## Fiscal Revenue Mobilization and Digitally Traded Products: Taxing at the Border or Behind It? (insea2023005 — excerpts)

## Source details

**Canonical URL:** [Fiscal Revenue Mobilization and Digitally Traded Products: Taxing at the Border or Behind It? (insea2023005 — excerpts)](https://www.imf.org/-/media/files/publications/imf-notes/2023/english/insea2023005.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/imf-notes/2023/english/insea2023005.pdf.md)
- [Structured JSON version](/-/media/files/publications/imf-notes/2023/english/insea2023005.pdf.json)

---

### Executive summary and context
- Digitalization shifts trade in digitizable products (for example, e-books, digital music) toward transactions via electronic transmissions (ETs) instead of physical goods crossing borders.
- WTO moratorium on customs duties on electronic transmissions:
  - Agreed in 1998; normally extended at two-year intervals.
  - In force until MC13 to be held in February 2024.
- Policy debate:
  - Some members (India, Indonesia, South Africa) have argued the moratorium should be reconsidered to increase policy space.
  - Others (including the EU and Group of Seven) support a permanent moratorium.
- Focus of the note:
  - Estimates of the portion of trade in digitizable products conducted electronically.
  - Upper bound static estimates of revenue potential from tariffs versus a value-added tax (VAT) on digitized flows.

### Key quantitative findings
- Relative revenue potential:
  - The maximal revenue potential of VAT on trade in digitized goods, based on a static analysis, is about 150 percent higher than that of tariffs at current rates.
- Absolute static estimates (upper-bound assumptions):
  - Estimated revenue potential in 2012: tariffs about $0.6 billion; VAT about $1.2 billion.
  - Estimated revenue potential in 2021: tariffs about $4 billion; VAT about $10 billion.
- Effective and standard tax rate observations:
  - Standard VAT rates in most countries are typically between 15 and 20 percent.
  - Effective VAT rates across country groups range between 3.5 percent and 7 percent on average.
- Country-group patterns (2021):
  - High-income countries: import largest share of world digitized products; tariffs (very) low; VAT rates imply significant VAT revenue potential.
  - Upper middle-income countries: revenue potential roughly the same for tariffs and VAT on average.
  - Lower middle-income countries: VAT has higher revenue potential, driven by digitized imports for final consumption (for example, audiovisual and gaming).
  - Low-income countries: import very little digital trade; VAT and tariff revenue potential are comparatively low and tend to be similar, with tariffs slightly higher due to comparatively high statutory tariff rates.
- Revenue potential as share of total tax revenue:
  - Static revenue estimates as a share of total revenue range between zero and 4.6 percent across all countries covered.
  - The largest differences in revenue potential in favor of VAT are observed among lower middle-income countries (up to 4.6 percent of total tax revenue).
  - In low-income countries the GDP-weighted average difference between VAT and tariff revenue potential is –0.1 percent of total revenue.
  - Among lower middle- and low-income countries, only five countries are estimated to have a significantly higher static revenue potential under tariffs (difference of more than 0.1 percentage points of total revenue).
- Country-specific dynamic estimates of VAT reforms on digital flows (emerging market and developing economies):
  - Short-term annual revenue gains range from about 0.03 percent of GDP (Chile, South Africa) to about 0.08 percent of GDP (Thailand).
- Other numeric findings and ranges referenced:
  - World internet users: about 400 million in 2000; 1 billion in 2005; almost 5 billion in 2020.
  - Upper bound counterfactual growth assumption for post-2010 trade in digitizable goods: constant rate of 10.8 percent.
  - VAT effective rates clustering: typically around 5 to 10 percent.
  - C-efficiency ratios indicate relatively low VAT effectiveness in low- and lower middle-income countries.
  - Estimates cited from earlier work: WTO 2016 — tariff revenue from digitizable products decreased from $1.2 billion in 2000 to $823 million in 2014 (0.26 percent of total tariff revenues in 2014).
  - Banga (2019) and Banga (2022) summarized estimates of foregone revenue for developing countries (figures reported in the literature section of the note).
  - General equilibrium modeling (Makiyama and Narayanan 2019; WTO 2020b) suggests tariffs on ETs could reduce imports and generate limited tariff revenue gains while harming investment, employment, growth and tax revenue in some cases (example: net loss of tax revenue for India estimated at about $2 billion in one study).
  - Repeated numeric references: "2.6 to 3.1 percent in developing countries and 10.4 to 11.7 percent in LDCs." (as appears in the source text).

### Methodology for estimating digitized trade and revenue potential
- Scope definition:
  - Adopted list from a recent WTO (2020b) study encompassing four categories: photographic and cinematographic film; print matter; media for sound, video, software, or other phenomena; and video games (Annex 1).
- Counterfactual construction and digitized trade measure:
  - Digitized trade defined as the difference between actual imports of digitizable products and counterfactual imports that would have grown at a constant rate (methodology based on UNCTAD 2017; Banga 2019, 2022).
  - Counterfactual post-2010 imports assumed to grow at the average annual growth rate in digitizable products over 2000–10: 10.8 percent.
  - Estimates are explicitly upper bounds: the constant growth assumption exceeds average GDP growth over this period and ignores administrative capacity and legal obligations that would reduce collectible bases.
- Revenue potential formulas (static):
  - Tariff revenue potential: τ_c,t,p_MFN × (M_c,t,p_COUNTERFACTUAL − M_c,t,p_OBSERVED)
  - VAT revenue potential: ε_c,t,i_VAT × (M_c,t,p_COUNTERFACTUAL − M_c,t,p_OBSERVED)
- Data sources and harmonization (Annex 2):
  - MFN tariffs, bound tariffs, and imports: WTO Integrated Database and Consolidated Tariff Schedules Database.
  - Multiple HS versions converted into a single HS version using United Nations conversion tables.
  - Matching to sectors: UN concordance tables link HS product codes to ISIC-based sectors.
  - World Input-Output Database (WIOD) coverage used for final consumption shares: 56 industrial sectors (ISIC revision 3), 43 countries plus a residual rest of the world category; WIOD used for years 2000 to 2014.
- Imputation and interpolation rules (exact formulas preserved):
  - Tariffs: missing tariff assumed equal to the first available prior datapoint; if none, use the next available datapoint.
  - Imports:
    - Linearly interpolate where gaps exist.
    - Fill forward missing values in subsequent years using a moving average:
      - 0.5M_{t–1}+0.3M_{t–2}+0.2M_{t–3} where available,
      - otherwise 0.6M_{t–1}+0.4M_{t–2},
      - or M_{t–1}.
  - For years 2015 to 2021 final consumption shares: assumed that the last observed share continues to apply.
- Effective VAT rate computation:
  - 휀_{c,t,i_VAT} = τ_{c,t_VAT} × σ_{c,t,i}
  - where σ_{c,t}^{i} are final consumption shares at sector level from WIOD; imputation for missing shares uses sector-year averages across country-income groups or rest-of-world values for low-income countries.

### Comparative analysis: VAT versus tariffs — efficiency and administration
- Efficiency:
  - VAT (broad-based, nondiscriminatory, with crediting of input VAT) is less distortionary than tariffs, raises revenue with minimal distortion to consumption and production decisions, and avoids cascading effects when administration functions well.
  - Tariffs distort production decisions by taxing intermediate inputs and final goods alike and can create cascading effects and efficiency costs.
- Administrative and implementation considerations:
  - Border collection becomes less practical as trade digitalizes; imposing customs duties on ETs is difficult in practice.
  - VAT administration has been adapted in many countries to capture digitalized flows; vendor collection, reverse charge, withholding regimes, and platform reporting are among deployed tools.
  - Vendor collection regimes require nonresident suppliers to collect VAT; simplified registration thresholds are commonly adopted for nonresident suppliers.
  - Reverse charge is typical for B2B supplies of imported digital products; resident businesses apply VAT as if they were suppliers and can claim input credits when the imports are intermediate inputs.
  - Withholding regimes and platform liability have been used as backstops or complements (examples: Chile, Colombia, Costa Rica, Mexico, Türkiye, India).
  - Practical implementation choices vary: some countries use lists of in-scope companies (Indonesia), others define scope broadly (South Africa changed from a prescribed list in 2014 to a broad definition in 2019).
- C-efficiency and compliance:
  - C-efficiency ratios (IMF Fiscal Affairs Department database) measure the departure from a perfectly enforced uniform-rate VAT and capture policy gaps and compliance/administration effectiveness.
  - C-efficiency is relatively low in low- and lower middle-income countries, indicating scope for revenue gains from improved policy design and administration.

### Policy conclusions and recommendations
- Core policy conclusion:
  - Broad-based nondiscriminatory VATs are preferable to tariffs for ETs on economic efficiency and revenue grounds.
- Specific policy recommendations and rationale:
  - Prioritize VAT-based approaches complemented by investment in administrative capacity to capture digitized flows and increase coverage and compliance.
  - Invest in modernizing VAT and customs infrastructure of developing countries, with priority on updating and enhancing domestic VAT infrastructure rather than implementing customs duties on ETs.
  - Use the WTO moratorium on customs duties on electronic transmissions as a commitment device to encourage countries to implement efficient VAT-based taxation of ETs.
  - Recognize that collecting customs duties at the border on digitally delivered products is impractical and more distortionary; VAT captures consumption more efficiently and builds on widespread administrative experience.
  - Global coordination and support are needed to modernize tax systems and to help developing countries implement vendor collection, reverse charge, platform reporting, withholding regimes and other practical measures tailored to country circumstances.

*Source: insea2023005 — References, Executive Summary, Box 1, Annex 2 (IMF Note 2023/005).*

### references.

### References

### Identifiers and Classification
- ISBN: 9798400253614 (paper)
- ISBN: 9798400253621 (ePub)
- ISBN: 9798400253645 (Web PDF)
- Subjects: LCSH: International Trade—Taxation. | International Trade—Effect of technological innovations on.
- Classification: LCC K4475.4.H3 2023

### Disclaimer and Purpose of the IMF Notes Series
- The IMF Notes Series aims to quickly disseminate succinct IMF analysis on critical economic issues to member countries and the broader policy community.
- The views expressed are those of the author(s) and do not necessarily represent the views of the IMF, its Executive Board, or its management.

### Recommended Citation
- Hanappi, Tibor, Adam Jakubik, and Michele Ruta. 2023. “Fiscal Revenue Mobilization and Digitally Traded Products: Taxing at the Border or Behind It?” IMF Note 2023/005, International Monetary Fund, Washington, DC.

### Publication and Ordering Information
- International Monetary Fund, Publications Services
- P.O. Box 92780, Washington, DC 20090, USA
- Tel.: (202) 623-7430 Fax: (202) 623-7201
- E-mail: publications@imf.org
- bookstore.IMF.org
- elibrary.IMF.org

### Acknowledgements
- Helpful comments from several IMF departments are gratefully acknowledged.
- The authors thank Hanif Alam, Enrique Aldaz-Carroll, Andrea Andrenelli, Antonia Carzaniga, Thomas Ecker, Simon Evenett, Anabel Gonzalez, Shafik Hebous, Kenneth Kang, Alexander Klemm, Julieta Ladronis, Jan Loeprick, Javier López González, Mario Mansour, Martin Molinuevo, Roberta Piermartini, Martin Sommer, Charles Vellutini for feedback and comments.
- Special thanks to Anna Konopatskaya and Cheryl Toksoz for outstanding production assistance.

### Executive Summary of the Note (Key Findings and Policy Conclusions)
- Digitalization transforms trade in digitizable products (for example, e-books, digital music) by increasing transactions via electronic transmissions (ETs) instead of physical goods crossing borders.
- WTO members agreed in 1998 to a moratorium on customs duties on electronic transmissions; the moratorium is normally extended at two-year intervals and is in force until MC13 to be held in February 2024.
- Some members (India, Indonesia, South Africa) have argued the moratorium should be reconsidered to increase policy space; others (including the EU and Group of Seven) support a permanent moratorium.
- The note focuses on fiscal implications of domestic and border tax instruments on ETs and provides:
  - Estimates of the portion of trade in digitizable products conducted electronically.
  - Upper bound static estimates of revenue potential from tariffs versus a value-added tax (VAT) on digitized flows.

Key quantitative and comparative findings:
- The maximal revenue potential of VAT on trade in digitized goods, based on a static analysis, is about 150 percent higher than that of tariffs at current rates.
- The difference in revenue potential between VAT and tariffs is largest for high-income countries due to:
  - A large share of global imports; and
  - Low tariffs relative to VAT rates.
- Upper and lower middle-income countries are heterogeneous:
  - Revenue potential is roughly the same for upper middle-income countries.
  - VAT has higher revenue potential in lower middle-income countries.
- Low-income countries have low revenue potential for both instruments due to low imports in digitized products.

Policy recommendations and rationale:
- Broad-based nondiscriminatory value-added taxes (VATs) are preferable to tariffs for ETs on economic efficiency and revenue grounds.
  - VATs: broad-based, exclude intermediate inputs, raise revenue with minimal distortion to consumption and production decisions, thus minimal impact on output and welfare.
  - Tariffs: more distortionary and less effective as a revenue instrument in this context.
- VATs are easier to implement and administer given extensive accumulated experience across countries of all income groups; little is known about implementing ET tariffs.
- The WTO moratorium on customs duties on electronic transmissions can serve as a commitment device to promote more efficient taxation via VAT.
- Transition to VAT-centered taxation of ETs requires investment by the global community in modernizing tax and customs infrastructure of developing countries.
  - Investment priority: update and enhance domestic VAT infrastructure rather than implementing customs duties on ETs.
  - Reasons: (1) greater economic efficiency (VAT creates less distortion per dollar raised), (2) easier administration (builds on existing tax infrastructure), (3) easier implementation (capturing all ETs “at the border” is impractical).

### Literature and Empirical Context (Selected Findings Summarized)
- Early WTO assessment (WTO 2016): tariff revenue collected from digitizable products decreased from $1.2 billion in 2000 to $823 million in 2014; this amounted to 0.26 percent of total tariff revenues in 2014.
- UNCTAD and Banga findings:
  - Banga (2019) estimated more than $10 billion per annum of foregone revenue for developing countries excluding LDCs, $1.5 billion for LDCs, and an order of magnitude less for high-income members (assuming tariffs on ETs could be collected at bound rates).
  - Banga (2022) updates: over 2017–20 developing countries excluding LDCs had foregone $12 billion per annum and LDCs $2 billion per annum (based on a scope of 49 HS 6-digit subheadings and counterfactual growth assumptions).
- Critiques and counterpoints:
  - Evenett and Fritz (2022) note only a handful of developing economies finance more than 20 percent of budgets through tariff revenue; half of the top 10 tariff-dependent developing economies are not WTO members and do not impose customs duties on ETs.
  - Many WTO members set tariffs below their maximum bindings; small tariff increases could offset estimated revenue loss in many cases.
- General equilibrium modeling results:
  - Makiyama and Narayanan (2019): imposing customs duties on ETs and reciprocal tariff increases by trading partners result in negative impacts on investment, employment, growth and tax revenue; net loss of tax revenue in the case of India estimated at about $2 billion.
  - WTO (2020b): imposing customs duties on ETs raises total tariff revenue by only 0.15 percent in developing countries and 0.25 percent in LDCs, and reduces imports.

### Contents Overview (from the Note)
- Introduction
- Literature Review
- Digitalization and Trade
- Tariffs and VAT on Digitized Products
- Revenue Effects of Tariffs versus VAT on Electronic Transmissions
- Policy Discussion and Conclusion
- Boxes:
  - 1. VAT on Digital Trade: Experiences in Emerging Market and Developing Economies
- Figures referenced in the Note:
  - 1. World Imports of Digitizable Goods
  - 2. Tax Mix by Income Group in 2021
  - 3. Standard VAT Rates and Tariff Rates on Digitizable Products
  - 4. Effective VAT Rates and Tariff Rates on Digitizable Products
  - 5. Revenue Potential of Taxing Electronic Transmissions
  - 6. Static Revenue Potential from Taxing Electronic Transmissions by Country Group
  - 7. C-Efficiency Ratios by Country Income
- Annexes referenced:
  - 1. List of HS 6-Digit Subheadings of Digitizable Products
  - 2. Data and Trade Estimates
  - 3. VAT on Digital Trade: Selected Design Features

*Source: insea2023005 - references (IMF Note 2023/005).*

### 2.6 to 3.1 percent in developing countries and 10.4 to 11.7 percent in LDCs.

### 2.6 to 3.1 percent in developing countries and 10.4 to 11.7 percent in LDCs.

### Digitalization and Trade
- Digital trade contributes to welfare gains by:
  - Enabling more efficient production through specialization according to comparative advantage.
  - Lowering the cost of production through economies of scale when serving global markets.
  - Increasing the variety of goods and services available to consumers.
- Digital inputs are special inputs that:
  - Contribute to the digitalization of services provision, manufacturing production, and agriculture.
  - Enable more efficient processes that can boost firms’ productivity and benefit consumers through lower prices.
  - Promote interconnectivity and transmission of existing knowledge and technology, contributing to productivity growth and richer consumption variety.
  - Foster inclusion by reducing trade barriers for small firms and women-led businesses (WTO 2018; WTO–WBG 2020).
- Trends in digitization of consumption:
  - Number of people using the internet: about 400 million in 2000; 1 billion in 2005; almost 5 billion in 2020 (WBG 2022).
  - Consumption shifting from physical digitizable goods (e.g., books, physical recordings, games) to digital products (e.g., e-books, digital music/videos, downloadable/online games) delivered via electronic transmissions.
  - Underpinned by proliferation of internet-connected devices, increasing devices per household, growing hours spent with mobile devices, and improving download speeds.

### Quantifying potential tax revenue implications of the WTO moratorium
- Two key elements required:
  - A working definition of the scope of products affected by the moratorium:
    - Adopted list from a recent WTO (2020b) study encompassing four categories: photographic and cinematographic film; print matter; media for sound, video, software, or other phenomena; and video games (Annex 1).
    - Goal: provide an approximate measure of revenue implications; policy recommendations also depend on economic theory of efficient taxation and practical considerations such as administration and implementation.
  - An estimate of trade in digitized products covered by the moratorium:
    - Trade in physical products with digitized equivalents is recorded in customs statistics; trade in digitized products is not recorded at the desired level and must be estimated.
    - Methodology: adopt UNCTAD (2017) and Banga (2019, 2022) approach—assume a constant growth rate of trade in these products to construct a counterfactual. Digitized trade defined as the difference between actual imports of digitizable products and the constructed counterfactual imports (Figure 1; see Annex 2).
    - Upper bound nature: estimates should be seen as an upper bound because the constant growth rate assumption exceeds average GDP growth over this period and ignores administrative capacity and legal obligations that reduce the portion of trade on which customs duties could potentially be collected.
- Additional considerations that may reduce magnitudes:
  - A portion of consumption may not be dutiable (e.g., treatment of streaming and subscription-based services where ownership does not change hands is ambiguous).
  - Existing rules in services trade (NT and MFN clauses in GATS and other agreements, bilateral tax treaties covering intellectual property fees) may limit the revenue potential of customs duties on electronic transmissions.
  - Expanding commitments on digital services in regional trade agreements have a similar limiting effect.
  - Considering these factors would strengthen the authors’ policy conclusions; the subsequent focus is on the theoretical maximum revenue potential assuming these factors do not come into play.
- Measurement note:
  - Digitally delivered trade should be recorded as part of services trade statistics per the WTO–OECD–IMF–UNCTAD Handbook on Measuring Digital Trade; measurement framework aligns with BPM6 and IMTS 2010 (WTO–OECD–IMF–UNCTAD 2023).

### Figure 1 and counterfactual construction
- Upper bound counterfactual estimates assume trade in digitizable goods would have grown at a constant rate of 10.8 percent after 2010 (methodology based on UNCTAD 2017; Banga 2019, 2022).
- Digitized trade measured as the difference between actual imports and the counterfactual imports of digitizable goods (see Annex 1 for the list of digitizable products).

### Tariffs and VAT on Digitized Products
- Tariffs:
  - Contribute substantially to tax revenues in some contexts but should not be viewed in isolation.
  - Historically convenient because levied on specific physical goods at the border with relatively low administrative and compliance costs.
  - Share of trade taxes in overall tax revenue is inversely related to development level (Figure 2).
  - Tariff rates vary by product and country and include intermediate inputs, implying significant efficiency costs.
  - Tariffs do not discriminate between intermediate inputs and final consumption, directly increasing costs of imported inputs and creating cascading effects that distort production decisions.
  - Tariff incidence is ultimately empirical; even large economies can experience almost complete pass-through of tariffs so domestic firms and consumers may bear the entirety of the cost (Amiti, Redding, and Weinstein 2020; Fajgelbaum and others 2020; Cavallo and others 2021).
- VAT:
  - Focus of the following analysis as the closest potential substitute for tariffs when assessing consumption taxes.
  - VAT is widely seen as efficient, especially if broad-based and single standard rate (De Mooij and Swistak, 2022).
  - Standard VAT rates in most countries are typically between 15 and 20 percent.
  - Two key principles:
    - Destination principle: imports taxed at the border and exports zero-rated; avoids distortions and aligns with WTO rules on National Treatment and subsidies.
    - Crediting of input VAT: ensures tax is levied only on domestic final consumption, avoiding cascading effects except where VAT administration is not fully operational (e.g., time lags in refunds).
  - Tariffs and VAT can be combined on the same cross-border transaction, but duplication reduces efficiency for no fiscal purpose. The VAT base should in principle include tariffs and excise duties to reflect final consumption accurately.

### General efficiency and administrative issues
- Digitization of trade reduces practicality of border collection, affecting government revenues and economic efficiency.
  - Increasing supply of intangibles and digitally traded products makes border tax collection difficult.
  - Digitization leads to foregone tariff and tax revenue for taxes applied to cross-border transactions (tariffs, excise duties, VAT).
  - Tax systems become less neutral if certain cross-border trade is not captured (e.g., domestic advertisers subject to VAT while foreign online platforms escape taxation), potentially harming domestic businesses’ competitiveness.
- VAT administration has been adapted successfully to digitalization challenges:
  - Authorities in advanced economies adapted to secure domestic tax bases (see Box 1 and Annex 3 for selected country approaches).

### Box 1. VAT on Digital Trade: Experiences in Emerging Market and Developing Economies
- Administrative approaches to ensure VAT taxation of digitally traded products differ across countries:
  - B2B supplies of imported digital products:
    - Typical approach: require resident businesses to apply the reverse charge method—resident businesses charge VAT on digital imports as if they were the supplier.
    - If imported digital products are intermediate inputs, resident businesses receive input VAT credits; avoidance risks are relatively low because reverse charge incentivizes correct identification and claiming of input VAT.
  - B2C supplies of imported digital products:
    - Vendor collection is the most common approach—non-resident suppliers liable for VAT charge and required to register if local sales exceed a certain threshold.
    - Simplified registration processes are often developed to increase voluntary compliance because final consumers cannot claim VAT credits.
  - Scope definition:
    - To implement vendor collection on remote supplies, the scope should be clearly defined in VAT legislation.
    - OECD VAT guidelines propose a broad definition of in-scope products, essentially including all intangible goods and services (OECD 2017a, 2017b).
    - Country examples:
      - South Africa (2014 legislation) originally limited scope to electronic services defined on a prescribed list; changed in 2019 to include all supplies that meet a certain definition.
      - Indonesia has opted for a list of in-scope companies.

*IMF Note: insea2023005 (excerpts provided).*

### Box 1. VAT on Digital Trade: Experiences in Emerging Market and Developing Economies

### Box 1. VAT on Digital Trade: Experiences in Emerging Market and Developing Economies

### Approaches to collecting VAT on digital trade
- Vendor collection regimes require nonresident suppliers to collect VAT; maintaining lists of compliant companies is challenging and can create distortions across firms.
- Withholding regimes place responsibility on financial intermediaries to withhold VAT on payments for digital imports; used as a backstop in Chile, Colombia, and Costa Rica when nonresident suppliers do not comply with vendor collection regimes.
- Digital platforms are leveraged for VAT collection:
  - Some countries require platforms to share tax information with authorities (example: Chile).
  - Full liability regimes make platforms liable to collect VAT on inbound supplies made through them (examples: Mexico and Türkiye).
  - India requires platforms to collect tax at source at 1 percent of the value of taxable supplies made through them by other suppliers.

### Rate structure and effective VAT rates
- Standard VAT rates are higher than tariffs across all country groups based on unweighted averages, with the gap largest in high-income countries and lowest in low-income countries.
- The analysis assumes the standard VAT rate is applicable to all products for current estimates.
- Effective VAT rates account for final consumption shares at the sector level; this yields “effective” VAT rates substantially lower than standard rates.
- Effective VAT rates, excluding high-income countries, are lower than tariff rates.
- Effective VAT rates across country groups range between 3.5 percent and 7 percent on average.

### Methodology and caveats for estimating revenue potential
- The methodology compares static revenue potential of tariffs versus VAT, holding current rates and demand constant.
- Caveats:
  - Two policy approaches are presented as mutually exclusive for illustrative comparison.
  - Actual imports of digitized products are not observed; estimates use an upper bound counterfactual growth projection.
  - Country-specific features of VAT legislation are not modeled in detail; realized VAT revenue may differ by country.
  - The exercise estimates maximum revenue potentials and does not measure incremental VAT gains under current tax regimes.

### Revenue potential: global and country-group patterns
- Definition of estimated electronic transmissions (ETs): difference between observed imports of digitizable products and the upper bound of counterfactual imports.
- Revenue potential formulas used:
  - Tariff revenue potential: τ_c,t,p_MFN × (M_c,t,p_COUNTERFACTUAL − M_c,t,p_OBSERVED)
  - VAT revenue potential: ε_c,t,i_VAT × (M_c,t,p_COUNTERFACTUAL − M_c,t,p_OBSERVED)
- Global results (static, upper-bound assumptions):
  - VAT on ETs would yield about 150 percent more potential revenue than tariffs.
  - Estimated revenue potential in 2012: tariffs about $0.6 billion; VAT about $1.2 billion.
  - Estimated revenue potential in 2021: tariffs about $4 billion; VAT about $10 billion.
  - The widening gap is driven by an increasing share of digitized products subject to relatively low tariffs while VAT rates remained broadly constant.
- Country-group patterns (2021):
  - High-income countries:
    - Import the largest share of world digitized products and tend to have (very) low tariffs.
    - VAT rates typically about 15 to 20 percent, implying significant VAT revenue potential.
  - Middle-income countries:
    - Import smaller shares across various sectors; tariffs vary and can be relatively high on some products.
    - On average, revenue potential from tariffs and VAT is roughly the same for upper middle-income countries.
    - VAT revenue potential dominates for lower middle-income countries, driven by digitized imports for final consumption (for example, audiovisual and gaming).
  - Low-income countries:
    - Import very little digital trade; revenue potential is comparatively low.
    - VAT and tariff revenue potential tend to be similar, with tariffs slightly higher due to comparatively high statutory tariff rates.

### Revenue potential as a share of total tax revenue
- Static revenue estimates as a share of total revenue range between zero and 4.6 percent across all countries covered.
- Revenue potential as a share of total revenue tends to be higher in lower middle- and low-income countries.
- The largest differences in revenue potential in favor of VAT are observed among lower middle-income countries (up to 4.6 percent of total tax revenue).
- In low-income countries the GDP-weighted average difference between VAT and tariff revenue potential is –0.1 percent of total revenue.
- Among lower middle- and low-income countries, only five countries are estimated to have a significantly higher static revenue potential under tariffs (difference of more than 0.1 percentage points of total revenue).

### Country-specific and dynamic considerations
- Country-specific estimates of VAT reforms targeted at digital flows suggest short-term annual revenue gains in emerging market and developing economies range from about 0.03 percent of GDP (Chile, South Africa) to about 0.08 percent (Thailand).
- These initial gains are expected to increase over time via indirect effects on compliance as reporting obligations of digital platforms expand.

### VAT administration, compliance, and C-efficiency
- Improving VAT administration and compliance is crucial to increasing revenues from domestic consumption in a digitalized economy.
- C-efficiency ratios (based on the IMF Fiscal Affairs Department database) measure departure from a perfectly enforced uniform-rate VAT and capture both policy gaps (preferential/reduced rates and incomplete coverage) and compliance/administration effectiveness.
- C-efficiency is relatively low in low- and lower middle-income countries, indicating potential for significant revenue gains from improved VAT policy design and investments in administration and compliance.

### Policy discussion and conclusions
- Comparison of two policy options to mobilize tax revenue from trade in digitized products:
  - Discriminatory measures such as tariffs.
  - Broad-based nondiscriminatory measures such as the VAT.
- Key conclusions:
  - VAT is less distortionary and more efficient than tariffs.
  - Static revenue estimates show that, globally, VAT revenue potential from digitized trade at current rates is 150 percent higher than potential tariff revenue.
  - VAT revenue potential is about the same magnitude or higher than tariffs across country groups at all income levels.
- Policy implications:
  - Prioritize VAT-based approaches complemented by investment in administrative capacity to capture digitized flows and increase coverage and compliance.
  - Investment in a well-functioning VAT system has long-term benefits for raising tax revenue efficiently.
  - The WTO moratorium on customs duties on ETs can serve as a commitment device to help countries focus reform efforts on implementing first-best policies.
  - Global policy coordination and support for developing countries are needed to modernize tax systems.

*Source: Box 1. VAT on Digital Trade: Experiences in Emerging Market and Developing Economies (insea2023005).*

### Annex 2. Data and Trade Estimates

### Annex 2. Data and Trade Estimates

### Data sources and harmonization
- MFN tariffs, bound tariffs, and imports: sourced from WTO Integrated Database and Consolidated Tariff Schedules Database and include all WTO members (Figure 9).
- Multiple HS versions in source data converted into a single Harmonized Commodity Description and Coding System (HS) version using United Nations conversion tables.
- Matching of digitizable products to end-use sectors uses UN concordance tables to link HS product codes to ISIC-based sectors.

### Imputation and interpolation rules for missing country-HS dyads
- Tariffs: missing tariff is assumed equal to the first available prior datapoint; if none, use the next available datapoint.
- Imports:
  - Linearly interpolate where gaps exist.
  - Fill forward missing values in subsequent years using a moving average:
    - 0.5M_{t–1}+0.3M_{t–2}+0.2M_{t–3} where available,
    - otherwise 0.6M_{t–1}+0.4M_{t–2},
    - or M_{t–1}.
- Counterfactual post-2010 imports: assumed to grow at the average annual growth rate in digitizable products over 2000–10: 10.8 percent.

### VAT data and estimation of revenue potential
- Standard VAT rates: sourced from the IMF Fiscal Affairs Department tax database (Figure 10).
- Revenue potential from VAT differs conceptually from tariffs due to VAT’s crediting mechanism under the destination principle (VAT levied on final domestic consumption; imports taxed at the border; exports zero-rated).
- Adjustment needed: estimated VAT revenue potential must be scaled by the share of digital trade imports that go into final domestic consumption (excluding intermediate goods and services that are re-exported or used in zero-rated production).

### Computation of final consumption shares and effective VAT rates
- Final consumption shares (σ_{c,t}^{i}) computed at the sector level using global inter-country input-output tables (World Input-Output Database) for years 2000 to 2014.
- WIOD coverage and structure:
  - 56 industrial sectors (ISIC revision 3 classification),
  - 43 countries plus a residual rest of the world category,
  - Constructed from national supply and use tables with explicit origin and destination of imports and exports.
- Effective VAT rates (휀휀_{푐푐,푡푡,푖푖 푉푉푉푉푉푉}) computed as:
  - 휀휀_{푐푐,푡푡,푖푖 푉푉푉푉푉푉} = 휏휏_{푐푐,푡푡 푉푉푉푉푉푉} × 휎휎_{푐푐,푡푡 푖푖}
  - allowing for variation across countries (c), years (t) and sectors (i).
- Imputation for missing final consumption shares:
  - Sector-year specific averages across country-income groups used where WIOD country coverage is missing.
  - For low-income countries, rest-of-the-world sector-year values used to impute missing points.
  - For years 2015 to 2021 it is assumed that the last observed share continues to apply.

### Comparison of effective VAT rates and tariffs on digitizable products
- Matching products to industrial sectors enables direct comparison of effective VAT rates with tariffs at the product level (Figure 11).
- Empirical pattern reported:
  - Tariff rates are much more dispersed than effective VAT rates and are often considerably higher.
  - Effective VAT rates tend to be more condensed and typically clustered around 5 to 10 percent, with only very few observations nearing zero.
- Note: effective VAT rates in the comparison are independent of the magnitude of estimated digitized trade flows.

### Selected practical implications from the methodology
- Tariff revenue estimation is straightforward: product-specific tariff rates applied to inbound transactions.
- VAT revenue estimation requires sector-level final consumption adjustments to avoid overstating revenue from imports that are used as intermediates or re-exported.
- Measurement would benefit from cross-classification tables between traded products and industries of production with up-to-date definitions and granularity.

*Source: Annex 2. Data and Trade Estimates, insea2023005 - Annex 2. Data and Trade Estimates.*

---


_Source: https://www.imf.org/-/media/files/publications/imf-notes/2023/english/insea2023005.pdf_