## wpiea2020245-print-pdf

## Source details

**Canonical URL:** [wpiea2020245-print-pdf](https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020245-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2020/english/wpiea2020245-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2020/english/wpiea2020245-print-pdf.pdf.json)

---

### I. Introduction — scope and motivation
- Examples of cross-border tax evasion cited:
  - An individual evaded Chinese tariff payments of $54 million on 1.3 million tons of oil products imported to China (late 2018).
  - A World Bank estimate of $1.2 billion in tariffs evaded by well-connected firms in Tunisia between 2002 and 2009.
  - MTIC (Missing Trader Intra Community) fraud in the EU estimated to incur between EUR 45 billion to 60 billion in annual tax losses.
- Key claims:
  - Cross-border trade fraud is non-trivial; exporters and importers consistently report different values for goods traded, and a crude estimate suggests it could represent up to 6.6 percent of GDP in low-income countries.
  - Digitalization can reduce trade fraud by improving availability, authenticity, accuracy, and timeliness of trade information at the border.
  - Potential revenue gains from digitalization could be substantial.
- Data coverage referenced: bilateral trade transactions using data on 28 intra-EU and 85 cross-country trade transactions over the period 2003–16.

### II. How digitalization affects tax compliance (summary of mechanisms)
- Digitalization benefits for tax and customs administration:
  - Electronic tax filing, pre-populated returns, verification of customs and business activity.
  - Reconciliation of payment differences, real-time revenue monitoring, audit facilitation, big-data taxpayer risk assessment.
  - World Bank (2016) estimates electronic filing and payments reduced time by 25 percent on average in the five years after digital system introduction.
- Country examples:
  - South Africa: electronic tax submissions, customs declarations, and payments rose from below 20 percent to close to 100 percent over the past decade.
  - Estonia: use of big data to identify high-risk and anomalous taxpayer behavior.
- Digitalization-specific advantages in trade:
  - Third-party digital records improve verification and authenticity of submissions; blockchain noted as a technology that can secure initial submissions and subsequent modifications (use remains limited).
  - Digital customs transaction data enable risk-based targeting of declarations, improving enforcement efficiency.
- Limitations:
  - Digitalization reduces authenticity and accuracy problems but obstacles remain for completeness of information, especially when payments (e.g., credit terms) do not match declared goods’ values.
- Fiscal importance:
  - Trade taxes represent close to 10 percent of total revenues on average in developing economies (Figure 3, Taxes on International Trade, 2015).

### III. Methodology and data (core empirical approach)
- Empirical strategy:
  - Estimate a trade gravity model where the dependent variable is the difference between importer-reported trade value (CIF) and exporter-reported trade value (FOB), normalized by the average reported trade flow (the “trade gap”).
  - Controls: bilateral CIF cost proxies (distance, common borders, languages), exporter and importer country characteristics (VAT rates, weighted average tariff rates), GDP and GDP per capita, inflation, exchange rates, regional trade agreement participation, GATT/WTO membership, and country-pair and year fixed effects.
- Main regressor:
  - Digitalization proxied by the UN’s Online Service Index (normalized 0 to 1), available since 2003.
- Instrumental variable strategy:
  - Instrument digitalization with R&D efficiency (ratio of patents to R&D intensity, where R&D intensity = R&D expenditure in percent of GDP).
  - Reported first-stage Kleibergen-Paap F-statistics exceed Stock and Yogo (2005) critical values.
- Data sources:
  - Bilateral trade: IMF’s Direction of Trade Statistics (DOTS).
  - Macroeconomic variables: IMF World Economic Outlook, World Development Indicators, IMF Tax Database.
  - Trade agreement and distance: CEPII Gravity Dataset.
  - Governance indices: World Governance Indicators (WGI).
  - Digitalization indices: UN Online Service Index, World Bank Digital Adoption Index, WEF Government Success in ICT Promotion.

### IV. Empirical results — key findings and magnitudes
- Trade misreporting magnitudes (median trade gap ratios, 2016):
  - Advanced Economies (AEs): -2.4 percent of GDP.
  - Low-Income Developing Countries (LIDCs): -6.6 percent of GDP.
- Intra-EU (28 countries, 2003–16) (selected estimates):
  - OLS (column 1): Im.Digitalization index coefficient = 0.041 (standard error 0.041) — positive but statistically insignificant.
  - TSLS second-stage (TSLS-2): importer digitalization coefficient estimates = 0.893** (standard error 0.350) and 0.936*** (standard error 0.341) in reported specifications.
  - Im.VAT rate coefficient (OLS) = -0.015***.
  - First-stage F-statistics reported = 68.60.
- All partners sample (selected estimates):
  - OLS (column 1): Im.Digitalization index = -0.051 (0.036); Ex.Digitalization index = 0.111*** (0.034).
  - TSLS second-stage (TSLS-2): Im.Digitalization index coefficient = 1.158* (standard error 0.670) and 1.163* (standard error 0.666) in columns 4 and 5; Ex.Digitalization index in second-stage = -1.463** and -1.442**.
  - First-stage F-statistics reported = 33.52.
  - Robustness: censoring dependent variable at 1st and 99th percentiles; controls include country and year fixed effects and governance controls.
- Interpretation:
  - Destination/importer digitalization is positively associated with higher reported import values relative to exporter-reported exports, implying improved digitalization in the importing country increases recorded imports and thus taxable base for VAT and tariffs.
  - Effects persist after controlling for tariffs, tax rates, development, and governance.

### V. Revenue gain calculation — assumptions and estimates
- Revenue Gain formula used:
  - Revenue Gain = τ_tax ⋅ Δ(V_imports - V_exports), rearranged and computed conservatively as Revenue Gain = τ_tax ⋅ 1/2 ⋅ (V_imports + V_exports) ⋅ β_imp ⋅ Δz_imp.
- Core assumptions in back-of-the-envelope exercise:
  - Halve the distance to the digitalization frontier for each importer: Δz_imp = 0.5 ⋅ (1 - z_imp).
  - Use country-specific VAT and weighted tariff rates and average 2016 trade flows.
  - Use β_imp = 1.158 (estimate from Table 3).
  - Conservative bias: assume importer digitalization increases reported imports without affecting exporter reports; denominator imports held constant.
- Median revenue gains per country group from closing half the distance to the digitalization frontier, 2016 (percent of GDP):
  - Advanced Economies: VAT Revenue Gains 0.7; Tariff Revenue Gains 0.04.
  - Emerging Market Economies: VAT Revenue Gains 0.7; Tariff Revenue Gains 0.2.
  - Low-Income Developing Countries: VAT Revenue Gains 1.1; Tariff Revenue Gains 0.4.
  - EU-28: VAT Revenue Gains 0.4.
- Additional summary points:
  - The paper reports "1.1 percent of GDP for low-income developing countries, 0.7 percent of GDP for emerging" as indicative potential VAT gains from halving the distance to the digitalization frontier.

### VI. Mechanisms, limitations, and risks
- Transmission mechanism:
  - How improvements in digitalization translate into better enforcement at the border requires further analysis.
- New risks:
  - Digitalization can introduce new fraud opportunities; example: digitalization of Estonia’s tax administration led fraudsters to create large numbers of ghost entities to generate multiple small credit claims that fell below the threshold for audit (IMF, 2018).
- Implementation constraints:
  - Realizing digital dividends requires adequate administrative and institutional capacity, significant fiscal resources, and political resolve.
  - The estimates are indicative and contingent on institutional reforms and capacity building to prevent new forms of fraud enabled by digital technologies.

### VII. Policy implications and research priorities
- Priorities:
  - Further work to analyze the transmission mechanism linking digitalization to enforcement effectiveness at borders.
  - Granular analysis to determine which specific digital tools (e.g., data matching, access to third-party information) are most useful for reducing evasion.
- Practical considerations:
  - Plan for significant fiscal resources required to close the distance to the digitalization frontier.
  - Implement institutional reforms and capacity building to mitigate fraud opportunities arising from digitalization.
  - Recognize the need for political commitment to implement and sustain digital reforms.

### Appendix: Data sources used in the analysis
- Bilateral exports — IMF: Direction of Trade Statistics
- Bilateral imports — IMF: Direction of Trade Statistics
- Common currency — CEPII: Gravity Dataset
- Common official/primary language — CEPII: Gravity Dataset
- Common religion — CEPII: Gravity Dataset
- Contiguity — CEPII: Gravity Dataset
- Control of corruption — WB: World Governance Indicators
- Digital adoption index — WB: World Development Report 2016
- E-Government index — UN: E-Government Survey 2016
- Exchange rate — IMF: World Economic Outlook
- GDP — IMF: World Economic Outlook
- GDP per capita — IMF: World Economic Outlook
- Government effectiveness — WB: World Governance Indicators
- Government success in ICT promotion — WEF: The Global Information Technology Report 2016
- Inflation rate — IMF: World Economic Outlook
- Online service index — UN: E-Government Survey 2016
- Origin is GATT/WTO member — CEPII: Gravity Dataset
- Patents filed by residents — WB: World Development Indicators
- Population weighted distance — CEPII: Gravity Dataset
- R&D expenditure (percent of GDP) — WB: World Development Indicators
- Regional trade agreement — CEPII: Gravity Dataset
- Rule of law — WB: World Governance Indicators
- Tariff rate (weighted mean) — WB: World Development Indicators
- VAT rate — IMF: Tax Rate Database

*Source: wpiea2020245-print-pdf (sections excerpted).*

### References .............................................................................................................

### References

### Tables
- 1. Pairwise Correlations of Digitalization Indices .................................................................................................. 9
- 2. Trade Gap Regressions Using Intra-EU Partners Trade Data .................................................................... 11
- 3. Trade Gap Regressions Using All Partners Trade Data ............................................................................... 12
- 4. Median Revenue Gains per Country Group from Closing Half the Distance to the  
    Digitalization Frontier, 2016 .................................................................................................................................. 14

### Figures
- 1. Number of Countries with Selected Digital Services ...................................................................................... 6
- 2. Digital Adoption Index for Governments Across Regions ............................................................................ 6
- 3. Taxes on International Trade, 2015 ........................................................................................................................ 7
- 4. Trade Gap Ratios, 2016 ........................................................................................................................................... 10

### Appendix
- Appendix

*wpiea2020245-print-pdf - References .............................................................................................................*

### 1. Data Sources ........................................................................................................

### 1. Data Sources

### I. Introduction — scope and motivation
- Examples of cross-border tax evasion cited:
  - An individual evaded Chinese tariff payments of $54 million on 1.3 million tons of oil products imported to China (late 2018).
  - A World Bank estimate of $1.2 billion in tariffs evaded by well-connected firms in Tunisia between 2002 and 2009.
  - MTIC (Missing Trader Intra Community) fraud in the EU estimated to incur between EUR 45 billion to 60 billion in annual tax losses.
- Key claims:
  - Cross-border trade fraud is non-trivial; exporters and importers consistently report different values for goods traded, and a crude estimate suggests it could represent up to 6.6 percent of GDP in low-income countries.
  - Digitalization can reduce trade fraud by improving availability, authenticity, accuracy, and timeliness of trade information at the border.
  - Potential revenue gains from digitalization could be substantial.
- Data coverage referenced: bilateral trade transactions using data on 28 intra-EU and 85 cross-country trade transactions over the period 2003–16.

### II. How digitalization affects tax compliance (summary of mechanisms)
- Digitalization benefits for tax and customs administration:
  - Electronic tax filing, pre-populated returns, verification of customs and business activity.
  - Reconciliation of payment differences, real-time revenue monitoring, audit facilitation, big-data taxpayer risk assessment.
  - Reduced administrative time burden (World Bank (2016) estimates electronic filing and payments reduced time by 25 percent on average in the five years after digital system introduction).
- Country examples of digital adoption in government:
  - South Africa: electronic tax submissions, customs declarations, and payments rose from below 20 percent to close to 100 percent over the past decade.
  - Estonia: use of big data to identify high-risk and anomalous taxpayer behavior.
- Digitalization-specific advantages in trade:
  - Third-party digital records improve verification and authenticity of submissions; blockchain noted as a technology that can secure initial submissions and subsequent modifications (use remains limited).
  - Digital customs transaction data enable risk-based targeting of declarations, improving enforcement efficiency.
- Limitations noted:
  - Digitalization reduces authenticity and accuracy problems but obstacles remain for completeness of information, especially when payments (e.g., credit terms) do not match declared goods’ values.
- Fiscal importance:
  - Trade taxes represent close to 10 percent of total revenues on average in developing economies (Figure 3, Taxes on International Trade, 2015).

### III. Methodology and data (core empirical approach)
- Empirical strategy:
  - Estimate a trade gravity model building on Kellenberg and Levinson (2019) where the dependent variable is the difference between importer-reported trade value (CIF) and exporter-reported trade value (FOB), normalized by the average reported trade flow (the “trade gap”).
  - Independent variables include bilateral CIF cost proxies (distance, common borders, languages), exporter and importer country characteristics (VAT rates, weighted average tariff rates), GDP and GDP per capita, inflation, exchange rates, regional trade agreement participation, GATT/WTO membership, and country-pair and year fixed effects.
- Main regressor of interest:
  - Digitalization proxied by the UN’s Online Service Index (normalized 0 to 1), available since 2003; correlated with other digital indices and broader sample coverage.
- Instrumental variable strategy:
  - Address endogeneity of digitalization using R&D efficiency (the ratio of patents to R&D intensity, where R&D intensity = R&D expenditure in percent of GDP) as an instrument for the digitalization index.
  - First-stage Kleibergen-Paap F-statistics exceed Stock and Yogo (2005) critical values in reported specifications.
- Data sources:
  - Bilateral trade: IMF’s Direction of Trade Statistics (DOTS).
  - Macroeconomic variables: IMF World Economic Outlook, World Development Indicators, IMF Tax Database.
  - Trade agreement and distance: CEPII Gravity Dataset.
  - Governance indices: World Governance Indicators (WGI).
  - Digitalization indices: UN Online Service Index, World Bank Digital Adoption Index, WEF Government Success in ICT Promotion (pairwise correlations reported in Table 1).

### IV. Empirical results — key findings and magnitudes
- Trade misreporting magnitudes (median trade gap ratios, 2016):
  - Advanced Economies (AEs): -2.4 percent of GDP.
  - Low-Income Developing Countries (LIDCs): -6.6 percent of GDP.
- Intra-EU (28 countries, 2003–16) results (Table 2, summary):
  - OLS (column 1): Im.Digitalization index coefficient = 0.041 (standard error 0.041) — positive but statistically insignificant.
  - TSLS second-stage (TSLS-2, censored and uncensored, used to address endogeneity): importer digitalization coefficient estimates reported at 0.893** and 0.936*** in specifications (standard errors 0.350 and 0.341 respectively) in columns 4 and 5.
  - Negative coefficient on importer VAT rate in OLS is consistent with higher incentive to underreport imports as VAT rises (Im.VAT rate coefficient -0.015*** in OLS).
  - First-stage F-statistics reported = 68.60.
- All partners sample (Table 3, summary):
  - OLS (column 1): Im.Digitalization index = -0.051 (0.036); Ex.Digitalization index = 0.111*** (0.034).
  - TSLS second-stage (TSLS-2, censored and uncensored): Im.Digitalization index coefficient = 1.158* and 1.163* in columns 4 and 5 (standard errors 0.670 and 0.666 respectively); Ex.Digitalization index in second-stage reported as -1.463** and -1.442**.
  - First-stage F-statistics reported = 33.52.
  - Robustness: censoring dependent variable at 1st and 99th percentiles to address outliers; controls include country and year fixed effects and governance controls.
- Interpretation:
  - Destination/importer digitalization is positively associated with higher reported import values relative to exporter-reported exports, implying improved digitalization in the importing country increases recorded imports and thus taxable base for VAT and tariffs.
  - Effects persist after controlling for tariffs, tax rates, development, and governance.

### V. Revenue gain calculation — assumptions and estimates
- Revenue gain framework:
  - Revenue Gain = τ_tax ⋅ Δ(V_imports - V_exports), where τ_tax is the tax rate of interest (VAT or tariff).
  - Rewritten using estimated impact β_imp on reported imports and change in digitalization Δz_imp: Revenue Gain = τ_tax ⋅ 1/2 ⋅ (V_imports + V_exports) ⋅ β_imp ⋅ Δz_imp (see equation rearrangement in text).
  - Conservative assumption: importer digitalization increases reported imports without affecting exporter reports; denominator imports held constant (biases estimate downward).
- Assumptions used in back-of-the-envelope exercise:
  - Reduce distance to the digitalization frontier for each importer by 50 percent: Δz_imp = 0.5 * (1 - z_imp).
  - Use country-specific VAT and weighted tariff rates and average 2016 trade flows.
  - Use β_imp = 1.158 (estimate from Table 3).
- Median revenue gains per country group from closing half the distance to the digitalization frontier (2016), Table 4 (percent of GDP):
  - Advanced Economies: VAT Revenue Gains 0.7; Tariff Revenue Gains 0.04.
  - Emerging Market Economies: VAT Revenue Gains 0.7; Tariff Revenue Gains 0.2.
  - Low-Income Developing Countries: VAT Revenue Gains 1.1; Tariff Revenue Gains 0.4.
  - EU-28: VAT Revenue Gains 0.4.
- Conclusion drawn in text:
  - Halving the distance to the digitalization frontier could raise the median VAT revenue by (text truncated in source), with the table providing the quantifications above.

*Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020245-print-pdf.pdf*

### 1.1 percent of GDP for low-income developing countries, 0.7 percent of GDP for emerging

### wpiea2020245-print-pdf - 1.1 percent of GDP for low-income developing countries, 0.7 percent of GDP for emerging

### Key findings on digitalization and trade fraud
- The paper documents a lower incidence of trade fraud when governments enhance information collection and processing through digitalization.
- Results indicate significant potential revenue gains from digitalization by reducing trade fraud.
- The estimates provide a broad range for revenue potential but do not identify which specific digital tools are most effective (for example, data matching, access to third-party information, etc.).

### Quantified potential revenue gains
- Potential additional revenue from reducing the distance to the digitalization frontier:
  - 1.1 percent of GDP for low-income developing countries
  - 0.7 percent of GDP for emerging market economies and advanced economies
  - 0.4 percent of GDP for the EU (Table 4)
- Median tariff revenue increases could be:
  - 0.4 percent of GDP for low-income developing countries
  - 0.2 percent of GDP for emerging market economies
  - 0.04 percent of GDP for advanced economies
- Note: These results are indicative of potential revenue gains because reducing the distance to the digitalization frontier is likely to require significant fiscal resources and the removal of institutional barriers.

### Mechanisms, limitations, and risks
- The transmission mechanism—how improvements in digitalization translate into better enforcement at the border—requires further analysis.
- Digitalization can itself introduce new fraud opportunities: individuals and firms may use new technology to hide sensitive information and evade taxes.
  - Example: Digitalization of Estonia’s tax administration led fraudsters to create large numbers of ghost entities to generate multiple small credit claims that fell below the threshold for audit (IMF, 2018).
- Governments need adequate administrative and institutional capacity and resources to realize digital dividends.
- The viability of policies that rely on digitalization ultimately depends on political resolve.
- The challenge is to adopt digital tools to enhance government policies while mitigating the risks associated with digitalization.

### Policy implications and research priorities
- Prioritize further work to analyze the transmission mechanism linking digitalization to enforcement effectiveness at borders.
- Conduct granular analysis to determine which specific digital tools (e.g., data matching, access to third-party information) are most useful for reducing evasion.
- Recognize and plan for:
  - Significant fiscal resources required to close the distance to the digitalization frontier.
  - Institutional reforms and capacity building to prevent new forms of fraud enabled by digital technologies.
  - The need for political commitment to implement and sustain digital reforms.

### Appendix: Data sources used in the analysis
- Bilateral exports — IMF: Direction of Trade Statistics
- Bilateral imports — IMF: Direction of Trade Statistics
- Common currency — CEPII: Gravity Dataset
- Common official/primary language — CEPII: Gravity Dataset
- Common religion — CEPII: Gravity Dataset
- Contiguity — CEPII: Gravity Dataset
- Control of corruption — WB: World Governance Indicators
- Digital adoption index — WB: World Development Report 2016
- E-Government index — UN: E-Government Survey 2016
- Exchange rate — IMF: World Economic Outlook
- GDP — IMF: World Economic Outlook
- GDP per capita — IMF: World Economic Outlook
- Government effectiveness — WB: World Governance Indicators
- Government success in ICT promotion — WEF: The Global Information Technology Report 2016
- Inflation rate — IMF: World Economic Outlook
- Online service index — UN: E-Government Survey 2016
- Origin is GATT/WTO member — CEPII: Gravity Dataset
- Patents filed by residents — WB: World Development Indicators
- Population weighted distance — CEPII: Gravity Dataset
- R&D expenditure (percent of GDP) — WB: World Development Indicators
- Regional trade agreement — CEPII: Gravity Dataset
- Rule of law — WB: World Governance Indicators
- Tariff rate (weighted mean) — WB: World Development Indicators
- VAT rate — IMF: Tax Rate Database

*Source: wpiea2020245-print-pdf (sections excerpted).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020245-print-pdf.pdf_