Contemplating a Multilateral Convention to Implement OECD Pillars 1 and 2

Mary C. Bennett

Mary C. Bennett is senior counsel at Baker McKenzie LLP in Washington and former head of the OECD’s tax treaty, transfer pricing, and financial transactions division.

In this article, the author analyzes the need for a multilateral treaty to fully implement the pillar 1 and 2 regimes under consideration by the OECD inclusive framework and explores various design features drafters should consider.

The views expressed in this article are those of the author alone and do not necessarily represent the views of Baker McKenzie or any of its clients.

Copyright 2021 Mary C. Bennett. All rights reserved.

The OECD inclusive framework is approaching critical decisions on whether to endorse some of the most fundamental changes to international tax rules in the past century. In October 2020 it released reports on two streams of work that have occupied its attention for the past several years: the pillar 1 blueprint, proposing new nexus and profit allocation rules to address the challenges arising from the digitalization of the economy, and the pillar 2 blueprint, calling for the introduction of global minimum tax rules. The inclusive framework has said the blueprints provide a solid basis for future consensus on the two work streams, and it committed “to swiftly address the remaining issues with a view to bringing the process to a successful conclusion by mid 2021.”¹ The blueprints acknowledge that implementation of any agreement would require changes to treaties and domestic law and would benefit from the development of a multilateral treaty to cover key elements of the pillars. While much ink has been spilled in the debate over the proposed substantive rules, there is relatively little public commentary on implementation challenges, particularly those associated with developing and ensuring the entry into effect of a multilateral convention. This article represents a first stab at identifying some of the issues to be addressed in that exercise.

I. Is a Multilateral Convention Needed?

An initial question, of course, is whether a multilateral convention is in fact needed to implement pillar 1 or 2. It is best approached by considering each pillar separately.

A. Need for a Pillar 1 Multilateral Convention

The pillar 1 blueprint notes several aspects of the pillar 1 proposals that would benefit from a multilateral convention, including removing treaty barriers to the determination of the new amount A tax; eliminating double taxation for that new tax; providing for the procedure, administration, and tax certainty processes for amount A; and implementing new rules on tax certainty beyond amount A. Each presents its own considerations.

There is no doubt that the permanent establishment provisions in bilateral treaties would prevent a market country from imposing the tax on residents of its treaty partners that benefit from those provisions. While countries could conceivably amend their treaties via individual protocols to remove that prohibition, the pillar 1 blueprint correctly notes that a multilateral treaty would be the best way to “ensure coordination, consistency and certainty, and operate in a speedy manner.”

Perhaps more fundamentally, the amount A tax has features that transcend mere bilateral relationships. Its base is calculated at the multinational group level, not at the level of group members resident in particular countries. The portion of the multinational group’s residual profit that will be allocable to a particular market jurisdiction is determined by applying a formulaic allocation key to the group’s total allocable tax base. That base is apportioned among all eligible market jurisdictions² based on the relative amount of in-scope revenue derived from each eligible jurisdiction, such that the aggregate allocations to eligible jurisdictions should equal the group’s total allocable residual profit. Thus, each market jurisdiction’s claim of entitlement to a share of the multinational group’s residual profit potentially affects claims of every other market jurisdiction where the group derives in-scope revenues.

Similarly, each market jurisdiction will identify at least one individual member of the multinational group as the paying entity chargeable with tax liability for that jurisdiction’s share of the group’s amount A allocable base. The pillar 1 blueprint suggests a process of up to four steps for identifying a group’s paying entities. The jurisdictions where those paying entities are resident must provide double taxation relief for their residents’ amount A tax liability. Thus, each market jurisdiction’s claim of entitlement to a share of the group’s residual profit potentially affects obligations of every residence jurisdiction where paying entities reside, and the aggregate of the double tax relief obligations in the residence states should equal the group’s amount A tax liability in the relevant market jurisdiction.

In other words, unlike traditional taxing regimes that take into account only a single legal entity’s taxable base and its potential liability in a single source state and a single residence state, the amount A taxing regime presents unprecedented possibilities for multijurisdictional tensions between states — market-to-market, market-to-residence, and residence-to-residence states — in the calculation of a single taxpayer’s amount A tax charges in a single market jurisdiction. That underscores the need for a truly multilateral treaty to implement amount A.

The pillar 1 blueprint suggests that as between countries where no bilateral treaty is in place, the amount A rules “could, at least in theory, be implemented purely under domestic legislation.” That suggestion ignores whether a market jurisdiction’s claim to exercise taxing rights over a foreign entity that has no presence in or other direct link with it (other than being a member of a multinational group whose members derive revenue from that jurisdiction) could be a violation of customary international law. Customary international law allows a state to exercise jurisdiction to prescribe law applicable to persons, property, or conduct when “there is a genuine connection between the subject of the regulation and the state seeking to regulate.”³ As commentators have noted, merely deriving sales revenue from a jurisdiction has not, at least until now, been considered a sufficiently genuine link under customary international law for that jurisdiction to impose income tax on the seller in the absence of any sort of presence in the market jurisdiction.⁴ That lack of genuine connection seems even more acute when the entity potentially being taxed is not even itself the seller, but merely an affiliate of the seller, as could be the case under the amount A process for identifying the paying entity in a multinational group. Market jurisdictions hoping to implement pillar 1 successfully may therefore want to have their right to impose an amount A liability on nonresident members of multinational groups confirmed by a treaty with the paying entities’ residence states, even if there is no treaty between the market and residence states that would limit that taxing right.⁵ While each market jurisdiction could conceivably approach any non-treaty partner to request a treaty confirming its amount A taxing rights, a multilateral treaty is clearly a more efficient solution for that purpose.

The tax certainty features of amount A, including the dispute resolution mechanism that includes mandatory arbitration binding on all relevant states, clearly need a multilateral treaty to function. The same appears true of the administrative implementation aspects — for example, harmonized approaches to centralized return filing, information exchange, and payment procedures — which will require multijurisdictional coordination.

Inasmuch as the OECD styles amount B as “a simplified means of establishing the arm’s length remuneration to narrow baseline activities,” there is a suggestion that its implementation would not require a change to existing treaties. However, the pillar 1 blueprint also contemplates the introduction of a mandatory, binding arbitration procedure for resolving amount B disputes, which could require changing treaty relationships or introducing new arrangements if no treaty relationship exists. Depending on the interaction between amount B and any marketing and distribution profits safe harbor that might be introduced to mitigate double counting of a multinational group’s profits in a market jurisdiction, the determination of the market jurisdiction’s amount B allocation could affect the allowable amount A allocation. That could affect the double tax relief obligations under the amount A regime for paying entities in the multinational group that are not resident in the same jurisdiction as the entity that is the counterparty to the market jurisdiction’s amount B allocation. Thus, the potential multijurisdictional impact of amount B allocations — on top of the need to quickly introduce new dispute resolution mechanisms for amount B disputes — militates for a multilateral treaty to cover amount B issues.

For unilateral measures, the pillar 1 blueprint simply states that “it is expected that any consensus-based agreement must include a commitment by members of the inclusive framework to implement this agreement and at the same time to withdraw relevant unilateral actions, and not adopt such unilateral actions in the future.” That sort of “standstill and rollback” obligation would ideally be enshrined in a binding legal instrument — and a multilateral convention makes sense.⁶ Presumably, that kind of agreement would include mutual commitments: an agreement by an entity’s residence state to allow for, and give the prescribed double tax relief for, any amount A tax liability imposed by a market state on that entity; and an agreement by the market state to refrain from applying unilateral measures and to limit its taxation of the entity’s in-scope revenue to that allowed under the multilateral treaty or any other treaty between the two states. Each state would also presumably commit to adhere to the multilateral treaty’s provisions relevant to the implementation of the pillar 1 regime by the market state, including compliance with administrative procedures and dispute resolution mechanisms.

There presumably would not need to be any provision in a multilateral treaty obliging a state, qua market jurisdiction, to enact the pillar 1 regime as part of its domestic law. In other words, pillar 1 need not become a minimum standard of the type that emerged from the OECD’s base erosion and profit-shifting project,⁷ except to the extent that inclusive framework members might be required to commit to take the steps necessary as residence states to accommodate the implementation of pillar 1’s new taxing rights by other inclusive framework members, including by removing PE restrictions, granting double taxation relief, and adhering to the administrative and dispute resolution measures necessary for that purpose. A multilateral treaty seems necessary to ensure compliance with the last commitment.

B. Pillar 2

Pillar 2 involves four rules, as well as a suggestion for rule coordination, that raise different issues regarding their interaction with existing treaties and their potential need for a multilateral treaty.

The pillar 2 blueprint asserts that like controlled foreign corporation rules, the income inclusion rule (IIR) can be imposed without any incompatibility with existing treaty obligations. It says that is the case regardless of whether the relevant treaty preserves the taxpayer’s state of residence’s right to tax the taxpayer without regard to the provisions of the treaty — that is, whether or not the treaty contains a saving clause comparable to article 1(3) of the 2017 OECD model tax convention. It relies on paragraph 81 of the commentary on model article 1, which in turn relies on rationales in the commentary on articles 7 and 10. Paragraph 14 of the commentary on article 7 says CFC regimes do not run afoul of the article 7(1) limitation on the taxation by one state of the business profits of an enterprise of the other state, on the theory that tax “levied by a State on its own residents does not reduce the profits of the enterprise of the other State and may not, therefore, be said to have been levied on such profits.” Likewise, paragraph 37 of the commentary on article 10 states that CFC rules do not run afoul of the article 10(5) limitation on the ability of one state to tax the undistributed profits of a company resident in the other state because “the paragraph is confined to taxation at source and, thus, has no bearing on the taxation at residence under such legislation or rules,” and “the paragraph concerns only the taxation of the company and not that of the shareholder.”

While language asserting the compatibility of CFC regimes with the OECD model has been in the commentary since 1992, not all courts and tax administrations have agreed with that.⁸ Concerns about those lingering contrary views regarding the compatibility of CFC rules with the OECD model contributed to the decision to introduce the saving clause.⁹ Thus, despite the pillar 2 blueprint’s assertions, countries that want to introduce an IIR but do not have universal inclusion of a saving clause in their treaty networks might want to be able to point to some legal instrument to confirm that the IIR does not violate their existing treaty obligations. A new multilateral treaty may be a useful way of getting that confirmation.

For the undertaxed payment rule (UTPR), the pillar 2 blueprint asserts that its denial of deductions on other than an arm’s-length basis does not violate article 9 of existing treaties on the grounds that once an expense is allocated to a resident taxpayer or local PE under the arm’s-length principle, domestic law governs the deductibility of that expense. The blueprint also analyzes whether application of the UTPR could be constrained by existing treaty provisions based on article 24(4) of the OECD model, which requires equal treatment of payments by a resident to a resident of the treaty partner and payments between resident taxpayers. According to the blueprint, article 24(4) does not constrain use of the UTPR because the rule applies only when a payment is made from a high-tax jurisdiction to a low-tax jurisdiction. It concludes rather dubiously that there cannot be a conflict with article 24(4) because a jurisdiction cannot be both high- and low-tax for a group in a particular year, and the only relevant criterion is whether a payment flows from a high-tax payer to a low-tax payee.

The analysis ignores the fact that the UTPR regime itself would mandate jurisdictional-level blending, thereby creating a resident-versus-nonresident discrimination for the very criterion that triggers nondeductibility. In other words, if a Country A entity (A1) made payments to both a Country A affiliate (A2) and a Country B affiliate (B), and if A2 and B both had low effective tax rates on a stand-alone basis, but Country A was a high-tax country for the multinational group because of the blending of A1’s ETR with A2’s ETR, the UTPR would deny A1’s deduction for the payment to B but not for the payment to A2. Even putting aside the blending point, there could arguably be some bad faith involved if a state, having entered into a treaty containing article 24(4) with a country that was known to have a lower tax rate, subsequently points to the resulting low ETR of the other state’s residents as a basis for asserting that article 24(4) has no effect on the application of the UTPR against the other state’s residents.

The author understands that lack of confidence in the strength of the pillar 2 blueprint’s conclusion on the article 24(4) point has caused some governments to conclude they would be better off seeking a multilateral treaty’s confirmation of their ability to apply the UTPR regime to payments made to residents of treaty partners that could potentially benefit from article 24(4).

The issues described above could conceivably be addressed in a multilateral treaty along the lines of the BEPS action 15 multilateral instrument — in other words, a multilateral treaty primarily designed to modify existing bilateral treaty obligations. However, like pillar 1, pillar 2 has features that transcend purely bilateral relationships and heighten the need to consider a more truly multilateral convention for its implementation.

For example, the UTPR operates in part by causing a high-tax-country payer (the UTPR taxpayer) that makes an otherwise deductible payment to a low-tax-country affiliate (the constituent entity) to compute the amount of top-up tax allocable to the constituent entity (for example, if the constituent entity has adjusted income of $1,000 and the group’s ETR in the constituent entity’s country is 1 percent below the pillar 2 minimum tax rate, the top-up tax allocable to that constituent entity is $10).¹⁰ Assuming that there is no parent entity above the constituent entity that will be picking up that $10 of top-up tax under the IIR, that top-up tax will be allocated in the first instance under the UTPR regime to those affiliates that make direct payments to the low-tax constituent entity. If there are several high-tax affiliates (UTPR taxpayers) that have made direct payments to a particular low-tax constituent entity during the tax year, under the UTPR’s first allocation key, that entity’s top-up tax is spread among those UTPR taxpayers in proportion to the size of their direct payments to the constituent entity. Thus, each UTPR taxpayer making payments to a low-tax affiliate in the multinational group must take into account the amount of payments made to the lowtax affiliate by all other UTPR taxpayers in the group to compute the amount of top-up tax that ultimately gets allocated to it. Conceptually, each UTPR taxpayer will lose enough of its deduction in its jurisdiction to increase its local tax liability by the amount of the top-up tax allocable to it from its low-tax constituent entity payee. The potentially multilateral sharing of the top-up tax allocable from any low-tax constituent entity highlights the need for a multilateral treaty to achieve a proper assignment of taxing rights.

Other mechanics of the UTPR regime also point to the potential need for a multilateral treaty for proper implementation. For example, the second allocation key for the UTPR causes the portion of the top-up tax associated with a low-tax affiliate that has not already been allocated under the first allocation key to be spread among group members in UTPR countries in proportion to their net intragroup deductible expenditures. That clearly requires a multilateral calculation, thus making it difficult to implement through a series of bilateral treaties. The pillar 2 blueprint acknowledges that multilateral coordination is critical to ensuring the appropriate application of the UTPR.

The pillar 2 blueprint strongly implies that the result of applying the UTPR regime in a particular jurisdiction cannot be greater than an effective denial of the deductions for intragroup payments made by entities resident in that jurisdiction, although the wording of the blueprint is not clear on that point.¹¹ In concluding that the UTPR regime is compatible with existing treaty obligations (for example, the article 9 arm’s-length and article 7 PE principles), the pillar 2 blueprint relies on the suggestion that the regime can do no more than effectively deny a local UTPR taxpayer’s deductions for intragroup payments. If the UTPR regime potentially has a broader effect — by effectively giving a UTPR jurisdiction the right to tax a resident entity on an amount that is greater than that entity’s gross income — the blueprint’s treaty compatibility analysis may be incomplete. In that case, as with pillar 1, a multilateral treaty may be viewed as the most efficient way to remove any bilateral treaty constraints on the full implementation of pillar 2.

The rule coordination aspects of pillar 2 also suggest the need for a multilateral agreement. That would likely include provisions confirming that the subject-to-tax rule (STTR) applies first (assuming the inclusive framework ultimately adopts that approach),¹² the IIR has priority over the UTPR,¹³ and the IIR is applied with “top-down” priority.¹⁴ Enshrining those coordination rules in a multilateral treaty could be challenging, particularly if the rules have not been fully fleshed out in any agreement reached by the inclusive framework.

For example, how will the restrictions on the UTPR interact with existing (or future) domestic legislation that may deny deductions on related-party payments on grounds that may be similar to those underlying the UTPR but without being styled as a UTPR regime? One might think here of, for instance, the U.S. base erosion and antiabuse tax, recently proposed German legislation denying deductions for payments into tax havens,¹⁵ or article 238A of the French tax code denying deductions for payments into low-tax jurisdictions.¹⁶ If, as the Biden administration proposes, the BEAT is replaced by a pillar-2-compatible SHIELD (stopping harmful inversions and ending low-tax developments) proposal,¹⁷ the question may be less difficult for the United States, but it will remain for other regimes and will presumably require consideration of appropriate “standstill and rollback” provisions for noncompliant regimes.

Similarly, how will the top-down restrictions on the IIR interact with existing (or future) domestic laws, including CFC regimes, that tax a parent company on some or all of a subsidiary’s earnings based in whole or part on the lower tax rate applicable in the subsidiary’s jurisdiction?¹⁸ The coexistence of the IIR standards with existing and future CFC and similar regimes is likely to prove an area of tension for some time, perhaps heightening the need for a multilateral treaty to provide effective dispute resolution mechanisms for pillar 2 issues.

The United States’ desire to ensure that the GILTI regime (whether in its current form or as amended by Biden administration proposals) is recognized as a pillar-2-compliant IIR regime could also lead to interest in having other countries commit to that recognition through a multilateral treaty. That could be necessary to protect U.S.-parented multinational groups from being exposed to the application of the UTPR by source countries where the group operates and from the application of the IIR by countries where the group has intermediate subsidiaries.

The STTR effectively requires changes to existing treaty obligations, which could theoretically be accomplished by individual protocols to existing treaties or, perhaps more efficiently, through a multilateral treaty. One question yet to be addressed by the inclusive framework is whether pillar 2’s STTR will be a minimum standard, in the sense that inclusive framework members would commit to changing their existing treaties to allow for the application of the STTR and ensure that any future treaties also allow for the rule. If those commitments are to be made, a further question is whether they will be reflected in a legally binding instrument such as a multilateral treaty. That kind of treaty could conceivably encourage reluctant jurisdictions to accommodate the STTR — for example, by allowing source countries to escape the pillar 2 limitations on the UTPR for payments within groups whose parent jurisdiction has not accommodated the STTR, even if it has implemented the IIR.

The switchover rule also requires a potential change to the double taxation relief obligations under existing treaties to allow for a head-office-country top-up tax on the low-taxed earnings of foreign PEs. That could likewise be efficiently achieved by a multilateral convention.

The pillar 2 blueprint does not suggest that inclusive framework members would be required to adopt pillar 2 as a minimum standard, in the sense that they would be obligated to enact either an IIR or a UTPR. The operating assumption has appeared to be that the threat of application of the UTPR by source countries to payments within groups headquartered elsewhere would be a sufficient incentive for the headquarters jurisdictions to reclaim their priority taxing rights by enacting an IIR.¹⁹ On April 7 U.S. Treasury Secretary Janet Yellen launched a strong campaign in favor of a global minimum tax:

What is not clear from Yellen’s statement is whether she is suggesting that the IIR (and perhaps also the UTPR) should become a minimum standard for inclusive framework members, or at least for some critical mass of “major economy” members. If that is what she is suggesting, a further obvious question is whether the minimum standard would call for merely a political commitment or a legally binding one, and if the latter, whether a multilateral treaty should be the vehicle to achieve that.

C. Interaction Between Pillars 1 and 2

The foregoing sections describe the likely need for a multilateral convention to implement pillars 1 and 2. A further issue that requires consideration is the interaction between countries’ commitments stemming from the pillars.

Throughout the project on pillars 1 and 2, it has been reported that some countries were more interested in reaching agreement on one of them than the other, with that preference varying according to the countries’ particular interests. However, there have also been reports that some countries would agree to one pillar or the other only if there was agreement on their preferred pillar, or that they would agree to either pillar only if there was agreement on both.²¹ That sentiment seems to be echoed in recent statements from the Biden administration. According to a widely circulated April 8 presentation by U.S. delegates to the Inclusive Framework Steering Group, pillar 2 “cannot be fully successful absent a stable multilateral international tax architecture,” and pillar 1 “provides the opportunity to stabilize the architecture.”

If the two pillars are to be implemented as a package deal, a single multilateral convention (or, at least, two linked multilateral conventions) may be needed to achieve that. That raises many questions of potential interaction between the two pillars. For example, what incentives could be built into the treaty to encourage reluctant countries to adopt one or another feature of pillar 1 or 2? Would a country interested in ensuring elimination of unilateral measures under a pillar 1 deal reserve the right not to apply the pillar 2 top-down IIR priority approach to groups headquartered in a country that failed to remove its unilateral digital tax measure? Could a market country interested in exercising its amount A taxing rights under pillar 1 reserve the right not to respect the pillar 2 priority of the IIR over the UTPR for payments to countries that did not agree to respect that market country’s enhanced taxing rights and to give double tax relief for them when necessary?

The above analysis leads to the conclusion that a multilateral treaty is almost undoubtedly necessary to achieve a full and coherent implementation of pillars 1 and 2, but the details of how its provisions will incorporate the core elements of those proposals (for example, permissive or mandatory, free-standing or interlinked?) must be hammered out.

II. Background on Multilateral Tax Agreements

Multilateral tax conventions are not a new idea, but they have been relatively few and far between. The introduction to the OECD model notes that the Committee on Fiscal Affairs gave thought to a multilateral convention when considering both the 1963 draft convention and the 1977 model convention but concluded the idea posed too many difficulties. Despite the later conclusion of the Nordic Convention on Income and Capital of 1983 (amended in 1987, 1989, and 1996); the Convention on Mutual Administrative Assistance in Tax Matters, which entered into force on April 1, 1995; and the MLI, which opened for signature in 2016, as of 2017 the OECD committee continued to think that “there are no reasons to believe that the conclusion of a multilateral tax convention involving a large number of countries that could replace the network of current bilateral tax conventions . . . could now be considered practicable,” and that “bilateral conventions are still a more appropriate way to ensure the elimination of double taxation at the international level.”²²

The multilateral tax agreements that do exist tend to be limited to a relatively small number of countries from a single region²³ or address a relatively narrow set of substantive or procedural issues.²⁴ The first type tends to fairly closely follow the provisions of bilateral tax treaties and the OECD or U.N. model.²⁵ Neither type provides a great deal of guidance on how to design a multilateral convention to implement pillars 1 and 2, given the novel substantive content of the proposals and their complicated interaction with participating states’ domestic laws and existing and future bilateral treaties.

As the most recent multilateral initiative of broad application, the MLI probably holds the most interest. It is, however, a fundamentally different document from what one would expect a BEPS 2.0 multilateral treaty to be. Its exclusive purpose is to modify existing bilateral treaty relationships efficiently and consistently. The new multilateral treaty will need to do that for specific provisions (for example, to introduce the pillar 2 switchover rule and STTR), but will largely be operating to create wholly new taxing rights and obligations, as well as procedural and dispute resolution mechanisms, that will need to coexist with existing treaties without being limited by them. The new multilateral treaty will also need to govern relationships between countries that do not have any existing treaty relationship, which means its drafters will need to think about the ancillary issues that might otherwise be left to the bilateral treaty. While ensuring an amount of consistency in the implementation of BEPS-related treaty changes, the MLI is full of optionality, providing contracting states with a huge number of choices to make about which of their treaties to allow to be amended, which provisions to adopt or leave aside, which variants of some provisions to select, and whether to make any of various permitted reservations. The result is an extremely complex and customized set of amended bilateral relationships, and it is only through the publication of synthesized texts and the OECD’s own toolkit that most users can feel confident they understand what modifications the MLI has made to existing treaties. One can only assume that the new multilateral treaty will have to be designed with much less optionality, given the need to have an essentially global and highly consistent approach to the implementation of pillars 1 and 2.

III. Issues to Be Addressed by the Treaty

The new multilateral treaty will obviously need to address all the core elements of the pillar 1 and 2 proposals discussed above. In thinking about how it will do that in a way that builds on existing concepts from bilateral treaties, it is useful to consider how the new multilateral treaty might handle specific issues typically covered by a bilateral treaty.

A. Persons Covered

Bilateral treaties typically cover persons who are residents of one or both contracting states; multilateral treaties frequently substitute the word “more” for “both” in that formulation. The new multilateral treaty may need some combination of phrases to describe the multinational groups covered because so many of the pillars 1 and 2 rules apply at the group level, as well as to define the individual entities covered. A decision to cover multinational groups could require attention to how the multilateral treaty might affect group members resident in jurisdictions that do not become parties to the treaty. The pillar 1 and 2 blueprints also tend to treat constituent entities of multinational groups as relevant persons, and that term includes PEs under some circumstances. Consideration will need to be given to whether to include those constituent entities in the definition of covered persons, in which case care will need to be paid to the definition of persons and to the determination of residence of those constituent entities.

The “persons covered” article of modern bilateral treaties tends to include language concerning whether income derived by or through an entity or arrangement that is treated as wholly or partly fiscally transparent under the tax law of either contracting state is considered income of a resident of a contracting state.²⁶ The pillar 2 blueprint contains language potentially treating fiscally transparent entities as stateless constituent entities, but allocating all their income, expenses, and taxes to other constituent entities in the multinational group. Consideration will need to be given to how to incorporate that approach, if deemed appropriate, into treaty language aimed at fiscally transparent entities.

Article 1(3) of the OECD model contains the so-called saving clause, which states that the treaty shall not affect the taxation by a contracting state of its own residents except for some specified benefits (for example, the right to double taxation relief under article 23). The drafters of the new multilateral treaty will need to consider whether to include a comparable saving clause and, if so, which benefits to exclude.

B. Taxes Covered

Most multilateral treaties tend to follow the general approach of article 2 of the OECD model, in that they affirm coverage of each contracting state’s income taxes as generically defined, include a list of covered taxes of each contracting state, and claim coverage of “any identical or substantially similar taxes that are imposed after the date of signature of the Convention in addition to, or in place of, the existing taxes.” While that approach may be appropriate for the new multilateral treaty as well, broader coverage will also be necessary for some purposes (for example, to activate the “standstill and rollback” obligation for offending unilateral measures).

Article 2(1) of the OECD model specifies that subnational taxes are covered taxes, but countries have different policies on whether to follow that approach; for example, the United States does not cover state or local income taxes in its tax treaties.²⁷ Questions may arise in the course of preparing the new multilateral treaty whether a party should be allowed to exclude subnational taxes from the treaty’s coverage, particularly in light of the U.S. insistence on the removal of unilateral measures by other countries and the tendency of a growing number of U.S. states to enact digital services taxes or their equivalent.²⁸

The drafters might also want to consider whether the new multilateral treaty should cover regional taxes. That could become an issue, for example, if the European Union should move to impose an EU-level digital levy as part of the “own resources” it is seeking to repay its COVID-19-related borrowing.

C. Definitions

Bilateral treaties typically define several specific terms and provide that unless the context otherwise requires or the competent authorities agree to a different meaning, any undefined term will have the meaning that it has at that time under the law of the state applying the treaty.²⁹ As discussed below, the new multilateral treaty may need to include some form of provision, comparable to model article 25(3), that would establish a procedure under which the parties to the treaty could agree on a definition of an undefined term. Regarding the fallback reference to the domestic law of the state applying the treaty, consideration might also be given to an approach like that in article 3(2) of the EU Arbitration Convention, under which “any term not defined in this Convention shall, unless the context otherwise requires, have the meaning which it has under the double taxation convention between the States concerned.”³⁰ While reference to a bilateral treaty definition rather than a domestic law definition might not be appropriate for all undefined terms, it could be appropriate in particular cases.

D. Residence

For those entities that have a tax residence under the domestic laws of at least one contracting state, one would expect that tax residence to be treated as the entity’s tax residence under the new multilateral treaty. The pillar 2 blueprint suggests that any entity not having a tax residence under that standard should be treated as resident in its jurisdiction of incorporation. For entities with dual residence, the OECD model indicates that the competent authorities of the contracting states should endeavor to resolve the issue, failing which the entity is generally not entitled to any treaty relief. The pillar 2 blueprint suggests that dual residence should be resolved under the tax treaty tiebreaker between the states concerned, and that further work will need to be done to develop rules when that solution is unavailable. Different multilateral treaties have provided various approaches for solving cases of dual residence, some relying on place of effective management,³¹ others on place of incorporation,³² and others calling for the issue to be settled by mutual agreement of the contracting states concerned.³³

E. Permanent Establishment

The pillar 1 and 2 blueprints raise different questions about whether a PE definition would need to be included in the new multilateral treaty and, if so, what that definition should be. For example, the pillar 1 blueprint refers to the existence of a PE in a market jurisdiction as a potential “plus” factor to be taken into account in determining whether a multinational group engaged in a consumer-facing business should be treated as having sufficient nexus there to be subject to the amount A regime and suggests the PE definition for that purpose should be “based on the commonalities of the UN and OECD Model definitions.” While the need for a PE definition for that purpose could be mooted if the inclusive framework moves away from a focus on “plus” factors (for example, if it abandons the consumer-facing business classification in favor of the Biden administration’s April 8 recommendation to the Inclusive Framework Steering Group), the commonalities of the OECD and U.N. models would presumably lead to a highest common denominator PE definition. For example, the definition would not include a services PE, a PE based on maintaining a stock of goods from which deliveries are made, or an insurance enterprise based on the collection of premiums, even though those PEs would exist under the U.N. model. One question might be whether the new multilateral treaty should rely on a bilateral treaty PE definition if one already applies.

The pillar 2 blueprint suggests its own definition based on the definition in any applicable tax treaty in force. If there is no applicable treaty, a PE will be deemed to exist if it has a sufficient business presence in a jurisdiction that the jurisdiction taxes the income of the operations on a net basis. A PE is also deemed to exist if the residence jurisdiction of the constituent entity that owns the PE treats it as a separate taxpayer from its resident taxpayer.

Putting aside the formal definition of a PE under the new multilateral treaty, another issue that could arise focuses on the extent to which the pillar 1 regime should treat the new amount A nexus as comparable to the existence of a PE for ancillary purposes of how the treaty might operate. For example, the OECD model is full of references to PEs other than in articles 5 and 7, because the existence of a PE can have implications for items such as the sourcing of interest payments, the taxability of gains from the disposition of business property, and the taxability of employee salaries. For the most part, it would seem inappropriate to treat amount A nexus as the equivalent of a PE for all those ancillary implications. As mentioned below, however, the amount A nexus probably should be guaranteed nondiscrimination protection comparable to that afforded PEs under article 24(3).

F. International Shipping and Air Transport

If the inclusive framework approves a carveout for international transportation groups, as the blueprints suggest might happen, the multilateral treaty will need to include a definition of the international transport business. There is a fair amount of deviation, even among model treaties, in the definition of the scope of international transport income that qualifies for the exemption from source-country tax under article 8. For example, the U.N. model article 8 includes two approaches, one of which limits the availability of the exemption for shipping operations that are “more than casual” in the source state. The U.S. model treaty is more expansive in its definition than either the U.N. or OECD model and includes broader coverage of income from the rental of ships or aircraft on a bareboat basis, as well as broader coverage of container leasing income. The pillar 1 blueprint indicates that the carveout from amount A would apply to profits that would fall under article 8 of the OECD model, whether or not a bilateral tax treaty exists between the jurisdictions in question. It certainly makes sense to have a single definition, but it will be interesting to see whether the United States will be prepared to live with the OECD model version of article 8. The difference in the versions of that article is the basis for one of the rare U.S. reservations on the OECD model.

G. Nondiscrimination

As mentioned, there are at least two nondiscrimination issues relevant to the new multilateral treaty. One involves market country nondiscrimination obligations regarding their exercise of amount A taxing jurisdiction based on the new nexus standard, whatever that might ultimately be under the inclusive framework’s agreement. Article 24(3) of the OECD model protects PEs: “The taxation on a permanent establishment which an enterprise of a Contracting State has in the other Contracting State shall not be less favourably levied in that other State than the taxation levied on enterprises of that other State carrying on the same activities.”

Given amount A’s creation of a new taxing right over a share of the business profits of a nonresident enterprise under a deemed nexus standard, it would seem appropriate for the multilateral treaty to guarantee comparable treatment. That would ensure, for example, that the amount A income could not be taxed at a higher rate than the generally applicable corporate rate and could not give rise to anything like a branch profits tax liability.

Second, there is some question whether the pillar 2 UTPR conflicts with treaty provisions based on OECD model article 24(4), which reads in relevant part:

If the new multilateral treaty authorizes the UTPR and simply overrides any existing treaty nondiscrimination obligations that might prevent the rule’s application, that should be sufficient to allow it to apply.

H. MAP and Dispute Resolution

The pillar 1 blueprint extensively discusses new and innovative mechanisms for preventing and resolving the kinds of multijurisdictional disputes that could arise in applying the amount A regime. Drafting multilateral treaty provisions to implement those mechanisms will be among the most challenging aspects of the pillar 1 implementation effort and is beyond the scope of this article.

One question worth discussing, however, is whether the new multilateral treaty should include a provision comparable to model article 25(3), which provides that the relevant competent authorities should attempt to resolve by mutual agreement any difficulties or doubts regarding the interpretation or application of the model, and if so, how it should work. According to the commentary, article 25(3) confirms that the competent authorities can, for example: (i) complete or clarify any ambiguous treaty definition to obviate any difficulty; and (ii) settle any difficulties that may emerge from any new system of taxation arising when a state’s laws have been changed without impairing the balance or affecting the substance of the convention.

Multilateral treaties present a particularly pressing need for some comparable mechanism to allow authoritative interpretations to be developed after the treaties become effective, in part because of the difficulty of introducing clarifying amendments into the treaty itself. Multilateral treaties with few contracting states can include provisions identical to article 25(3),³⁴ or they might tweak the language to reflect the multilateral nature of the agreement — for example, to ensure that all parties to the treaty are able to participate in negotiating an agreement.³⁵

If a multilateral treaty is expected to have dozens, if not upward of 100, parties, a more formal mechanism may be expected. For example, article 24 of the 1988 OECD/COE Mutual Administrative Assistance Convention provided for the establishment of a coordinating body composed of party representatives with authority to: (1) monitor the implementation and development of the convention under the aegis of the OECD; (2) recommend any action likely to further the convention’s general goals; (3) act as a forum for the study of new methods and procedures to increase international tax cooperation; (4) recommend revisions to the convention when appropriate; and (5) furnish opinions on the interpretation of the convention’s provisions when requested by a party.

Similarly, MLI article 31 provides for a conference of the parties, and article 32(2) authorizes the conference to address questions of interpretation. For example, the conference recently published an opinion interpreting the entry into effect of provisions of MLI article 35(1)(a), as well as one setting out guiding principles for interpreting and implementing the MLI.

The pillar 1 blueprint refers to the likely need for guidance to support and supplement provisions in the multilateral treaty, citing as examples multilateral competent authority agreements, commentary on the multilateral convention, and guidelines for the determination and application of amount A. It adds that that guidance could be revised or updated periodically.

The drafters of the multilateral convention will undoubtedly want to include some form of model article 25(3) to allow the parties to reach agreements on the interpretation of the convention after its entry into effect. The parties must be aware, however, that those agreements are not necessarily binding on courts. As noted in the commentary:

The reluctance of some courts to follow interpretations agreed by the competent authorities under an international agreement may reflect a broader backlash against the delegation to international bodies of authority traditionally held exclusively by states.³⁷ That being said, the parties’ common agreement on the interpretation of a treaty is to be taken into account in interpreting the treaty under article 31(3) of the Vienna Convention on the Law of Treaties (VCLT). Scholars have also pointed to the important way subsequent agreements among the parties to a multilateral treaty may effectively develop new law at the margins without the formality of amending the treaty text.³⁸

I. Exchange of Information

The multilateral treaty will of course need to include exchange of information provisions to ensure the proper implementation of the pillar 1 and 2 regimes. That is particularly the case for pillar 1, because the global formulary aspect of some of the calculations means that each party is likely to have an interest in information originating in a large number of other parties, and each multinational group will be filing with a lead tax administration returns that are to be shared with other tax administrations where the group has operations or derives revenue.

That may pose policy challenges for a country like the United States, which typically rigorously examines the procedures and confidentiality standards of any prospective exchange of information partner before agreeing to transmit taxpayer information. That has meant, for example, that the United States was one of very few inclusive framework countries to decline to participate in the OECD’s Multilateral Competent Authority Agreement for the Exchange of Country-by-Country Reports and that it has failed to ratify the 2010 protocol to the 2008 OECD/COE Multilateral Convention on Mutual Administrative Assistance in Tax Matters. Whether the United States will be sufficiently confident with its potential partners under the new multilateral treaty to agree to exchange information with them remains to be seen.

J. Assistance in Collection

One might also expect that the new multilateral treaty would need to include a strong assistance-in-collection provision, particularly because pillar 1 grants amount A taxing rights to market jurisdictions for multinational group paying entities that may have no presence in those market jurisdictions. Here, too, though, U.S. policy raises doubts about the country’s willingness to participate. The United States entered a reservation on the assistance-in-collection provisions of the 1988 OECD/COE Mutual Administrative Assistance Convention and has agreed to only a few of those kinds of provisions in its bilateral treaties.³⁹ As a country likely to be a lead tax administration that must accept centralized return filings (and perhaps payments) from its multinational groups with amount A liabilities abroad, the United States may be hard-pressed to refuse to accept assistance-in-collection obligations, at least insofar as they involve taxes due under the pillar 1 regime.

K. Limitation on Benefits

Action 6 of the OECD’s BEPS project established a minimum standard requiring participating countries to include anti-treaty-shopping provisions in their treaties in the form of a limitation on benefits article, a principal purpose test provision, or a combination of the two. The principal purpose test has been the preference by far among the countries that have signed on to the MLI, even though the United States firmly opposes it. The inclusive framework will presumably need to decide whether to include an anti-treaty-shopping provision in the new multilateral treaty and, if so, how it should operate. One could argue, however, that an anti-treaty-shopping provision is redundant in a multilateral treaty that is intended to be signed by, and introduce a consistent set of rules for, a large number of participating parties, and is likely to expand countries’ taxing rights, rather than provide relief from taxation. If the inclusive framework insists on including an anti-treaty-shopping provision, the selection of the principal purpose test would be a policy challenge for the United States.

L. Entry Into Force

The pillar 1 blueprint says the implementation phase of the new regime will include work on the different challenges that could arise in developing the multilateral treaty, including “to ensure that parallel and conflicting rules for the taxation of in-scope multinational groups do not arise and that the multilateral convention starts to take effect only once a critical mass of jurisdictions have fully implemented it.” The blueprint further provides that the work will include clarifying the meaning of “critical mass.”

Multilateral treaties generally do not enter into force until a minimum number of signatories have completed their ratification procedures. There are at least a few problems that could affect the selection of the threshold number for pillar 1 purposes. One involves how amount A liability will be allocated within a multinational group during the — almost inevitable — transition period when the countries of some group members have joined the multilateral treaty and others have not.

Suppose Country A (home of Parent A) and Country B (market jurisdiction) have joined the multilateral treaty but Country C (home of Sub C) has not. Suppose Country B determines that it is entitled to an amount A liability for the group of $100, of which $75 would be allocable to Parent A and $25 would be allocable to Sub C as paying entities. Will Country B collect only $75 from the group until Country C has joined the multilateral treaty and thus agreed to Country B’s new taxing right? Will Country B seek to reassign liability for the $25 to Parent A and collect the full $100 from it? What will Country A’s obligation to give double taxation relief be during the interim period — that is, must it exempt only $75 of Parent A’s income or the full $100?

The pillar 1 blueprint is silent on how the rules would apply during the transition period, but the answer could very much affect Country A’s views on how to determine the appropriate critical mass of countries that would have to implement the multilateral treaty before its provisions would start to impose obligations on Country A and its residents.

Likewise, countries that are being asked to withdraw their unilateral measures on implementation of an agreement will want to know what the rules will be for the timing of their withdrawal. For example, if they ratify the multilateral treaty, must they withdraw their unilateral measure across the board on the treaty’s entry into force? Will they be allowed to continue to apply their unilateral measure to companies resident in jurisdictions that have not yet ratified the multilateral treaty? The answer to that question will also affect what countries with unilateral measures view as the appropriate critical mass needed to effectuate the multilateral treaty.

For pillar 2, will countries (particularly the headquarters jurisdictions of major multinational groups) be inclined to accept a multilateral treaty that exposes its residents to a global minimum tax only once they are confident that their competitor countries will simultaneously do the same? Would that suggest that the relevant critical mass should be determined by reference to a particular class of countries?⁴⁰

All those questions will presumably generate political debate in the inclusive framework.

M. Termination and Amendment

Multilateral treaties typically allow individual parties to withdraw at will, although frequently only after the agreement has been in effect for a minimum period and only after a notice period. VCLT article 55 provides that “unless the treaty otherwise provides, a multilateral treaty does not terminate by reason only of the fact that the number of the parties falls below the number necessary for its entry into force.”

VCLT article 40 provides rules for amending a multilateral treaty, unless the treaty provides otherwise, including:

• all contracting states must be notified of any proposal to amend and have the right to take part in the decision on the action to be taken on the proposal, as well as in the negotiation and conclusion of any agreement to amend;

• the amending agreement does not bind any party to the treaty that does not become a party to the amending agreement; and

• as between a state party to both an earlier and later treaty and a state party to only the earlier treaty, the earlier treaty governs the parties’ mutual rights and obligations.

Those rules demonstrate the hurdles that must be overcome to update a multilateral treaty in a way that will be effective for all parties, which heightens the importance of having an effective mechanism for ongoing treaty interpretation.

IV. Reservations and Understandings

A critical part of designing a new multilateral treaty will be deciding what reservations the parties can make. As noted, the MLI authorized and prohibited various reservations the parties could make, which allowed for a highly customized approach, reflecting the MLI’s nature as essentially a vehicle for modifying hundreds of individual bilateral treaties.

The new treaty would be a very different instrument and would have to reflect the truly multilateral nature being sought and the need for essentially global, standardized rules for the BEPS 2.0 regimes. VCLT article 19 provides the basic rules for treaty reservations:

One would expect the new multilateral treaty to specify the reservations that can be made (presumably far fewer than the MLI) and will prohibit any others.

Commentators have noted the increasing tendency of the U.S. Senate to attach conditions in the form of reservations, understandings, or declarations to multilateral treaties.⁴¹ Understandings include “interpretive statements that clarify or elaborate, rather than change, the provisions of an agreement and that are deemed to be consistent with the obligations imposed by the agreement,” and the Senate may use declarations as “statements of purpose, policy, or position related to matters raised in a treaty in question but not altering or limiting any of its provisions.”⁴²

Limiting the ability to enter reservations obviously creates tension for countries contemplating joining a treaty, particularly those, like the United States, whose ratification procedures require the consent of a legislative body that has not been involved in the negotiations and might not share the executive branch’s policy perspectives. In those circumstances, attaching conditions may be an attractive option for senators hoping to influence the effect of the treaty. However, those conditions create uncertainties about the exact nature of the legal obligations undertaken by the United States.

In recent testimony before the Senate Finance Committee, former Treasury Assistant Secretary for Tax Policy Pamela F. Olson suggested that Congress might want to consider using a mechanism to guide the pillar 1 and 2 negotiations such as the Trade Promotion Authority, under which the executive branch’s trade negotiations must follow guidelines and objectives set by Congress. That intriguing suggestion seems unlikely to be taken up in the near term, which means there will likely need to be unprecedented coordination between Treasury negotiators and lawmakers if the United States is going to be able to reach agreement on a new multilateral treaty — especially one that revolutionizes the international tax system — that will survive the Senate advice and consent process.

V. Conclusion

The issues discussed in this article, which merely scratch the surface, reveal that the drafters of the new multilateral treaty to implement pillars 1 and 2 have a herculean task in front of them. The weight of the task certainly influences expectations about the timing of an ultimate outcome. One recalls that the MLI — which simply incorporated previously agreed, BEPS-related treaty changes into a legal instrument — was not opened for signature until more than a year after the BEPS final reports were released. The time needed to draft and reach agreement on the BEPS 2.0 multilateral agreement, which must address many more concerns and develop substantive language largely from scratch, could reasonably be expected to be a multiple of that. Thus, interested parties likely have a few more years to ponder the anticipated intricacies of such an innovative agreement.