Digital trade tariffs could raise revenue—but the gains depend on uncertain assumptions

A comparison of hypothetical tax regimes

In a preprint submitted to arXiv on November 3, 2026, Sonja Mitikj and Viktor Stojkoski compare digital services taxes, or DSTs, with hypothetical customs duties on digital products. The supplied metadata establishes preprint status, not peer-reviewed publication.

The authors frame their analysis around what they describe as the March 2026 lapse of the WTO moratorium on customs duties on electronic transmissions. They report that no government had yet imposed such duties and that some members retained duty-free commitments among themselves. Their calculations therefore concern possible policies, not measured outcomes from digital tariffs already operating.

DSTs generally tax revenue from selected digital activities, subject to coverage rules and business-size thresholds. The study uses a broader grouping that also includes equalisation levies and certain gross-basis or withholding taxes. These instruments are not identical to tariffs, nor to value-added tax, which generally applies to domestic and imported consumption alike.

What the researchers measured

The underlying dataset estimates trade across 189 countries and 31 digital product categories during 2016–2024. It covers software, games, advertising and platform intermediation fees, among other activities—not every service delivered online. Its bilateral flows are constructed from corporate revenue information and observed or model-imputed consumption patterns, rather than directly observed customs transactions.

The main regression and scenario sample is narrower: 112 supplier origins and 68 destinations, covering approximately $652 billion in estimated trade in 2024.

To construct hypothetical tariff rates, the researchers match digital categories with physical goods and use their applied tariffs. Software, for example, is matched with recorded media. These “tariff equivalents” are benchmarks, not enacted digital tariff schedules. Some matches, including printed materials for online services, are necessarily imperfect.

The authors then estimate gravity models: statistical models relating trade to economic size, trading frictions and taxes. They also compare trade before and after DST adoption.

Association is not a demonstrated tax effect

Across destinations, one model associates a one-percentage-point higher DST rate with approximately 16% lower trade. But that relationship largely disappears when the analysis accounts for persistent differences between destinations or country pairs. Taxing destinations may differ from other markets in ways the tax itself does not explain.

The main adoption study follows 12 destinations from two years before to two years after adoption. Relative to the preceding year, trade is 7.7% lower in the adoption year and 9.7% lower the following year. Two years after adoption, the estimate is no longer statistically distinguishable from the reference level.

Crucially, trade was already declining before adoption. That could reflect an existing trend or businesses anticipating announced taxes; the data cannot distinguish those explanations. Alternative weighting gives larger, more persistent declines, while broader services data show a smaller decline lasting into the second year. The evidence does not establish that DST effects universally disappear after two years.

Revenue gains depend on the trade response

For their headline scenarios, the authors translate the adoption-year estimate into roughly 2% lower trade per additional percentage point of tax. They assume the same response for DSTs and tariff equivalents, holding other economic conditions fixed.

On the covered 2024 trade base, their calculations imply:

• DST baseline: approximately $2.7 billion in revenue.
• DSTs plus tariff equivalents: approximately $6.9 billion, with trade $9.3 billion lower, or 1.4%.
• Tariff equivalents replacing DSTs: approximately $4.3 billion, with trade $3.8 billion lower.

These are implied gross revenues excluding VAT, not observed government receipts or forecasts. They use the historical 2024 tax configuration; the supplement notes subsequent changes, including Canada’s retroactive DST repeal in 2026.

With the much stronger cross-destination association substituted for the headline response, adding tariffs reduces trade by 7.6% and raises about $4.0 billion. Tariffs alone then raise only $1.8 billion—less than the DST baseline. Separate responses derived from goods trade also overturn the tariff-only revenue advantage.

Who collects—and who ultimately pays

Tariff receipts concentrate in destinations with higher physical-goods tariffs, including Brazil and India. DST payments are attributed more heavily to subsidiary locations such as Ireland; tariff payments concentrate more on supplier origins such as the United States.

Those are accounting allocations, not evidence about who bears the economic cost. Assigning sales to parent-company headquarters changes the payment geography, and firms may pass taxes to customers.

The calculations omit administrative and enforcement costs, relocation, broader price and income adjustments, and substitution between digital and physical products. The supplied extraction includes the main paper and supplement, but not readable chart graphics; this account relies on their prose, tables and captions. The authors disclose Hinrich Foundation support. Their results illuminate a conditional revenue trade-off, not a demonstrated policy optimum.