Taxing data when the United States disagrees

Abstract

What is the best way to tax data-driven business models without contravening the existing global quasi-constitutionalist order on tax, trade, and investment law? A number of countries in Europe and around the world have begun imposing standalone digital services taxes pending multilateral agreement on a coordinated reform of bilateral income tax treaties (aka the OECD-G20/Inclusive Framework’s ‘Pillar 1’). But if Pillar 1 fails to materialise and countries go forward with unilateral digital services taxes, the United States and U.S. firms will likely seek redress using domestic measures as well as trade and investment treaties where applicable. This Article argues that in the face of such U.S. resistance, EU member states and countries elsewhere ought to reconsider using the income tax system to achieve their goals instead. We first review the events that led countries to avoid the income tax in favour of standalone taxes only to become embroiled in domestic U.S. trade policies. We then explain how European and other source jurisdictions for business services related to data collection, mining, and commercialisation could revisit the income tax to get to the same tax base, namely by taking an ambulatory interpretation approach to provisions in existing tax treaties in a way that renders possible to accommodate withholding taxes on those services. We show that an ambulatory interpretation approach could achieve the underlying goals of taxing data-driven businesses, in some cases even without any domestic law or treaty reform, with treaty-based rules for dispute resolution a ready tool to draw upon if and when the United States disagrees.

1. Introduction

Why might countries seek to tax data-driven firms, and if there are good reasons to do so, what kind of rules would be most feasible to adopt without contravening the existing global quasi-constitutionalist architecture on tax, trade, and investment law? An independent surtax on a specified base, such as the digital services taxes that have proliferated in Europe and throughout the world over the past several years, appeared at one point to be the most viable option for taxing highly digitalised companies, ¹ given the absence of a multilateral agreement to reform the international treaty-based tax regime. Yet the United States quickly demonstrated its opposition when it declared those taxes to be tariffs deserving of retaliatory measures. The Organisation for Economic Co-operation and Development (OECD) met the U.S. opposition by formulating a reform plan called Pillar 1. However, this multilateral solution now appears equally vulnerable to U.S. resistance, and if the United States obstructs Pillar 1, other countries will revert to unilateral digital services taxes. The foreseeable scenario is that the United States will retaliate with countervailing tariffs.

This Article shows that there is (and always has been) a potentially more viable alternative to standalone digital services taxes, namely: bringing specified data-related income streams under domestic income tax withholding provisions. If existing national regimes are insufficiently clear on the matter, lawmakers at both the domestic and the EU level could consider explicitly defining such income streams as a distinct income category subject to withholding in their own right. Although some observers might protest that this kind of reform conflicts with various international tax norms or practices, the OECD at one point explicitly acknowledged that withholding might be a feasible tool for taxing the data economy. Nevertheless, the OECD quickly abandoned the idea with virtually no analysis. In light of the subsequent stalling of progress on the proposed multilateral solution and the clear risks that continue to attend to unilateral digital services taxes, revisiting the withholding option appears to be the better strategy.

This Article accordingly argues that given the range of possible legal challenges that the United States and U.S. firms might bring against new taxes created by other countries outside the international tax regime, extending to data-related service fees the withholding income-based tax treatment traditionally applied to outbound payments might currently be the best alternative to tax data. In Section 2 we investigate possible barriers to data taxation arising from the global constitutional-like order on tax, trade, and investment, by examining legal and political challenges that the United States and U.S. firms have pursued or are likely to pursue in opposition to foreign countries’ attempts to tax data firms via unilateral digital services taxes or a multilaterally coordinated redesign of tax treaties. In Section 3 we argue that returning to the international income tax system, specifically by working strategically with treaty provisions that allow withholding taxes at source, such as royalties, other income, and technical services, might forestall such retaliatory action. In Section 4 we demonstrate that the proposed approach is legally defensible because international law allows various methods of treaty interpretation, including contextual and purposive readings that could be used to validate the withholding solution even when a standard tax treaty is involved.

2. The global quasi-constitutionalist order of tax, trade, and investment

The early adopters of digital services taxes designed them to fall outside the income tax system in order to avoid direct clashes with tax treaties on income and capital. ² India took the lead with a 6 per cent-rated ‘Google tax’ (formally called an equalisation levy) in 2016, followed by Hungary’s introduction of a 7.5 per cent-rated digital services tax that was, as of July 2019, reduced to 0 per cent. ³ In 2017, the United Kingdom issued a position paper accompanied by a consultation process and the year after the European Union proposed but did not adopt a digital services tax directive on revenues from targeted advertising and digital interface services. ⁴

In January 2019, Belgium proposed a 3 per cent tax on the sale of user data that was first rejected but then reintroduced in July of the same year. ⁵ New Zealand and France released proposals in February and March of 2019, with France adopting, later in the year, a retroactive 3 per cent-rated digital services tax on gross revenues from the provision of digital interfaces, targeted advertising, and user data transmission. ⁶ Entering into force of the French tax was delayed to the end of 2020 due to US threats of retaliation. ⁷

In November 2019, the Czech Republic proposed a 7 per cent tax on revenues from targeted advertising, user data, transmission, and digital services. ⁸ In December 2019, Canada announced a plan to impose a 3 per cent tax on sales of online ads and user data that would only enter into force in 2024 after stalled negotiations on a multilateral solution at the level of the OECD. Also in December 2019, Slovenia opened a consultation on a proposed digital services tax and Slovakia announced it would submit a draft legislation. ⁹

In January 2020, Austria adopted a 5 per cent tax on revenues from online advertising or any type of software or website rendered in the country, Italy adopted a 3 per cent tax on advertising on a digital interface, multiplatform for buying or selling goods and services, and transmission of user data generated from a digital interface, Latvia commissioned a study on a 3 per cent-rated digital services tax, and Norway announced the intention to introduce a tax in 2021 in case a multilateral agreement was not reached by the OECD by the end of 2020. ¹⁰ In February 2020, Spain adopted a 3 per cent tax on revenues from online advertising and user data transmission. ¹¹ In March 2020, Turkey adopted a 7.5 per cent tax on advertising services via digital platforms, digital sales, and services provided on digital platforms, anticipating the possibility of raising the rate to 15 per cent. ¹² Also in March 2020, Indonesia introduced a digital permanent establishment coupled with a specific tax on e-transactions (program-making and special events) applying to foreign merchants, service providers, and digital platforms. ¹³

By April 2020, India had announced the expansion of its equalisation levy to include e-commerce supply and services, the United Kingdom adopted a 2 per cent-rated digital services tax on revenues of search engines, social media platforms, and online marketplaces, and Poland announced a plan to introduce a 1.5 per cent surcharge on revenues from video-on-demand platforms. ¹⁴ In May 2020, Brazil proposed to apply its contribution for intervention in the economic domain (CIDE—contribuição de intervenção no domínio econômico) to digital advertising, service, and user data transmission revenues, with a progressive rate structure of 1 per cent, 3 per cent, and 5 per cent. ¹⁵ In June 2020, the European Union reintroduced a proposal for a 3 per cent digital services tax on revenues from targeted advertising and digital interface services. This proposal failed to obtain unanimous approval by member states and in 2023 the Commission submitted but then suspended an EU digital levy directive pending completion of Pillar 1 negotiations. ¹⁶

Summing up the state of digital services tax adoption in Europe in 2024, Austria, Denmark, France, Hungary, Italy, Poland, Portugal, Spain, Switzerland, Turkey, and the United Kingdom have all implemented the tax, Belgium and the Czech Republic have so far only put forward proposals, and Latvia, Norway, Slovakia, and Slovenia have either announced a plan or shown intentions to do so. ¹⁷

Against this background, the following sections examine the legal and political barriers that the United States and U.S. firms have leveraged or are likely to leverage in opposition to these taxes. Such barriers arise from the confluence of global power imbalances with various treaty-based tax, trade, and investment norms that form a constitutional-like order that sets a framework for cross-border business activities while constraining individual state action and reform. ¹⁸ Beyond what countries voluntarily agree to under treaties on tax, trade, or investment, there is no overarching applicable legal system that restricts the exercise of the jurisdiction to tax. ¹⁹ However, a combination of practical enforcement difficulties as well as the historical and contemporary tax policy choices of economically influential states form a recognisable transnational legal order that effectively influences the range of policy choices states view as feasible at any given time. It is within this order that barriers to digital services taxes have arisen.

A. Barriers to digital service taxes

In a technical sense, digital services taxes are distinguished from income taxes in that the former are flat taxes on gross receipts (commonly referred to as an excise tax) while the latter are traditionally imposed on receipts net of expenses and losses. That said, this technical distinction is incoherent in that income tax regimes commonly feature gross-basis withholding taxes on payments made by domestic payors to foreign recipients. ²⁰ Gross-basis withholding taxes are all but economically indistinguishable from any other excise tax. From an economic perspective, gross receipts are not income because income is a net concept, yet these gross-basis taxes are a constitutive feature of the international income tax landscape because of the basic logistical problems states face in collecting taxes from persons whose assets are beyond the jurisdiction of the tax authority.

The United States initiated an internal investigation into France’s digital services tax in mid-July 2019, publishing its findings in a report by end of that year. ²¹ The report declared that France’s digital services tax ‘discriminates against U.S. companies, is inconsistent with prevailing principles of tax policy and [is] unusually burdensome for affected U.S. companies’. ²² The United States accordingly threatened to impose ‘retaliatory’ tariffs, potentially amounting to $2.4 billion, on various French products, including champagne, wine, cheese, and handbags. ²³ But the punishment was delayed in January 2020, when then-U.S. President Donald Trump and French President Emmanuel Macron negotiated a truce, agreeing to suspend the dispute until the end of 2020. ²⁴

The intention was that the controversy would dissipate when, by the end of 2020, the OECD was expected to have brought the Inclusive Framework countries to a multilateral solution. A core element in this process is that member countries should dismantle their unilateral digital services taxes, thus eliminating any alleged violation of U.S. law. While the multilateral solution continues to be delayed, France and other countries that introduced digital services taxes have continued to collect revenues in amounts that increase every year. Estimates of collected revenues from the French digital services tax, for example, were close to €300 million in 2019, to €400 million in 2020, and to €500 million in 2021. ²⁵

Within six months of the introduction of the tax in France, then-U.S. Treasury Secretary Steven Mnuchin publicly reaffirmed U.S. opposition to digital services taxes and similar unilateral measures, in an announcement declaring that the United States was at an impasse with European countries in the OECD discussions. ²⁶ Mnuchin warned that the United States would respond with appropriate measures if countries continued to collect or adopt such taxes. Consequently, on 10 July 2020, then-U.S. Trade Representative Robert Lighthizer announced his decision to impose an additional 25 per cent tariff on French products valued at $1.3 billion as a response to the French digital services tax. ²⁷ Simultaneously, Lighthizer initiated a second set of internal investigations into digital services taxes of another 10 countries in Europe and elsewhere plus the European Union as a block. ²⁸

The pattern of U.S. resistance to foreign digital services taxes suggests that adopting new standalone taxes aimed at data is an unstable policy choice for most countries. Avoiding the income tax as a policy instrument not only failed to insulate European and other countries from controversy, but it evidently expanded the legal pathways for resistance by the United States and U.S.-based firms. In considering what policy strategies might be more feasible, it is helpful to understand the legal components in place that enable the United States to label foreign taxes as unfair trade practices and impose retaliatory tariffs without any process for review or appeal. This requires some familiarity with the interaction between U.S. domestic trade law and World Trade Organization (WTO) law, as described in the next Section.

B. The ‘aggressive unilateralism’ of the U.S. trade authority

The legal framework for U.S. resistance to foreign digital services taxes lies at the intersection of domestic U.S. trade law and WTO dispute resolution processes. The domestic component of this framework is section 301 of the US Trade Act of 1974, ²⁹ now codified in Title 19 of the U.S. Code (but still colloquially referred to as section 301 or §301). ³⁰ This law grants the U.S. President the right to unilaterally suspend or withdraw trade agreement provisions or impose import restrictions on foreign goods and services under specified circumstances. ³¹ Today, the authority to investigate and determine whether a foreign practice is objectionable, and to establish and carry out retaliatory actions, rests with the U.S. Trade Representative, currently Katherine Tai. ³² The range of available retaliatory measures afforded by §301 includes imposing duties or other import restrictions, withdrawing or suspending trade agreement concessions, or entering into binding agreements with foreign governments to ‘either eliminate the conduct in question (or the burden to U.S. commerce) or compensate the United States with satisfactory trade benefits’. ³³

Following the 1994 establishment of the WTO, the United States committed to using the WTO’s Dispute Settlement Understanding rather than its unilateral §301 authority whenever it investigated an issue involving an alleged violation of a WTO agreement. ³⁴ This is substantively as well as procedurally impactful because the §301 rules address general notions of ‘unreasonable or discriminatory’ practices that ‘impede or restrict U.S. commerce’, all as defined within U.S. domestic law, while the WTO documents set out specific parameters for trade practices among the members. Prior to the introduction of the WTO Dispute Settlement Understanding process, authors had criticised the §301 investigative process as a form of ‘aggressive unilateralism’. ³⁵ In agreeing to the WTO framework, the United States ostensibly answered these criticisms by voluntarily reducing its own range of policy space for unilateral trade retaliation, at least in the case of WTO Member States.

The level of U.S. fidelity to this commitment has been inconsistent over the years, but the Dispute Settlement Understanding process has acted as an effective check in the past. For example, the United States used §301 during 1998–2000 to compel other countries to eliminate trade barriers and open up their markets to U.S. suppliers. ³⁶ But some countries successfully deployed the Dispute Settlement Understanding process to have the U.S. actions declared WTO-inconsistent for their explicit breach of WTO obligations, thus permitting countries to impose punitive duties against the United States. ³⁷ Similarly, in a 2018 dispute between China and the United States over another §301 investigation, a WTO panel found that certain U.S. tariffs violated WTO terms. ³⁸

However, since July 2017 the United States dismantled the Dispute Settlement Understanding by blocking appointments to its Appellate Body. ³⁹ A WTO panel may still be assembled to review allegations of trade violations among the member states, but now an opposing party in any dispute can obstruct any WTO-sanctioned retaliatory measures simply by lodging an appeal. Since there is no one to hear the appeal, the matter is suspended indefinitely. For this reason, authors have come to describe the WTO dispute resolution process as one that involves appealing ‘into the void’. ⁴⁰ This means that until the WTO process is restored, the U.S. Trade Representative has effectively unopposable authority to impose measures against any foreign actions that she deems unfair or detrimental to U.S. commerce. ⁴¹

With this dispute resolution impasse in place, the United States is using its unilateral §301 process to apply what it characterises as ‘retaliatory’ tariffs whenever its internal investigations find foreign conduct to be objectionable. This approach operates outside the scope of WTO dispute resolution procedures, focusing solely on the U.S. domestic legal framework.

Specifically regarding recent §301-based actions against digital services taxes, various scholars have opined that the U.S. Trade Representative has repeatedly failed to make a compelling legal and factual case for trade retaliation. For example, Stephen Shay calls the U.S. Trade Representative’s report on France’s digital services tax ‘weak and unpersuasive’, as the presented arguments ‘range from dubious to wrong’ and reflect a ‘misalignment of expertise with the task’. ⁴² Further, Shay notes that the investigative report employed no ‘objective standard’ in its analysis. ⁴³

Chris Noonan and Victoria Plekhanova have examined issues raised by the U.S. Trade Representative with respect to discriminatory intent or outcome, deductibility against the corporate income tax for domestic companies, retroactivity, incompatibility with international tax principles, economic impact on businesses, and the justifications provided by France for introducing the tax. ⁴⁴ Nooman and Plekhaova find little support for using §301, further noting that ‘[t]he “discriminatory” effect is also within the power of the US to correct through permitting DST [digital services tax] paid in a foreign jurisdiction by a US resident to be a deductible expense for the purpose of US CIT [corporate income tax]’. ⁴⁵

In a thorough analysis in light of the WTO’s international trade law regimes, Alice Pirlot and Henri Culot conclude that ‘arguments based on WTO law do not provide a convincing ground to oppose such taxes’. ⁴⁶ Pirlot and Culot outline three main points. First, it is not obvious that the services targeted by digital services taxes are ‘like’ or ‘comparable’ to services provided by other companies that are not taxed. Second, allowing the digital services tax to be deducted from the corporate income tax base is a regular practice in accounting for deductible expenses when calculating taxable profits, which therefore cannot be equated with a ‘gain’ or ‘advantage’ for domestic companies. Third, the fact that multiple countries agree that the international tax regime is not fit for the digital economy weakens the claim that the tax was introduced with a discriminatory purpose.

Despite the views expressed by these and other scholars, a destabilised multilateral trade regime and an aggressively unilateral U.S. stance creates a world in which resistance to digital services taxes will likely continue to be expressed in the form of effectively unappealable trade-based retaliation measures. Since there exists no legal process to resolve the matter, some countries have delayed the effect of these measures by negotiating delayed implementation of their digital services taxes pending the multilateral adoption of coordinated reform via the OECD’s Inclusive Framework, as explained in the next Section.

C. Barriers to Pillar 1

A postponement of various countries’ digital services taxes and the U.S. trade retaliation measures emerged in October 2021 with a moratorium set by the OECD Inclusive Framework. The moratorium required all parties to remove or abstain from enacting digital services taxes or similar measures from 8 October 2021, to 31 December 2023, while a multilateral solution was underway. But the United States itself complicated the success of these multilateral efforts, not least by signaling its own disinclination to adopt conforming legislation. ⁴⁷ On 20 February 2023, Pillar 1 became blocked once again by disagreements voiced by the United States, India, and Saudi Arabia.

However, the clock on the moratorium has passed. Once the OECD’s Pillar 1 multilateral convention was not in effect by 31 December 2023, members of the OECD Inclusive Framework were politically free to restore or impose new digital services taxes, ending their pledge to refrain from doing so. ⁴⁸ Unsurprisingly, as a result of stalling negotiations and the U.S. reluctance to find a multilateral resolution, some countries signaled their intention to revert to digital services taxes as the moratorium expired. For example, as of 2024, the previously delayed Canadian digital services tax legislation entered into force, despite the multilateral commitment. Canada’s Parliamentary Budget Office estimates that over the course of five years, the tax would increase federal revenues by CA$7.2 billion. ⁴⁹

In July 2023, the OECD released a new statement announcing the moratorium on new digital services taxes would be extended by one year to 31 December 2024, provided certain conditions are met, with the possibility of another extension to 31 December 2025, depending upon progress on the multilateral convention to implement Pillar 1. ⁵⁰ Some jurisdictions, including Canada, rejected the one-year extension, despite growing U.S. threats of retaliation under §301. ⁵¹

Since the United States is the home jurisdiction of most big data companies and the world’s largest exporter of data-related services, it is easy to see why the United States has been so resistant to both multilateral and unilateral solutions to the taxation of digital profits. ⁵² It is also easy to see why the United States, as a superpower within the global political economy, has not shied away from trade-related threats against a multitude of countries, despite the questionable grounds of the §301 investigations. Accordingly, the content and scope of U.S. cooperation in the rollout of Pillar 1 remains to be seen. However, if countries are committed to taxing data as a policy choice, navigating the potential legal and geopolitical barriers to doing so in the likely scenario of no multilateral reform is key.

In particular, owing to the impasses created by the United States with respect to both Pillar 1 and digital services taxes, it might be time to reconsider the possibility of countries adopting income-based tax measures to capture more of the wealth generated by data-driven firms. Engaging the income tax system, including tax treaties, potentially disentangles the taxation of data firms from the trade and investment regimes. In particular, withholding taxes on advertising fees and other forms of income associated with the sale of customer data may be more resilient to a legal challenge through trade and investment agreements, depending on the country and the content of their relevant cross-border agreements. When imposed under a general income tax system, such taxes are usually respected under trade and investment agreements. ⁵³ Under the General Agreement on Tariffs and Trade (GATT), a longstanding view posits that, since the regime concerns import barriers to goods, covered taxes are mostly indirect ones, ⁵⁴ but this view is not universally shared among scholars. ⁵⁵ Under the General Agreement on Trade in Services (GATS), Articles XIV(e) and XXII(1) establish that the nondiscrimination principles, namely most favoured nation (Article II) and national treatment (Article XVII), cannot be invoked in respect to measures dealt with by double tax treaties.

Withholding has ample precedents. Many non-OECD countries, notably in Latin America, have long imposed withholding taxes on technical services and technical assistance fees paid to taxpayers abroad even though some OECD countries (and presumably many taxpayers) would argue that such fees should be taxed only when the payee is physically present in the relevant territory. ⁵⁶ Brazil and India have been among the key players from the Global South that have experimented with different legal instruments (eg, internal regulations, interpretive acts, and treaty protocols) to expand source-based taxation by ‘dehydrating’ Article 7. ⁵⁷ Brazil, in particular, has been the locus of much national and international debate regarding its tax policy approach of equating outgoing payments for technical services with royalties or ‘other’ income within the language of tax treaties, ⁵⁸ as discussed in greater detail in Section 3 below.

3. Why withholding is better

As seen above, standalone digital services taxes are likely to lead to a trade war as the U.S. turns to unilateral countermeasures. The OECD is currently in the process of developing a multilateral solution in the form of a negotiated expansion of taxation at source but there is no guarantee that this solution will conclude in a timely manner, or with the cooperation of the United States. As such, it is wholly appropriate for EU member states and other countries to consider income-based measures to achieve their shared domestic policy goals. In this Part, we demonstrate that income tax-based withholding on specified payments is commendable as a well-established, as well as normatively and legally justifiable, approach. We acknowledge the range of arguments against such reform, but argue that on balance, it is more feasible to use the income tax than a standalone excise tax to get at the desired income stream.

First, we explain why withholding is a common way of taxing the kind of income that is at stake in the debate over taxing data, and we note the role of tax treaties in both implementing and restricting source-based withholding. Second, noting that the space for policy reform is limited by existing tax treaties, we suggest three possible categories for an income-based withholding tax on data fees, namely as royalties, as ‘other’ income, and as technical services, in each case carving these income streams out of what would otherwise presumably be categorised as business profits. Our argument is that states could tax data income via withholding even in the presence of tax treaties. In Section 4 thereafter, we explain why withholding on data-based income is consistent with the overall context and purpose of tax treaties, as applied within the context of the modern, highly digitalised economy. ⁵⁹

A. Withholding as an appropriate income tax strategy

Withholding is a familiar way to tax cross-border income flows because it attaches to a local payor that can be linked to a local source. ⁶⁰ The justification for withholding taxes in general is to ensure the state can enforce national tax laws in a consistent manner. With respect to cross-border transfers, the idea is that a source state will find it difficult to pursue payment directly to a nonresident who lacks sufficient physical contact with the jurisdiction. ⁶¹ The source state’s best option is therefore to turn to the local payor of the income to impose an obligation to report the payment to the tax authority and, in some cases, withhold tax as well. Typically, withholding is imposed on a gross basis since the payor is presumed to lack sufficient information to calculate the taxpayer’s net income. ⁶²

Even though gross-basis taxes are in nature more of an excise than an income tax, the withholding mechanism is a staple in income tax systems around the world. Residence states, even in the absence of an agreement to mitigate double taxation, frequently provide foreign tax credits for amounts withheld at source. ⁶³ If there are historical barriers to withholding, they would generally be compliance based, in that countries might find it challenging to attach tax burdens to individuals or assets in respect of foreign entities whose physical ties to the jurisdiction are low or non-existent. Domestic law, accordingly, should be able to place data income – as a special form of private earning that is deeply connected to the territory of the source state – under existing withholding tax regimes. ⁶⁴

There are, however, conceptual barriers to doing so. The most challenging of these may be found in the delineation of different types of income into categories, which are reflected in tax treaties. Under most income tax systems, the types of income that data-driven firms earn from most market countries in Europe and elsewhere would almost certainly be characterised as business profits as opposed to property income. This is an uncontroversial observation since the profits generally arise from the act of combining labour and capital (a classic marker of business income) to gather and mobilise user data in various ways and sell advertising space on media platforms to vendors seeking to reach a customer base in a given location. As such, a threshold difficulty in building the case for withholding is to explain why it would be appropriate for countries to extricate from business profits only those income streams earned through commercialising user data or selling advertising.

An answer to this challenge is that it has always been the case that a given stream of income may be simultaneously characterised as both business profits and a subcategory of such profits, such as royalties, and when both apply, treaties prioritise the subcategory. ⁶⁵ This hierarchy is clearly seen in the way that tax treaties reflecting the OECD, UN, and U.S. models deal with the interaction of the business profits and property income provisions. In each model, the article on business profits starts with the familiar threshold rule that a state will only tax the business profits of a non-resident to the extent connected to a local permanent establishment. Later in the article, however, the model explicitly preserves the source-based taxation of specific forms of business profits when covered by another article, namely those delineating negotiated withholding rates on specified property income streams. Thus, each of the models states that when profits include items of income ‘dealt with separately in other Articles of this Convention’, those Articles ‘shall not be affected by the provisions of this Article’. ⁶⁶

OECD Model Commentary notes as a ‘rule of interpretation’ that this provision ‘gives first preference’ to the specific income articles on dividends, interest and so on, such that the business profits article applies to business profits that ‘do not belong to categories of income covered by’ the other articles. ⁶⁷ The OECD further notes its understanding that ‘the items of income covered by the special Articles may, subject to the provisions of the Convention, be taxed either separately, or as business profits, in conformity with the tax laws of the Contracting States’. ⁶⁸

Therefore, to the extent the specific types of income data-driven firms earn may be covered by the other articles, the structure of tax treaties already accommodates withholding at source even in the absence of a permanent establishment. The question is whether the parties that negotiated a particular tax treaty can be said to have intended that data-related income streams be covered by the property income articles as opposed to the business profits articles. Judging from the decision European lawmakers made to avoid entangling themselves in tax treaty interpretation by adopting standalone digital services taxes, it seems that the parties cannot be said to have intended this outcome.

Yet, by adopting digital services taxes, the relevant tax treaties have not been tested to determine how they would interact with a domestic extension of the withholding regime to cover data-based income streams. Instead, by adopting standalone digital services taxes, lawmakers have tested the conformity of these taxes with existing trade and investment regimes. The ensuing procedural battles described above suggest that the prospects for conformity are uncertain at best.

For this reason, it seems appropriate to alter the approach. The question is whether lawmakers can apply new withholding taxes under existing tax treaties (in particular, but not exclusively, those with the United States). To answer this question requires a detailed analysis of the relevant agreements in play. Some treaties are more restrictive than others, and some treaty terms are less ambiguous than others. ⁶⁹ In the discussion that follows, we explore which of the ‘special’ income categories as described in the OECD Model Commentary might encompass data-based profit streams, and we provide a contextual and purposive interpretation to make the case for countries in Europe and around the world to trigger this interpretive exercise to separately delineate data-based profit streams.

B. Three ways to withhold on data-based income (when a treaty is involved)

There are many ways to alter tax systems to cover data services via bilateral or multilateral agreement. In 2017, the European Commission rejected a digital services tax, and then proposed a directive in 2018 that failed to obtain unanimous support. ⁷⁰ In June 2020, the Commission reopened discussions and by 2023 it had put forward a EU digital levy that was, nonetheless, postponed until finalisation of the Pillar 1 process. ⁷¹ Given that this EU digital levy would create the same kinds of trade frictions with the United States as unilateral digital services taxes, this section explores the possibility of EU member states – or any other country – imposing withholding taxes at source even when the traditional treaty-based permanent establishment standard is not met. The approach could be pursued unilaterally or promoted via an EU withholding directive on ‘special’ income categories encompassing data-related payments. The first two of these payments, namely royalties and other income, are already laid out in tax treaties so the question is one of interpretation. The third, technical services, is contemplated in the UN Model but we argue could also be read into some tax treaties via dynamic contextual and purposive interpretation.

Royalties

Some types of data services income comprise location-specific rent, which are structurally similar to the proceeds from the exploitation of natural resources that are taxed under resource rent taxes. ⁷² As a form of economic rent that arises within the territory of the source state, ⁷³ domestic law ought to be able to assign such earnings to treaty provisions that allow source taxation, specifically the royalties article. ⁷⁴ That said, domestically classifying data fees charged by foreign-based companies as royalties only makes sense if the relevant treaty provision allows source-based withholding.

For this approach, the interpretive work to be done is in the relevant definitions. In defining royalties, tax treaties often mention ‘payments of any kind received as a consideration … for information concerning industrial, commercial or scientific experience’. Data companies make profits by selling or monetising information concerning the experience of users and consumers in digital platforms that are operated and maintained by those companies. This unique attribute makes data-based income distinct from the traditional category of business profits that is characterised by productive processes carried out by an enterprise through the application of capital and labour without intense user participation. ⁷⁵ Indeed, data-based firms engage in profit-seeking activities that are highly dependent on gathering, monitoring, and processing valuable information related to the active engagement of customers in digital platforms, which can be conceptually assimilated to the activity of extracting a royalty from the commercial experience of users and consumers. Domestic law could, therefore, clarify the meaning of ‘commercial experience’ in royalty provisions as inclusive of income from data activities.

It is likely that bringing data fees into the royalty definition will require adopting a reform to statutory law, but the conventions surrounding the treaty-based definition of royalties make room for at least some movement in applying the domestic definition. The Commentary to Article 12 in the UN Model, for instance, recognises the difficulties in distinguishing services income from royalties in the context of the expression ‘information concerning industrial, commercial or scientific experience’, given the broad meaning of that phrase. ⁷⁶

Where source-based withholding is preserved under a treaty, a broadly defined provision will be most amenable to interpretation that includes service fees of various kinds. ⁷⁷ But even tax treaties with royalty provisions modeled after Article 12(3) of the UN Model (or the OECD Model) may encompass domestic law including data fees within the concept of ‘payments of any kind received as a consideration for … the right to use industrial, commercial or scientific equipment or for information concerning industrial, commercial or scientific experience’. ⁷⁸ Here, the relevant experience could be that of the local user who provides monetisable data or the local customer who uses advertising space, in both cases by actively engaging in online search engines, social media platforms, online gaming, cloud computing, and the like.

Other income

If the tax treaty contains an ‘other income’ article and allows source-based withholding (as is the case in treaties that follow the UN Model rather than the OECD Model), a second option is to classify data fees charged by foreign-based, data-driven firms as falling under the category of ‘other’ income instead of traditional business profits. This approach, however, would likely require formal statutory law reform in order to survive judicial review.

The main difficulty with this approach lies in a conventional understanding among legal scholars that the categories of (international) income that may be subject to withholding comprise a closed list, to which new types of income cannot be simply added without re-negotiating the tax treaty-based network. ⁷⁹ Yet several factors mitigate against this view.

To read the ‘other’ category as impervious to change is to detach the interpretation of the law from the evolving reality of business profit production. Reading the business profits articles without considering the holistic context in which they were drafted would result in an unreasonable and absurd restriction on states’ power to tax cross-border transactions, counter to the good-faith intentions of treaty partners in accepting the idea of physical permanent establishments to begin with.

In 2000, for example, Brazilian federal tax authorities issued a declaratory act according to which ‘[r]emittances under contracts for the provision of technical assistance and technical services without the transfer of technology’ were to be classified as ‘other income’ under Brazil’s tax treaties even when the treaty did not have an Article 21. This position, as observed by Vanessa Arruda Ferreira, was based on the ‘interpretative argument’ that ‘since the expression “business profit” is not defined by tax treaties, the meaning to be considered shall be the one under domestic law, following the interpretation rule of the treaty corresponding to Article 3(2) of the OECD Model Convention (2010).’ ⁸⁰ For years, Brazilian courts accepted this understanding, but after much criticism from taxpayers and some scholars, courts changed their position. ⁸¹

In 2012, the Brazilian Superior Court of Justice decided the famous Copesul case, ruling that the Brazilian federal tax administration could not include technical services fees under the other income articles of the Brazil-Canada and Brazil-Germany tax treaties. ⁸² But in reaching this decision, the Court seemingly accepted the Brazilian tax authority’s ‘interpretative argument’ because the decision relied on the internal law concept of profit to determine the application of Articles 7 and 21 of Brazil-signed tax treaties. ⁸³ Since this concept makes no reservations in respect to technical services, the Court found that technical services fees were to be classified as business profits according to domestic law, consequently attracting the application of Article 7 of tax treaties, which only allows source taxation in case of a local permanent establishment.

In 2020, the Court confirmed the same understanding in respect to the Brazil-Spain tax treaty, ⁸⁴ but in this case the Court also declared that business profits treaty provisions (Article 7) should not be automatically applied to all profits arising from the provision of technical services. The Court recognised that there are situations of ‘hybridism’ where the relevant income resembles royalties or independent personal services fees. To avoid similar challenges in respect to data-related payments, EU countries and others could consider reforming their statutory tax laws, specifically in regard to the legal concept of business profits.

Technical services

Finally, a third approach involves the definition of technical services. The concept of technical services, when present in tax treaties, is not typically defined comprehensively, but domestic law could clarify the concept to include data-related fees.

In the context of a tax treaty that includes Article 12A of the UN Model, domestic law could include data services fees within the concept of technical services fees, specifically within the defining language ‘any payment in consideration for any service of managerial, technical or consultancy nature …’. Along these lines, Andrés Báez Moreno has argued that Article 12B of the UN Model is an unnecessary addition because that expression, which is already present in Article 12A, can be interpreted to accommodate automated digital services. ⁸⁵

Even if the relevant tax treaty contains only an Article 12 that both allows withholding taxation and mentions ‘technical services’ or an Article 12A of the UN Model, domestic law could define data fees charged by foreign-based companies as falling within the concept of technical services. In Brazil, for example, 28 of the 33 tax treaties contain the expression ‘technical services’. ⁸⁶ According to Revenue Normative Instruction 1455 of 6 March 2014, technical services are defined as ‘the execution of a service that depends on specialised technical knowledge or that involves administrative assistance or consultancy, carried out by independent professionals or with an employment relationship or, even, resulting from automated structures with clear technological content …’. ⁸⁷ Since the Copesul case discussed above, which denied the possibility of the tax administration using internal regulations to reclassify income differently from domestic law definitions, Brazil has opted to negotiate protocols with its treaty partners in order to amend royalty provisions to include the expression ‘technical services’ (often accompanied by ‘technical assistance’). In a recent protocol to the Brazil–Argentina tax treaty, the parties even included in the royalties provision the expression ‘automated structures with clear technological content’. ⁸⁸

Nevertheless, countries would do better in using internal statutory law reform instead of administrative regulations or treaty protocols to clarify the concept of technical services for treaty purposes. Statutory reform could be supported by the explicit reference to the domestic law of the treaty partners for interpreting undefined terms in Article 3(2) of both the OECD and UN Models as well as Article 2(2) of the Multilateral Instrument, respecting the purpose of such agreements in cases of undefined terms whose meanings cannot be derived from context or specific mutual agreement procedures. ⁸⁹

4. The contextual and purposive case for withholding

In considering whether and how a treaty could be viewed as covering data-related payments, we start with the premise that many treaties are built upon a vocabulary that dates to a century ago, that treaties are broadly written and slow to change, and that the categories laid out in treaties are modified and adapted by domestic and international practices in various ways to account for the advanced complexity of economic entities and transactions in the modern economy. As such, to work with a treaty is to work with categories of income that are rarely perfectly or completely defined. Reforming domestic law and correspondingly interpreting treaties to encompass data services should be understood as a reasonable reading of treaty concepts in light of the modern context in which these instruments operate. ⁹⁰

The interpretation puzzle involves investigating the source of the income, deciding whether income can be said to have alternative sources, and if so, determining what ordering rule applies. In particular, the logic enshrined in the business profits and permanent establishment threshold provisions of either the OECD or the UN Models (and in business profits articles in tax treaties around the world) is one that assumes that a firm cannot substantially participate in another country’s economic life without crossing a threshold of permanence, and that the income thus produced, under assumptions of physicality, is to be classified as ‘business profits’. ⁹¹ At its core, the classification and assignment achieved by this distributive rule presupposes brick-and-mortar businesses that require, at some point in time, to be physically present to be meaningfully involved in a country’s economy. ⁹² Physicality was, on this view, the best available proxy for nexus in a world in which physicality was assumed to be fundamental to the production of business income in a jurisdiction.

Yet, the permanent establishment threshold is not necessarily designed to prevent source-based taxation of different business income items by drawing a rigid line between taxation and exemption. It could instead be the case that the threshold merely defines the parameters for net taxation versus gross taxation. This construction is borne out in the Australian case of Tech Mahindra Limited v Commissioner of Taxation, in which an Indian taxpayer received payments that could be alternatively described in both Article 7 (business profits) and Article 12 (royalties) of the Australia-India treaty. ⁹³ The taxpayer argued that the payments were not taxable in Australia because they were business profits that were not attributable to the local permanent establishment, such that Australia could not return to Article 12 and impose gross-basis withholding instead. ⁹⁴ The Federal Court of Australia rejected this argument on grounds that ‘no reason or purpose is identified which would be served’ by the taxpayer’s proposed construction. To follow the taxpayer’s logic would lead to an absurd result in that Australia would have ‘no entitlement to tax the income at all’. The Court thus agreed with the Australian Tax Commissioner’s construction that in the case of payments covered by Articles 7 and 12 without an effective connection to a permanent establishment, Australia is entitled by Article 12(4) ‘to impose tax at the potentially more generous rates permitted under Article 7(1)’. The Court characterised this interpretation as ‘manifestly … the purpose which Article 12(4) is intended to serve’. ⁹⁵

If this view is conceptually difficult to accept, we should ask ourselves why that is so. If the answer is nothing more than tradition and historical practice, we should interrogate how antiquated traditions serve as interpretation standards. The permanent establishment threshold is drawn from an age in which there was virtually no way to become established in a jurisdiction without physical presence of labour and capital. ⁹⁶ That is not the world of business today.

Even if it may be clear that data services income constitutes business profits, it is also possible to view such fees as royalties, other income, or a defined category of technical services. It is a permissible act of interpretation to look for treaty wording that allows for expansion (eg, the part of the royalties article that states ‘or for information concerning industrial, commercial or scientific experience’) to encompass the underlying purpose of tax treaties, namely to prevent double taxation (rather than to facilitate non-taxation). ⁹⁷ This is an argument for an evolutionary interpretation, which is a teleological or purposive interpretative approach. ⁹⁸

An evolutionary approach is appropriate because as a legal text, a treaty provision requires interpretation, and interpretations may change over time. ⁹⁹ To read tax treaty provisions today without considering the context in which they were drafted and the evolving reality of business profit production could lead, in the words of the Vienna Convention on the Law of Treaties of 1969, to ‘a result which is manifestly absurd or unreasonable’. ¹⁰⁰ Further, the OECD and United Nations have long advocated for an ambulatory or dynamic interpretive approach to tax treaties, ¹⁰¹ and this view has been supported in the literature. ¹⁰² This view is also reflected in domestic U.S. tax law jurisprudence. ¹⁰³ Even a strained reading of an ambiguous provision can be acceptable in certain circumstances, so long as it meets the lawmakers’ intentions. ¹⁰⁴

An ambulatory approach means that the treaty should follow a contemporary meaning of its terms, rather than reference only to those concepts and economic realities existing at the time of ratification. Drawing from policy learning and exogenous factors such as technological change, a dynamic interpretation of treaties could reasonably establish that treaty partners never agreed to restrict their otherwise unlimited power to tax (at source) profits that do not depend on the firm’s physical presence to carry on its business.

To some extent, the Vienna Convention acts as a constraint on the ways in which authorities may interpret or apply tax treaties. ¹⁰⁵ On the other hand it facilitates the emergence of a revised understanding over time in that the interpretive rules of Articles 31–3 accept an evolutionary or evolutive approach. ¹⁰⁶ The analogous ambulatory approach advocated by the UN and OECD for tax treaty interpretation has been extensively discussed in the literature, particularly in relation to updated OECD commentaries to its model convention. ¹⁰⁷ Even though the appropriateness of this doctrine has raised disagreement amongst legal scholars, ¹⁰⁸ a dynamic approach is in line with the general rules set forth in the Vienna Convention that require that a treaty be interpreted ‘in good faith’ and ‘in light of its objects and purpose’ so as to avoid ‘a result which is manifestly absurd or unreasonable’. ¹⁰⁹

Reading tax treaties in a contextual and purposive manner is a core interpretive requirement, and doing so should assist lawmakers seeking to defend the expansion of withholding taxes at source against a categorical rejection on doctrinal grounds. The goal is to preserve taxing rights that would never have been surrendered had the rise of digitalisation been anticipated. The underlying rationale is that relevant conceptions have changed in such a way that to continue to adhere to conventional definitions or to rely on plain meaning readings of treaties would be contrary to the broader good-faith intentions of the parties as well as the object and purpose of designing different categories of income and physical presence thresholds in the first place.

Accordingly, an interpretation of treaty terms that validates a legal characterisation of data service fees as royalty, other income, or technical services (grounded on Article 3(2) of tax treaties in combination with Articles 31 and 32 of the Vienna Convention) need not be viewed as permissive of treaty override. Rather, such an approach may be viewed as accounting for an evolving consensus surrounding new economic realities in order to realise the treaty’s broader mandates.

5. Conclusions

When policymakers consider the choice between adopting a standalone tax on data service-related transfers or working with the existing corporate income tax regime, bilateral and multilateral tax, trade, and investment agreements are ever-present factors. Using income taxes instead of a standalone digital services tax will not immunise the European Union and countries elsewhere against U.S. resistance, but it would likely circumvent the immediate trade retaliation that seems sure to arise.

In 2015, the OECD first acknowledged a gross-basis final withholding tax as a possible measure to capture the profits created by highly digitalised firms, ‘provided they respect existing treaty obligations or [adopt changes] in their bilateral tax treaties.” ¹¹⁰ It cautioned that countries and firms might raise international trade law and EU law challenges in response yet recognised that the different agreements carry distinct implications for tax measures. In particular, the OECD noted that the GATS provides broad exceptions in situations involving tax treaties and for the imposition of direct tax provisions aimed at ensuring the equitable or effective imposition of direct taxes, while the GATT prohibits parties from subjecting imported products to taxes in excess of those that would apply to similar products produced domestically. ¹¹¹ Regardless of these distinctions, it seems clear that the OECD’s overall preference is to sidestep trade and investment entanglements in favor of forging a coordinated multilateral solution, even if doing so requires renegotiation of existing income tax treaties. ¹¹² But a coordinated multilateral solution is by no means the only possible approach involving income taxes; unilateral and EU-level tools exist and in many circumstances are a more feasible option.

To the extent that there are concerns about violating treaties, it is useful to recall that disputes about the reach, scope, and meaning of treaties are very often resolved through a diplomatic process that involves designated officials whose discretion to resolve matters with treaty partners is broad according to the terms of the treaty. Thus, in the OECD Model, the designated treaty dispute resolution personnel – the so-called ‘competent authorities’ of the respective treaty partners – are authorised not only to endeavor to resolve ‘any difficulties or doubts arising as to the interpretation or application of the Convention …’, but they ‘may also consult together for the elimination of double taxation in cases not provided for in the Convention’. ¹¹³ This is an expansive delegation of authority that in effect empowers the treaty partners to achieve ends not necessarily laid out in the treaty. The scope of competent authority powers should accordingly inform lawmakers who are considering how to bring data-based payments within scope of the taxing power.

In sum, digital services taxes and similar measures are meant to offset income tax revenue losses that market states face in an increasingly digitised world. But imposed as stand-alone excise taxes, these measures have attracted retaliatory trade-based action from the United States which cannot currently be effectively challenged before the WTO. Bringing digital services taxes into the income tax system would reduce some of this retaliation risk. A dynamic approach to interpreting existing tax treaty concepts may thus provide a more reliable way of achieving the objectives of digital services taxes, with less friction.