The countdown to December 31 is on. If the OECD’s pillar 1 multilateral convention (MLC) is not in effect by that date, members of the OECD inclusive framework will be free to impose new digital services taxes, ending their pledge to refrain from doing so.
However, there are concerns the OECD might not meet that deadline. According to the OECD’s most recent estimate, the MLC will be finalized in mid-2023 and will enter into force some time in 2024. Some stakeholders are asking the inclusive framework to extend the moratorium past December 31. Meanwhile, some countries are waiting in the wings to spring digital services measures into action. Colombia just enacted a significant economic presence rule that will go into effect in January 2024; Canada and the United States continue to spar over Canada’s planned DST.
For several years, there’s been speculation about how the OECD might define DSTs and similar measures. There has also been speculation about how that framing might affect DSTs that have already been enacted. Inclusive framework members will be expected to commit to the standstill and rollback of unilateral DSTs under pillar 1, but as we have noted several times, DSTs may be difficult to remove and may have lasting power. Much of that hinges on how the OECD might define DSTs and relevant similar measures. That day is here.
In late December 2022 the OECD released a draft definition of DSTs and relevant similar measures and invited public comments on the language and approach. For such an important issue, it is surprising that fewer than three dozen commentators provided feedback. Most of those commentators represent large business interests; Kenya was the only national government to respond. However, those who responded had quite a bit to say. It is clear that stakeholders are divided over what will be more destabilizing to the international tax system: Some argued that a pillar 1 regime that mostly deters DSTs but allows room for some exclusions will not balance the system, while others argued that a pillar 1 regime that completely bars DSTs will likely conflict with jurisdictions’ constitutional laws and will be highly damaging.
These turf battles strike at an unresolved central question: What exactly is pillar 1 designed to do? This leads to a related question: What does the international tax community think pillar 1 is designed to do?
For business interest groups, such as the German Chamber of Commerce and Industry (DIHK), the United States Council for International Business (USCIB), and the Silicon Valley Tax Directors Group (SVTDG), the answer to the first question is straightforward: Pillar 1 was designed to protect the international tax system from destabilizing unilateral measures. As such, it should require the complete withdrawal and abolition of DSTs and relevant similar measures.
“If the mechanism for identifying destabilizing taxes contains gaps, then it will be ineffective and the objective of stabilizing the international tax system cannot be achieved,” USCIB said in its comment letter.
However, entities representing developing country interests, such as the government of Kenya, the South Centre, and the Latin American Tax Policy Forum (LATPF), see the situation a bit differently. Yes, pillar 1 is meant to stabilize the international system, but not at the expense of sovereignty and practicality. In separate comments, they each cautioned the OECD to be mindful of jurisdictions’ constitutional limitations and warned the OECD against imposing blanket bans on DSTs before countries are legally and practically ready to comply. Otherwise, inclusive framework delegates may sign on to a deal that their local courts could later dismantle, the LATPF told the OECD.
“The prospect of signing the MLC and having to witness its demise by the decision of local courts is devastating for the stability of the global business environment,” the LATPF said in its comments.
“More importantly, no degree of political pressure should be imposed on Inclusive Framework jurisdictions so that they modify their constitutions or governing laws to accommodate Pillar One,” it added.
What Is a DST or Relevant Similar Measure?
The OECD’s definition and treatment of DSTs and relevant similar measures will be split across two articles in its upcoming pillar 1 MLC. The rules for handling DSTs and relevant similar measures will be found in article 37, “Removal of Existing Measures,” and the definition will be contained in article 38, “Provision Eliminating Amount A Allocations for Parties Imposing DSTs and Relevant Similar Measures.” We’ll start with article 38 because that contains the definition.
According to article 38(2), the term “digital services tax or relevant similar measure” would mean:
any tax imposed by a Party, however described, if it meets all of the following criteria and is not described in paragraph 3 [which addresses exclusions]:
a. the application of such tax, or the amount of tax imposed, is determined primarily by reference to the location of customers or users, or other similar market-based criteria;
b. such tax either:
i. is applicable by its terms solely to persons that:
1. are not residents of that Party (“non-residents”); or
2. are primarily owned, directly or indirectly, by non-residents of that Party (“foreign-owned businesses”); or
ii. is applicable in practice exclusively or almost exclusively to non-residents or foreign-owned businesses as a result of the application of revenue thresholds, exemptions for taxpayers subject to domestic corporate income tax in that Party, or restrictions of scope that ensure that substantially all residents (other than foreign-owned businesses) supplying comparable goods or services are exempt from its application; and
c. such tax is not treated as an income tax under the domestic law of the Party, or is otherwise treated by that Party as outside the scope of any agreements (other than this Convention) that are in force between that Party and one or more other jurisdictions for the avoidance of double taxation with respect to taxes on income.
Commentators had several opinions on how the OECD can refine the definition and make it a sharper tool in combating unilateral DSTs.
The draft language clarifies that a unilateral measure must satisfy points a, b, and c (exceptions notwithstanding) to qualify as a DST or relevant similar measure. Several commentators, including the International Chamber of Commerce, the Digital Economy Group, USCIB, and SVTDG, argued that the tripartite test is far too strict because it could exclude some taxes that arguably are DSTs but fail to meet one or two of the OECD’s elements.
The groups want a disjunctive test: If a tax meets one element of the OECD’s definition, it should be considered a DST in violation of pillar 1. They also want an expansive test. The USCIB offered an example of what that could look like:
- the measure applies based on location of users, customers, or market-based criteria;
- the measure targets a particular industry, either by its terms or in practice;
- the measure discriminates against nonresidents or foreign-owned business, either by its terms or in practice;
- the measure creates an unlevel playing field through the application of arbitrary distinctions, such as between offline and online or between domestic and foreign;
- the tax is extraterritorial and based on gross income or income imputed from gross revenue;
- the measure otherwise deviates from international norms; or
- the measure has been determined to be discriminatory under the provisions for review of such taxes under the MLC.
Several commentators also objected to the definition’s third element, article 38(2)(c), which would exempt measures that are treated as an income tax under domestic law or treaty. Their general argument is that unilateral measures, such as withholding taxes and significant economic presence rules, are more likely than not to be packaged as income taxes. According to the SVTDG, these measures tend to be enacted or considered by countries that lack broad treaty networks, which would mean relief would be unavailable. The SVTDG, explaining the depth of its concern, wrote that article 38(2)(c) will likely be the “sole reason” that withholding taxes and significant economic provisions escape the OECD’s definition.
“Failing to include such measures within the definition makes the definition inherently incomplete,” SVTDG said. The Digital Economy Group suggested the OECD use the third element as a general factor in determining whether a unilateral measure is a DST but not treat it as a required element.
Beyond that, several commentators, including DIHK, pointed out that some DSTs apply to both domestic and foreign enterprises, rendering the domestic law part of article 38(2)(c) ineffective in addressing harmful DSTs.
And what exactly does the OECD mean by “similar market-based criteria” in article 38(2)(a)? The text doesn’t offer any background or explanation, and the South Centre asked the OECD to either delete the phrase or specify the criteria in the article, arguing that the language is vague and could generate disputes over its meaning.
Creating a Flexible Definition
While the OECD’s definition must be firm enough to capture existing unilateral measures, it also must be flexible enough to capture future ones. The OECD is planning to create an enumerated list of prohibited unilateral measures and attach it to the MLC in an annex (annex A). However, stakeholders wishing to share input on that annex will be disappointed; the OECD Task Force on the Digital Economy will be responsible for negotiating the list, but the negotiation will not be open for public input. Also, the OECD has not confirmed whether annex A will be periodically updated or what that process might look like.
Although annex A will carry a lot of weight in identifying DSTs, it won’t be the definitive word on whether a unilateral measure violates pillar 1. This is because any measure that meets article 38’s definition could also be subject to review by a conference of the parties to the pillar 1 MLC.
There are a few open questions concerning this approach. One is how the OECD will treat unilateral measures belonging to jurisdictions that join the MLC after it enters into force. Will the measures be added to the list? Another question is how the OECD will treat new measures that are labeled as DSTs or relevant similar measures based on article 38.
The SVTDG said it endorses the OECD’s approach but also asked the OECD to take things a step further and treat annex A as a guideline or interpretive tool for assessing future unilateral measures. That way, the organization wrote, if a new unilateral measure does not meet the article 38 definition, it still could be treated as a unilateral measure if it is “substantially similar” to other measures listed in annex A.
In the meantime, several commentators insisted that the OECD expand the definition of DSTs and relevant similar measures so that it applies to significant economic presence or digital permanent establishment measures.
Then there’s the issue of cultural contribution taxes on digital companies. The Digital Economy Group asked the OECD to consider cultural levies on streaming service providers, such as Netflix, which have been introduced across Europe under the EU audiovisual and media services directive. Because they are imposed on gross revenue, they have the same financial impact as DSTs, the group said.
How Much Choice for Jurisdictions?
Under article 38(1), jurisdictions cannot receive an amount A allocation if they maintain a DST or relevant similar measure:
Any Party for which a digital services tax or relevant similar measure, or a measure listed in Annex A (List of Existing Measures Subject to Removal), is in force and in effect during a Period:
a. shall not be allocated any profit under [the MLC provision allocating Amount A] with respect to that Period; and
b. shall not impose tax with respect to that Period under any domestic law provision implementing the provisions of [the MLC provision allocating Amount A].
However, the entire approach of article 38(1) is problematic for the USCIB, which wrote that the OECD’s draft would allow inclusive framework jurisdictions to make a choice: They can choose to maintain a DST and forgo their amount A allocation, or they can choose to remove all unilateral measures and receive their amount A allocation. The mere existence of that choice means that the international tax system won’t be stabilized as pillar 1 intends, USCIB wrote. The group encouraged the OECD to impose additional barriers and disincentives that steer countries away from implementing new unilateral measures and induce them to withdraw any existing measures.
Meanwhile, the BEPS Monitoring Group commended the OECD for building some choice into the rules, especially because there is uncertainty about how much revenue inclusive framework members will each gain or lose under pillar 1. Even after the MLC goes live, it may not be immediately clear to individual states what their revenue numbers will look like.
In this context, the BEPS Monitoring Group emphasized that inclusive framework members must carefully weigh all their taxing options, ranging from pillar 1 to DSTs to treaty-based withholding taxes on services based on articles 12A and 12B of the U.N. model tax convention.
That said, the issue of “choice” is far from resolved in a few different areas, and in each one, commentators have differing opinions on what approach would be most destabilizing. One issue the OECD is negotiating is whether inclusive framework jurisdictions that choose to maintain a DST will be fully denied from receiving an amount A allocation or could receive a partial allocation based on the scale of the measure.
The LATPF wrote that a partial allocation scheme would “push the application of the MLC into a very controversial space, forcing jurisdictions to attribute ‘scores’ or ‘degrees’ to the DSTs or similar measures implemented by their counterparts.” The organization believes full denial is a better solution, as do several others, including the National Foreign Trade Council, which wrote that a partial denial will fail to discourage unilateral measures and will continue to destabilize the international tax system. In contrast, the International Chamber of Commerce said it supported partial denial but did not provide a reason why.
The OECD mentioned in a footnote (footnote 1) that it is discussing exactly how inclusive framework members should refrain from implementing unilateral measures. It is unclear whether that commitment would be legal or political, but the Kenyan government urged the OECD to be mindful of members’ sovereignty, writing that it would be “impractical and difficult for a country to transfer its sovereignty by committing not to enact any legislation that impacts the digital economy.” It prefers that the OECD abandon that approach and rather address any infractions through international law.
Should Exclusions Be Allowed?
To the chagrin of several commentators, the OECD has drafted several proposed exclusions from the definition of DSTs and relevant similar measures.
Under article 38(3), the term “digital services tax or relevant similar measure” would exclude:
a. a rule that addresses artificial structuring to avoid traditional permanent establishment or similar domestic law nexus requirements that are based on physical presence (including both direct physical presence and the physical presence and activity of an agent);
b. value added taxes, goods and services taxes, sales taxes, or other similar taxes on consumption; or
c. generally applicable taxes imposed with respect to transactions on a per-unit or per-transaction basis rather than on an ad valorem basis.
Several commentators wrote that the rules need to clearly provide double taxation relief if exemptions are included. The USCIB advised that all jurisdictions joining the pillar 1 MLC should agree to provide double taxation relief and that the MLC itself should expressly allow jurisdictions to apply either a double taxation exemption mechanism or a credit mechanism, including to tax measures allowed by the OECD.
Several commentators also voiced concerns about the OECD’s planned exemption for artificial structuring-related rules in article 38(3)(a). A few commentators, including the Digital Economy Group and SVTDG, wrote that the OECD’s exemption could easily enable jurisdictions to enact destabilizing measures, using artificial structuring as a pretext. More importantly, these commentators wrote, jurisdictions would be able to do so because article 38(3)(a) is not grounded in existing abuse of law principles. Without that foundation, unilateral measures could proliferate. The SVTDG urged the OECD to install “robust guardrails” that limit the exemption to actual instances of abuse of law.
Should Pillar 1 Apply to All Companies?
Article 37(1) would prohibit participating jurisdictions from applying any unilateral measure listed in annex A to “any company.” This language particularly rankled commentators representing developing jurisdictions. The South Centre called it “one of the most egregious and unfair aspects of the Amount A rules.” Both the South Centre and the Kenyan government asked the OECD to limit that provision to amount A in-scope companies, which is an approach that developing countries have supported for some time. Also, the Kenyan government urged the OECD to implement a “transition provision” that it said may encourage jurisdictions to sign on to the MLC.
However, others approve of this blanket approach, in some cases encouraging the OECD to advance the definition further. EY, which supports the “any company” approach, wrote that “the obligation to withdraw existing measures and not to enact new measures also must be comprehensive. A measure that is in violation of this obligation should not be permitted to be applied to any company under any circumstance, consistent with the October 2021 agreement. This is the case without regard to whether the company is subject to Amount A.”
Deloitte added that the pillar 1 “multilateral convention should be updated to refer more broadly to all entities whatever their legal form including partnerships [and] trusts.”
Treatment of Nonparties
One looming question is whether jurisdictions may be allowed to apply unilateral measures against a multinational whose ultimate parent entity (UPE) is in a jurisdiction that has not signed the MLC. Some commentators said the jurisdiction that should matter for purposes of determining whether the amount A rules apply is the jurisdiction of the multinational entity, not the UPE. Others, such as the South Centre, said unilateral measures should be allowed to apply to multinationals with UPEs in non-amount A jurisdictions.
Others see room for negotiation. The Kenyan government urged the OECD to engage in more dialogue on the issue, which it said “will affect the adoption of the 2-pillar solution.”
The Withholding Tax Showdown
The treatment of unilateral measures that are covered by treaties is far from settled — the OECD says that the standstill and rollback commitment would not include withholding taxes treated as covered taxes under tax treaties, which are excluded by article 38(2)(c). A review of the comments reveals that this will be a bitterly negotiated issue. Commentators such as the USCIB are concerned that the withholding tax exclusion will allow jurisdictions to apply withholding taxes in novel ways that discriminate against particular sectors. As such, several commentators told the OECD that significant economic presence rules and withholding taxes, including those enacted under article 12B of the U.N. model tax convention, should not be excludable under article 38(2)(c).
Some commentators, such as the Kenyan government and the BEPS Monitoring Group, also want more clarity on article 38(2)(c). In particular, the BEPS Monitoring Group said developing countries need to know whether and in what circumstances withholding taxes on payments to nonresidents would be forbidden under pillar 1.
“The scope of Amount A has now been extended to cover all services, but it is not clear whether or to what extent states participating in the MLC are expected in exchange to abjure taxation of revenues from all services through withholding taxes,” the BEPS Monitoring Group wrote. “The term ‘relevant similar measures’ is indeterminate, so it is important for the scope of the proscription in article 38 to be made as clear as possible.”
However, the OECD did indicate in footnote 10 that it will consider whether and under what circumstances the standstill and rollback treatment should apply to certain unspecified measures that are covered by tax treaties.
The SVTDG said it is encouraged that the OECD is discussing the possibility that taxes within the scope of treaties may nonetheless be destabilizing. “We encourage the [inclusive framework] to recognize that the possibility a tax may be covered by a treaty does not necessarily ameliorate the destabilizing features and effects of a unilateral measure for the international tax framework generally,” it said.
The South Centre had harsh words for the footnote, calling it an “illogical, unjustified and dangerous proposal” that could prevent jurisdictions from considering or enacting various solutions to the taxation of the digital economy. The organization is concerned that this approach could be used against article 12B of the U.N. model tax convention and could trample upon the right of treaty partners to apply article 12B in their bilateral treaties.
“There is no reason why taxes covered under existing bilateral tax treaties should be included in the scope of prohibited measures. It goes against the fundamental sovereign right of countries to decide their tax policy for themselves,” the South Centre wrote.
The Kenyan government also noted concerns with the interplay of article 38(2)(c) and footnote 10, writing that they contradict each other.
“Digital income would still fall within Article 7, 12A(UN) and 21(3)(UN). That would simply mean that paragraph 2(c) would only apply to any sort of indirect taxes,” it said. “Kenya would take an exception with the current DST law being categorized as a unilateral measure but still covered under the relevant treaty article.”
Judging by the comments submitted to the OECD, stakeholders have divergent ideas on what might be stabilizing or destabilizing in a pillar 1 world. How can there be consensus when all parties fear the results if their favored approach is not adopted? It may be helpful to revisit a central question underpinning this whole two-pillar project: What exactly is pillar 1 designed to do? But as the consultation has made apparent, this is not an easy question to answer.