Digital Trade Liberalization Under Low Tariff Regimes: Heterogeneous Effects of Information and Communication Technologies in Vietnam

Abstract

This paper examines how the composition of digital trade liberalization shapes trade adjustment patterns in a low-tariff environment. While digital technologies are often treated as a unified category, different components may generate distinct economic effects and policy implications. Using Vietnam as a case study, the analysis distinguishes between information technologies (ITs), which are embedded in production processes, and communication technologies (CTs), which primarily reduce coordination costs. A partial equilibrium simulation is conducted using the WITS-SMART model based on 2023 trade and tariff data. The results show that tariff elimination in IT products leads to a substantially stronger import response and a larger reduction in tariff revenue compared to CT. However, the overall magnitude of these effects remains modest, reflecting Vietnam’s already low tariff structure. The findings highlight the importance of accounting for heterogeneity within digital technologies when assessing trade liberalization in emerging economies.

1. Introduction

Over the past three decades, the organization of production has become increasingly fragmented across borders, with global value chains enabling countries to specialize in particular stages of the production process (Feenstra, 1998; De Backer & Miroudot, 2013). For emerging economies such as Vietnam, integration into GVCs has supported rapid trade expansion and industrial upgrading. Recent developments, including geopolitical tensions and technological change, have altered the structure of global production networks (Antràs, 2021; Rodrik, 2018).

Among the forces reshaping GVCs, digitalization plays a complex and often ambiguous role. On the one hand, advances in digital infrastructure reduce trade and coordination costs, making it easier to manage fragmented production networks across borders. On the other hand, technologies such as automation, artificial intelligence, and advanced manufacturing systems are altering the organization of production itself, potentially reducing the reliance on labor-intensive activities that have traditionally supported industrialization in developing economies. The coexistence of these two dynamics creates a tension that is not easily resolved in policy terms.

A key limitation in the existing literature is that information and communication technology (ICT) is often treated as a single, uniform driver of trade integration. While influential contributions have emphasized the role of digital technologies in enabling the fragmentation and coordination of production (Baldwin, 2013; Elms & Low, 2013), relatively little attention has been paid to the possibility that different components of ICT may generate distinct economic effects. As a result, the implications of digital trade liberalization are often discussed in aggregate terms, which may obscure important differences in how technologies interact with production structures.

This paper addresses this gap by distinguishing between two broad components of ICT based on their functional roles. Communication technologies primarily reduce coordination costs and facilitate the management of dispersed production processes, whereas information technologies are more directly embedded in production and are associated with automation and technological upgrading. This distinction suggests that digital trade liberalization may not generate uniform outcomes across technology groups.

Focusing on Vietnam, the study examines how tariff liberalization in these two categories affects trade adjustment patterns. The analysis employs a partial equilibrium simulation using the WITS-SMART model, based on bilateral trade and tariff data for 2023. By simulating the removal of tariffs on information and communication technology products separately, the study compares their effects on import-side adjustments, trade creation and diversion, and welfare outcomes.

Recent trade developments further underscore the relevance of this analysis. Vietnam’s imports increased from approximately USD 237 billion in 2018 to over USD 327 billion in 2023, reflecting sustained demand for intermediate and capital goods (United Nations, 2024). ICT-related imports remain concentrated in machinery, electronics, and communication equipment, indicating that participation in global value chains is closely tied to access to imported technological inputs. In this context, the effects of digital trade liberalization cannot be evaluated solely in terms of aggregate trade expansion but must also be understood in relation to changes in import dependence and production structure.

Rather than treating ICT as a uniform driver of trade integration, this paper examines how different components of digital technology generate distinct adjustment patterns and policy trade-offs. The analysis suggests that digital trade liberalization may raise sequencing considerations, in the sense that different components are associated with distinct adjustment patterns and policy trade-offs. Measures that enhance connectivity tend to reinforce existing production networks with relatively limited disruption, whereas those that support technological upgrading are associated with stronger import responses and higher fiscal costs.

The remainder of the paper is structured as follows. Section 2 reviews the relevant literature. Section 3 presents the methodology. Section 4 reports the results. Section 5 discusses the findings. Section 6 concludes.

2. Literature Review

The relationship between digital technologies and international trade has been widely examined in the context of global value chains. Early contributions emphasize the role of ICT in reducing coordination costs and enabling the fragmentation of production across borders (Baldwin, 2013; Elms & Low, 2013). This perspective builds on broader GVC literature, which highlights how trade costs, firm capabilities, and task specialization shape international production networks (Feenstra, 1998; Cattaneo et al., 2010; De Backer & Miroudot, 2013).

GVC participation is also associated with upgrading opportunities and structural transformation. Access to imported intermediate inputs and capital goods is particularly important for developing economies seeking to move into higher-value production segments (Sturgeon & Gereffi, 2009; Barrientos et al., 2011; Rodrik, 2018).

Recent research suggests that digital technologies are not homogeneous. Empirical evidence indicates that ICT development can enhance GVC participation through both backward and forward linkages (Peng, 2022). At the same time, advances in digital capital goods, including automation and AI-related technologies, may reshape comparative advantages and alter development trajectories (Rodrik, 2018).

Empirical work in trade policy analysis has long relied on partial equilibrium frameworks to capture product-level adjustment. The SMART model, widely implemented through WITS, draws on this tradition and allows trade responses to be decomposed into trade creation and trade diversion under the Armington assumption (Laird & Yeats, 1986; Francois & Hall, 2003; Armington, 1969).

Despite extensive applications, existing studies typically treat ICT as a single category. However, product-based classifications from OECD and UNCTAD show that ICT goods include multiple subgroups with distinct economic roles (OECD, 2011; UNCTAD, 2018). This suggests that the effects of liberalization may differ across ICT components.

This paper contributes by combining a SMART simulation framework with a functional disaggregation of ICT, allowing heterogeneous trade responses to be identified.

In addition to the general literature on ICT and global value chains, a growing body of empirical work has applied partial equilibrium models, particularly the SMART framework, to assess the impact of trade liberalization at the sectoral level. Early applications by Laird and Yeats (1986) established the methodological foundation for trade policy simulations, which was later operationalized in the SMART model and further developed by Francois and Hall (2003). These studies demonstrate the usefulness of partial equilibrium approaches in capturing detailed product-level adjustments under tariff changes.

More recent applications of the SMART framework provide useful points of comparison. Pasara (2021), for example, examines trade creation and diversion effects in the context of African regional integration, while Zhang et al. (2024) analyze trade pattern adjustments under the Regional Comprehensive Economic Partnership (RCEP). These studies generally find that the magnitude of trade effects depends strongly on initial tariff levels and sectoral composition.

In the case of Vietnam, existing research has primarily focused on aggregate trade dynamics or the role of ICT in global value chain participation (Dang, 2024). Less attention has been paid to the internal composition of ICT and the possibility that different technological components may generate distinct trade responses. This gap motivates the present study.

3. Methodology

This study employs a partial equilibrium simulation to examine the effects of digital trade liberalization in Vietnam. The analysis is conducted using the SMART model implemented in the World Integrated Trade Solution (WITS) platform. The SMART framework is widely used in applied trade analysis to evaluate the impact of tariff changes on trade flows, tariff revenue, and welfare at a detailed product level.

3.1. Model Framework

The SMART model evaluates the effects of tariff reductions by decomposing import adjustment into trade creation and trade diversion under the Armington assumption of imperfect substitution among exporters (Armington, 1969). The model builds on partial equilibrium trade policy simulation frameworks developed by Laird and Yeats (1986) and further applied in industry-level trade policy simulations by Francois and Hall (2003). It is implemented in the World Integrated Trade Solution (WITS) platform developed by the World Bank (2024).

In this framework, Vietnam is treated as the importing market, and the analysis focuses on import-side adjustments.

The total change in imports can be expressed as:

$${\Delta M}_{\mathit{ijk}} = {\mathit{TC}}_{\mathit{ijk}} + {\mathit{TD}}_{\mathit{ijk}}$$

(1)

where ΔM_ijk denotes the change in Vietnam’s imports of product i from trading partner j under scenario k, TC_ijk is the trade creation effect, and TD_ijk is the trade diversion effect.

Trade creation reflects the expansion of import demand resulting from lower domestic prices after tariff reduction and can be expressed in simplified form as:

TC_ijk = M_ijk · η_i · (ΔP_ijk/P_ijk)

(2)

where M_ijk is the initial import value, η_i is the import demand elasticity for product i, and ΔP_ijk/P_ijk is the proportional change in import prices associated with tariff reductions.

Trade diversion captures the reallocation of imports across foreign suppliers when relative prices change and may be represented as:

$${\mathit{TD}}_{\mathit{ijk}} = M_{\mathit{ijk}} · {\sigma }_i · (\Delta R_{\mathit{ijk}}/R_{\mathit{ijk}})$$

(3)

where σ_i is the elasticity of substitution across exporters and ΔR_ijk/R_ijk denotes the change in relative prices across supplying countries.

At the aggregate level, the import effect for each scenario is obtained by summing across products and trading partners:

$${\Delta M}_k = {\Sigma }_i {\Sigma }_j ({\mathit{TC}}_{\mathit{ijk}} + {\mathit{TD}}_{\mathit{ijk}})$$

(4)

The numerical simulation is implemented in the WITS-SMART system using the default elasticity parameters embedded in the platform. Accordingly, the equations above provide a conceptual representation of the model structure, while the reported results are generated directly through the SMART simulation algorithm.

3.2. Welfare Clarification

Within the SMART framework, welfare gains primarily reflect changes in consumer surplus resulting from price reductions following tariff elimination. Changes in sourcing patterns may capture limited terms-of-trade effects, but these remain secondary in a partial equilibrium setting. The welfare estimates should therefore be interpreted as partial-equilibrium measures rather than economy-wide welfare changes (Francois & Hall, 2003).

3.3. Data Sources and Coverage

The simulation is based on Vietnam’s bilateral import data at the HS 6-digit level for the year 2023. Trade data are obtained from the UN Comtrade database, while tariff data are drawn from the TRAINS and WTO databases as integrated within the WITS platform.

The choice of 2023 as the benchmark year reflects a more recent trade structure following major global disruptions, including the COVID-19 pandemic and increased geopolitical tensions. This allows the analysis to capture the effects of digital trade liberalization in a relatively mature and low-tariff environment.

To provide broader context, aggregate trade trends are examined using UN Comtrade data for the period 2018–2024. These data indicate a sustained increase in imports of intermediate and capital goods, particularly in technology-related sectors, highlighting the importance of digital inputs in Vietnam’s participation in global value chains.

3.4. Classification of Technology Groups

A key step in the analysis is the classification of ICT products into information technology (IT) and communication technology (CT) groups. This distinction follows the functional perspective proposed by Baldwin (2013) and Elms and Low (2013), which differentiates between technologies that primarily reshape production processes and those that reduce coordination costs across fragmented production networks.

IT products include machinery, equipment, and electronic components embedded in production processes, often associated with automation, upgrading, and structural transformation. CT products, by contrast, include telecommunications equipment and related technologies that facilitate information exchange, logistics, and coordination across geographically dispersed production stages.

The classification is implemented using HS-based product groupings consistent with the SMART framework. It is intended as an analytical distinction rather than a formal industry classification. To reduce classification risk, the grouping is cross-checked against international ICT goods classifications, particularly the OECD definition of ICT goods and the UNCTAD correspondence table for HS-based ICT classifications (OECD, 2011; UNCTAD, 2018). The detailed product mapping is provided in Appendix A.

The classification remains imperfect. Some HS headings, such as electric motors or ignition equipment, may be viewed as conventional industrial goods rather than narrowly digital products. They are retained in the IT group only when their role in the simulation is interpreted as production-embedded electrical or electronic components within broader manufacturing systems.

This does not imply that these goods are inherently digital. Rather, their inclusion reflects their function within production systems that increasingly integrate electronic and digital elements. Alternative classification approaches, such as the OECD ICT goods list, could produce different groupings. However, given the need for consistency with the SMART framework and HS-based tariff data, the present approach prioritizes analytical coherence over strict technological delineation. The sensitivity of the results to alternative classifications is therefore acknowledged as a limitation and a direction for future robustness analysis.

3.5. Simulation Design

Two policy scenarios are constructed to examine the effects of digital trade liberalization.

• Scenario 1 (IT liberalization): import tariffs applied by Vietnam on IT products are reduced to zero.
• Scenario 2 (CT liberalization): import tariffs applied by Vietnam on CT products are reduced to zero.

These scenarios are designed to isolate the effects of tariff liberalization across different components of digital technology. The model generates estimates of import changes, trade creation, trade diversion, tariff revenue effects, and welfare gains for each scenario.

The results are analyzed both at the aggregate level and across major trading partners to identify differences in adjustment patterns.

3.6. Limitations

The SMART model provides a useful framework for analyzing the trade and welfare effects of tariff changes; however, several limitations should be acknowledged.

First, as a partial equilibrium model, it focuses on sector-specific adjustments and does not capture broader general equilibrium effects, such as interactions across sectors or changes in factor markets (Francois & Hall, 2003).

Second, the results depend on the Armington assumption of imperfect substitution and on the elasticity parameters embedded in the model, which may not fully reflect country-specific conditions (Armington, 1969).

Third, the simulation considers tariff changes only and does not account for non-tariff measures, regulatory barriers, or dynamic effects such as technological learning and investment responses.

Finally, the analysis is based on a single benchmark year and therefore provides a static representation of trade structure.

Despite these limitations, the model remains appropriate for identifying the direction and relative magnitude of trade adjustments associated with different components of digital trade liberalization.

4. Results

This section presents the results of the SMART simulations for the two liberalization scenarios, focusing on import-side adjustments in Vietnam. The analysis examines the magnitude of import changes, the decomposition into trade creation and trade diversion, and the associated effects on tariff revenue and welfare.

4.1. Overall Import Effects

Table 1. Summary of SMART Simulation Results for Vietnam, 2023 (thousand USD).

Indicator	Scenario 1: IT Liberalization	Scenario 2: CT Liberalization
Imports before simulation	287,427,000	169,675,000
Import Change	+583.201	+81.432
Trade Creation	+583.160	+81.424
Trade diversion	−0.001	+0.011
Tariff Revenue Change	−540.214	−66.198
Welfare gain	+29.899	+3.191

Source: Author’s calculations based on WITS-SMART simulation results.

reports the aggregate simulation results. Both scenarios lead to an increase in imports, although the magnitude differs substantially. IT liberalization produces a much stronger import response than CT liberalization.

However, the overall scale of the effects remains small relative to Vietnam’s import base. This reflects the already low tariff structure of ICT goods, consistent with international commitments such as the WTO Information Technology Agreement.

Under IT liberalization, imports increase by approximately USD 583.2 thousand, compared with USD 81.4 thousand under CT liberalization. These changes are small in absolute terms. This is not unexpected. When tariffs are already low, further reductions have limited scope to generate large price effects. In this setting, the adjustment occurs at the margin, through shifts in sourcing rather than expansion in overall demand.

This pattern is consistent with the broader trade literature, which suggests that the marginal impact of tariff reductions declines as tariff levels approach zero (Wacziarg & Welch, 2008). As a result, the policy relevance of the simulation lies less in the absolute scale of import changes and more in the comparative structure of adjustment across technology groups. This reinforces the importance of examining how different technology groups respond, rather than focusing solely on aggregate effects.

4.2. Trade Creation and Diversion

The decomposition of import changes shows that trade expansion in both scenarios is driven almost entirely by trade creation rather than trade diversion. Under IT liberalization, trade creation accounts for virtually the entire increase in imports, while trade diversion is negligible and slightly negative. A similar pattern is observed under CT liberalization, where trade creation dominates, and diversion effects remain very small.

Although aggregate diversion effects are limited, the partner-level results reveal differences in sourcing patterns.

Table 2. Selected Partner-Level Trade Effects under IT and CT Liberalization, 2023 (thousand USD).

Partner	IT Scenario: Trade Total Effect	CT Scenario: Trade Total Effect
Taiwan, China	+452.666	+25.846
Korea, Rep.	+195.727	+11.946
United States	+39.563	+6.668
India	+20.293	+3.378
Germany	+10.747	+8.771
China	−75.708	+35.358
Japan	−18.516	−3.583
Thailand	−9.091	−1.978
Indonesia	−10.220	−3.984
Malaysia	−6.018	−3.181

Source: Author’s calculations based on WITS-SMART simulation results.

reports selected trading partners with notable changes.

Under IT liberalization, the largest positive effects are concentrated among major technology suppliers, including Taiwan (China), Korea, the United States, and India. At the same time, several regional partners such as Japan, Thailand, Indonesia, and Malaysia experience reductions in trade flows.

In contrast, CT liberalization generates smaller and more contained adjustments. Positive effects remain concentrated among key suppliers, but the magnitude of both positive and negative changes is lower. These results indicate that IT liberalization is associated with stronger reallocation toward technologically advanced suppliers, whereas CT liberalization leads to more limited adjustments within existing trade networks.

4.3. Revenue and Welfare Effects

The fiscal and welfare implications differ across the two scenarios. IT liberalization leads to a reduction in tariff revenue of approximately USD 540.2 thousand, compared with USD 66.2 thousand under CT liberalization. In both cases, welfare gains are positive but modest, amounting to USD 29.9 thousand for IT and USD 3.2 thousand for CT.

These results indicate that, in a context where tariffs are already relatively low, further liberalization produces limited welfare gains relative to the scale of existing trade. However, the comparison between the two scenarios remains informative. IT liberalization involves both a stronger import response and a larger fiscal cost, whereas CT liberalization yields smaller and more contained adjustments.

Taken together, the findings show that digital trade liberalization generates different adjustment patterns across technology groups, even when the overall magnitude of the effects remains limited.

These results provide the basis for interpreting the broader policy implications of digital trade liberalization, which are discussed in the next section.

5. Discussion

The relatively small magnitude of the simulated effects is consistent with findings from previous SMART-based studies. For example, Pasara (2021) and Zhang et al. (2024) report similarly modest adjustments in settings where tariffs are already low. In such contexts, tariff reductions tend to generate incremental changes rather than structural shifts.

What matters here is less the size of the effects than their composition. The comparison between IT and CT reveals adjustment patterns that are not visible in aggregate analyses of ICT trade. Technologies embedded in production systems are associated with stronger reallocation across suppliers, whereas communication-related products tend to follow existing trade structures.

This contrast underscores the importance of functional differences within ICT. Production-embedded technologies are more closely linked to changes in sourcing and input structure, while communication technologies primarily reinforce coordination within established networks.

These differences also raise a question of sequencing. While the present analysis is not designed to identify an optimal temporal ordering of reforms, it suggests that different components of digital liberalization are associated with distinct adjustment patterns and policy trade-offs.

From a policy perspective, the results indicate that digital trade liberalization should not be treated as a uniform package. Liberalization of connectivity-related goods may support network integration with limited disruption, whereas liberalization of production-embedded technologies may be more closely linked to upgrading but also involves stronger import responses and short-term fiscal effects. These outcomes are not necessarily adverse; they may reflect access to higher-quality inputs and technologies. However, they point to the need for complementary policies aimed at strengthening domestic capabilities and absorptive capacity.

The contribution of the paper lies in making these differences visible. By disaggregating ICT into functionally distinct groups, the analysis shows how aggregate approaches can mask variation in import responses, supplier reallocation, tariff revenue, and welfare effects.

The analysis is deliberately conducted within a partial equilibrium framework to allow for a detailed product-level perspective. While this approach does not capture general equilibrium interactions, it provides a clearer view of how specific components of digital liberalization affect trade structure. Future research could extend this approach using general equilibrium models, firm-level data, or dynamic panel analysis based on longitudinal customs data.

6. Conclusions

This paper examines the heterogeneous effects of digital trade liberalization by distinguishing between information and communication technologies in Vietnam. Using a SMART partial equilibrium simulation based on 2023 trade and tariff data, the study evaluates the effects of tariff elimination on imports, trade creation, trade diversion, tariff revenue, and welfare.

The results show that liberalization in production-embedded technologies generates stronger import responses and larger fiscal effects than liberalization in connectivity-oriented technologies. Although the overall magnitude of the effects remains modest, the observed asymmetry provides insight into how digital trade policies interact with production structures in a low-tariff environment.

The contribution of the paper lies less in the magnitude of the estimated effects than in the comparison itself. Treating ICT as a single category masks differences that become visible once the components are separated. The results suggest that even in a low-tariff setting, the composition of liberalization matters.

The study is subject to several limitations. The analysis is based on a partial equilibrium framework and a single benchmark year. It does not capture dynamic responses, non-tariff measures, investment effects, or firm-level heterogeneity. Future research could extend the analysis using general equilibrium models, dynamic panel approaches based on longitudinal customs data, or firm-level evidence to examine how digital liberalization affects technological upgrading and production capability over time.