Zhang et al., 2021; Simon & Pingfang, 2021).

The endogeneity analysis, which uses lagged forms of DST and the interaction between the one-period lag of Internet-user rates and the penetration rate of fixed telephone lines in 1984 as instrumental variables, demonstrates that the main findings are not affected by endogeneity concerns. The results of the POIVLR, Cross-fit POIVLR, and 2SLS models consistently confirm the validity of the IV strategy. The 2SLS estimates further support H₃, which is in line with Rodriguez-Crespo et al. (2019), who also found a positive and significant relationship between Internet use and trade in both developing and developed economies.

The heterogeneity analysis reveals that the impact of DST on GDP is contingent upon countries’ income levels, digital-infrastructure readiness and regulatory-quality conditions. The income-based heterogeneity results show that low-income and upper-middle-income economies experience modest positive effects, whereas high-income developing countries benefit the most from DST. Conversely, lower-middle-income countries display no statistically significant effect, suggesting potential barriers to harnessing digital trade, such as limited digital infrastructure, skill shortages, or regulatory constraints. The heterogeneity analysis based on digital infrastructure and regulatory quality further indicates that only economies with high levels of digital infrastructure experience positive and significant effects of DST on GDP. In contrast, countries with low digital infrastructure, as well as those with both high and low regulatory quality, do not exhibit significant effects. Collectively, these findings suggest that the capacity to benefit from digital trade is determined by structural factors, including digital readiness, institutional quality and the broader enabling environment.

Finally, the results highlight the reinforcing roles of traditional trade and structural transformation in shaping GDP outcomes. Goods exports (GEX) and goods imports (GIM) both contribute positively to economic growth, while structural change (SCH) emerges as a significant driver, suggesting that broader economic transformation complements the growth-enhancing effects of digital trade. The negative association between population (POP) and GDP may reflect diminishing returns in more populous countries, underscoring the need for effective resource allocation and targeted policies to maximize the benefits of digitalization.

This study examines the role of DST in fostering economic growth across 87 developing economies from 2005 to 2023. Using state-of-the-art ML techniques, including the Cross-fit POLR model, the findings consistently demonstrate that DST exerts a positive and statistically significant impact on GDP. These results reinforce theoretical expectations from both classical and endogenous growth frameworks, underscoring the importance of digitalization, technological advancement and knowledge diffusion as engines of economic expansion.

The mechanism analysis further reveals that employment generation and technological innovation serve as key channels through which DST contributes to economic performance. Digital engagement enhances labor market opportunities and encourages firms to adopt advanced technologies, which improve productivity and support growth. These findings highlight the critical role of human capital development and innovation systems in translating digital trade participation into tangible economic gains. Additionally, the endogeneity analysis demonstrates that the main findings are not affected by endogeneity concerns.

Heterogeneity analyses show that the benefits of DST are not uniform across countries but depend heavily on structural characteristics such as income level, digital infrastructure and regulatory quality. High-income developing economies and those with advanced digital infrastructure enjoy the strongest gains, while lower-middle-income countries and those with weak infrastructure benefit less. This underscores the necessity of targeted policy interventions aimed at strengthening digital connectivity, improving regulatory environments and enhancing digital readiness to ensure more inclusive participation in the digital economy.

Traditional trade flows and structural transformations continue to play a reinforcing role in shaping economic outcomes. This suggests that digital trade can complement, rather than replace, broader development strategies. The negative association observed between population size and GDP highlights the importance of effective resource allocation and policies that allow economies to benefit from demographic trends.

The findings of this study, indicate that, first, developing countries should officially recognize digital services trade (DST) as a strategic driver for economic growth. This can be achieved through the establishment of dedicated agencies to monitor DST expansion. These agencies should provide targeted incentives, such as tax breaks for digital service providers. They should also set clear, measurable goals for DST’s contribution to the GDP. Second, based on the results of the mechanism analysis, policymakers should consider investing in the development of digital skills for the workforce, vocational training, and research and development of digital services in order to improve productivity, labor market outcomes and innovation at the firm level. Third, according to the heterogeneity analysis results, governments should prioritize expanding broadband networks, modernizing customs procedures, promoting digital literacy and implementing regulatory reforms to balance foreign investment and domestic protections. These measures will enhance countries’ capacity to participate effectively in digital trade and reduce structural barriers that limit growth potential in lower-income or lower-infrastructure economies. Finally, in line with the moderation analysis, policymakers should prioritize investments in nationwide broadband expansion, ICT infrastructure and digital connectivity to strengthen the technological backbone of the economy. Simultaneously, targeted programs aiming to improve digital literacy, vocational training and higher education in the field of ICT and related areas can build a human capital that is capable of fully exploiting digital trade opportunities.

This study is not without its limitations. First, the availability of the underlying data may pose challenges, as the accuracy and representativeness of the databases used can affect the validity of the empirical results. Second, although the analysis identified significant associations between key variables, the observational nature of the study limited the ability to make definitive causal inferences. Third, the findings of this study may be context-dependent, and caution should be exercised when attempting to generalize the results to other geographic regions or economic contexts, given potential differences in structural, institutional and developmental factors. Finally, the measurement of digital trade (DST) relies exclusively on UNCTAD’s BoP-based aggregates, which, in line with concerns raised by the OECD–WTO–IMF Handbook on Measuring Digital Trade (2023), may conflate digitally delivered services with digitally enabled but non-digital services. This measurement constraint introduces the risk of classification error, which could affect the accuracy of the estimated relationship between DST and GDP.