Anti-Abuse Rules in International Tax Law and their Interactions

1. Introduction

Over the past few decades, geographic boundaries for companies have faded. Building complex organizations and supply chains by integrating technologies, data, and sourcing from multiple countries have become a common practice. Digitalization, globalization, and increased mobility have allowed businesses to quickly scale up and enter new markets. It has also led to the centralization and automation of many functions and processes reshaping the local footprint of multinationals.

S. 172These developments have exposed vulnerabilities in the international tax framework built upon a “brick-and-mortar” principle. Aside from genuine business-driven planning, some exploited the gaps to avoid taxation of business profits in the source states by deliberately staying below the permanent establishment (PE) threshold using artificial arrangements. In response to this, the Organisation for Economic Co-operation and Development (OECD) introduced the Base Erosion and Profit Shifting Project (BEPS) Action 7 targeting the artificial avoidance of a PE status. This action called for a review of the definition of a PE included in the tax treaties to prevent the use of various abusive tax strategies.

Nevertheless, while being a stand-alone PE-focused initiative, Action 7 is closely connected with other anti-abuse rules - general anti-avoidance rules (GAARs) introduced in tax treaties as well as domestic GAARs and specific anti-avoidance rules (SAARs). Each of these measures has its purpose in the context of broader anti-avoidance efforts. Yet, the relationship between them and Action 7 as well as potential overlap, deserves a separate analysis which is the primary goal of this work. By examining how these anti-avoidance rules are implemented and how they co-exist, the work seeks to provide a comprehensive understanding of their joint impact on preventing the artificial avoidance of a PE status.

2. PE threshold under OECD and UN Model Conventions and its role in preventing abuse

The notion of a PE is one of the cornerstones in treaty-based international tax law. Tax treaties include distributive rules for allocating taxing rights between states in respect of various income streams. These distributive rules - Article 7 of the OECD and the UN MCs as well as most of the current tax treaties - deal with allocation of taxing rights over business profits. The primary rule, as stated in this article, provides that the business profits earned by an enterprise are taxable in the residence state except insofar as the enterprise has a PE in the source state. In such circumstances, those business profits attributable to the PE are primarily taxable in the source state. The concept behind the use of a PE as a threshold for allocating taxing rights to the source state is that the presence of a PE amounts to sufficient participation of a foreign enterprise in the economic life of the source state to justify taxation there.

S. 173While there is near-unanimous agreement among countries that the existence of a PE constitutes sufficient taxable presence, the approach to conditions for a PE and the interpretation of its definition can vary depending on the model (OECD/UN) and fiscal objectives that a specific country follows. This can potentially influence choices in anti-abuse measures different countries make when trying to combat PE avoidance and thus needs to be outlined.

The dominant definition of a PE in international taxation follows the one reflected in Article 5 of the OECD MC. However, it is often criticized for primarily reflecting the interests of developed, capital-exporting countries by prioritizing residence taxation over source taxation. Having recognized that, in an effort to accommodate broader taxing rights at source for developing countries, the UN MC often interprets the definition of a fixed place of business more broadly with a lower threshold for what constitutes permanence and regularity.

For example, according to the UN MC, a building site, construction, assembly, or installation project constitutes a PE if it lasts more than six months compared to 12 months in the OECD MC. Moreover, under the UN MC provision of services, having employees or other personnel even without a fixed place of business could lead to a PE. Also, according to Article 5(5)(b) of the UN MC, maintenance of stock by the dependent agent can lead to a PE even when such an agent does not conclude contracts for the principal.

Hence, by capturing a broader range of activities and definitions that are more deliberate, the UN MC leaves less room for ambiguous interpretations. It can be assumed, therefore, that countries following the UN MC are slightly better equipped against some PE avoidance techniques compared to those following the OECD MC. In the next sections of this work, it will be investigated which additional measures are available to protect the taxing rights of states from avoidance of a PE primarily based on the OECD MC and how different anti-abuse measures interplay with each other.

3. BEPS Action 7. SAARs revising the PE threshold

3.1. Introduction

As already mentioned in the introduction, BEPS Action 7 called for enhancement of the PE definition to prevent various planning strategies aiming at avoidance of the PE status. This targeted focus is a hallmark of the SAARs designed to address specific harmful practices. It has been organized around weaknesses of different S. 174paragraphs of Article 5 of the OECD MC that required amendments further discussed in this section.

The implementation of Action 7 required a change to the OECD MC and bilateral tax treaties. To do so, the MLI has been chosen to translate the results of the proposed SAARs from call to action, giving them legal force.

3.2. Avoidance through commissionaire arrangements

Under the definition of a PE in the previous version of the OECD MC, foreign entities selling their products in domestic markets were able to avoid the agency PE thanks to commissionaire arrangements. These arrangements involved a local agent selling products in its name, but the risk and account remained with the non-resident related entity (principal) owning such products. In this way, the commissionaire was able to place the products on the local market while not being entitled to the goods sold. As a result, the commissionaire did not make any profits from sales and, hence, such profits were not taxed in the state of its business activities. Instead, only the principal fee received by the commissioner for its services was taxable.

This structure was possible because the definition that existed in the previous version of the OECD MC explicitly required the agent for being qualified as a PE to have and habitually exercise authority to conclude contracts in the name of the enterprise. Hence, it has been proposed to amend the definition of the agent to remove this vulnerability.

Article 12 of the MLI proposed moving away from the authority to legally bind criterion. Instead, a new condition provides that the person acting on behalf of a foreign enterprise in the jurisdiction will constitute a PE when habitually concluding contracts or habitually playing the principal role leading to the conclusion of contracts that are routinely concluded without material modification by the enterprise, and these contracts are:

S. 175In addition to above, Article 12 introduces stricter criteria for qualifying as an “independent agent”. The previous version of the OECD MC stated that a company would not have a PE merely by operating through a broker, general commission agent, or other independent agent as long as the agent acted in the ordinary course of business. The amendment proposed in the MLI removes specific references to brokers and commission agents to avoid confusion across civil and common law systems focusing solely on whether the agent operates in the ordinary course of business (e.g., performs activities that are related to that agency business). Moreover, the new version imposes that the number of principals is one of the criteria when assessing the independence of the local agent. Article 12(2) of the MLI provides that a person acting exclusively or almost exclusively on behalf of one or more enterprises to which it is closely related will not be considered an independent agent concerning any such enterprise.

3.3. Specific activity exemptions

Both the general PE and the dependent agent PE clauses are conditioned by the exclusion for preparatory and auxiliary activities including, for example, warehousing facilities, facilities where products are displayed or delivered, and facilities maintained for holding goods or merchandising or collecting information. The rationale for the exclusion is that such activities are not substantial enough to raise a taxable presence in the source state. However, the Final Report on Action 7 notes that these types of activities nowadays often constitute core activities, particularly in the field of e-commerce. The proposed amendment stated in Article 13 of the MLI aims to address this issue. It is proposed to tackle it in two possible ways.

The first option (Option A) suggests modifying Article 5(4) of the OECD MC by requiring listed activities such as warehousing, storage, delivery, and collecting information to qualify for an exemption only if they are truly of a preparatory or auxiliary character. In other words, there are no more automatic PE exemptions merely because the activity is in the list regardless of its economic significance.

The second option (Option B) proposes maintaining the automatic exemption but addressing BEPS issues with the anti-fragmentation rule instead. According to this rule, the specific activity exemption is not available to a fixed place of business of an enterprise if the same or a closely related enterprise carries on business activities at either the same place or at another place in the state if:

S. 176In either case, to apply this rule, the activities carried out by the (two or more) enterprise(s) must be complementary and form part of a single cohesive business operation.

3.4. Contract splitting and fragmentation

According to the pre-2017 PE definition in the OECD MC, non-residents involved in building, construction or installation projects were able to avoid the construction PE by artificially splitting up contracts into several parts to circumvent the 12-month threshold. Article 14 of the MLI addresses this issue by stating that activities performed over periods of more than 30 days should be accumulated to determine whether the 12-month period after which they constitute a PE has been exceeded. Activities performed by closely related enterprises should also be aggregated for this purpose similar to the exceptions mentioned in section 3.3 under the anti-fragmentation rule.

Based on a proposed SAAR, the assessment of whether a PE exists requires not only an analysis of the activities of a specific non-resident but also activities conducted by its related parties. Therefore, it has also been proposed to clarify the circumstances in which a person is closely related for the purposes of the PE definition. Such clarification was reflected in Article 15 and effectively meant that, in the case of adoption of any of the Articles 12-14, incorporation of this one becomes mandatory.

3.5. Overview of implementation

As noted in the previous sections, the implementation of the suggested changes to the PE clause was proposed to be done through the MLI. The uniqueness of this instrument lies in enabling countries to quickly modify their tax treaty networks by signing and ratifying a single multilateral convention rather than renegotiating each treaty individually. At the time of this analysis, there were 102 signatories (countries) to the MLI which applied to over 1,850 matched agreements. The downside of such a high number of signatories is that the MLI had to have a high degree of flexibility to keep its attractiveness. As a result of this, the instrument provides a broad list of options that countries can use to accommodate their preferences in treaty policy. Jurisdictions can choose which tax treaties would be covered by the MLI, select specific provisions, make reservations on others (with the exception of the minimum standards), and select between alternatives. When it S. 177comes to Action 7, none of the proposed Articles 12-15 were included in the minimum standard. Hence, it is sensible to assess how many countries actually opted for these SAARs.

For the MLI to effectively modify a tax treaty, the following four conditions must be simultaneously fulfilled: (1) both contracting jurisdictions must sign and ratify the MLI; (2) both contracting jurisdictions must define the treaty covered as covered tax agreements; (3) both contracting jurisdictions must refrain from making a reservation on the relevant article (if applicable); and (4) both contracting jurisdictions must choose an equal (or at least compatible) normative option among those allowed by the MLI (“matching position”).

Based on the information on the OECD website, out of the 102 countries that signed up for the MLI, only 53 (52 %) opted for the new PE definition of the dependent agency PE (Article 12). When analysing the countries that refrained from this article, many of them belong to the so called capital exporting countries. On the other hand, many developing countries that are often source or market jurisdictions opted for this article. This is self-explanatory considering that the lower threshold provides these states with additional tools to claim taxing rights over business profits. However, the fact that, for the treaties to be amended, a “matching position” is required means that it will be difficult for these source states to succeed without resident states on-board.

As of Article 13, in the case of countries opting for the new approach on the PE exemption, they had to choose between Option A and B as described in section 3.3. When assessing the acceptance rate of this article, it should first be looked at how many countries rejected it as a whole which is 35 % (36 out of 102 jurisdictions). Many of these countries are often labelled as conduit jurisdictions. Out of those countries that opted for the adjustment, the majority went for Option A, i.e., the exception applies only if the activities are of a truly preparatory or ancillary nature. At the same time, the remaining seven states selected Option B.

When it comes to measures addressing avoidance by fragmentation addressed in Article 14, the majority of countries reserved the right not to apply it. Like with Article 12, out of the 40 (39 %) countries that opted for this SAAR, many more S. 178often act as source rather than residence states. Among these countries were also a few for which offshore activities are particularly relevant like Azerbaijan, the Netherlands, and Norway. These states reserved the right for Article 14 not to apply concerning provisions of their covered tax agreements relating to the exploration for or exploitation of natural resources.

Finally, the implementation rate of Article 15 is highly dependent on the implementation of Articles 12-14 to which it relates. In line with this, countries that reserved their right for respective articles also filed a reservation not to apply this one. At the same time, 64 (63 %) of all countries amended the definition of a related party for the purposes of the PE definition as per the proposal in the MLI.

Overall, the quantitative analysis of implementation by different countries may suggest a limited success/acceptance rate of the PE anti-avoidance offered by Action 7. However, this conclusion would be premature. First of all, aside from the MLI, countries can also agree on the provisions in bilateral (re-)negotiations outside the ambit of the MLI. Moreover, it should also be assessed whether and if artificial avoidance of a PE is addressed by alternative measures. It could well be that countries that reserved their right for some or all of the above articles still commit to the principles of Action 7 through anti-abuse rules implemented in their domestic legislation or GAARs in bilateral treaties.

4. Domestic SAAR addressing PE avoidance -Australian example

4.1. Introduction

Unlike the treaty SAARs described in the previous sections, domestic SAARs are derived from the provisions of domestic tax law and can be introduced unilaterally. These rules reflect the specific anti-avoidance measures of the country enacting them and do not require international agreements to be applied. This section is dedicated to the Australian example of such a SAAR and will outline its role in addressing issues similar to those addressed by Action 7.

4.2. Australian MAAL and its role in preventing PE avoidance

Although the Australian Government has clearly stated that their country should adopt a leading role in international anti-base erosion and profit-shifting efforts, it has been considered that immediate, unilateral action is necessary to strengthen its law against tax avoidance. Consequently, the government has introduced unilatS. 179eral measures to counter international tax avoidance in its domestic tax law, including the Multinational Anti-Avoidance Law (MAAL). With this rule, the Australian Government amended the general anti-avoidance provisions in Part IVA of their Income Tax Assessment Act 1936 to prevent foreign corporations from using complex, contrived, and artificial schemes that enable them to (i) have substantial sales activities in Australia but (ii) book revenue offshore and pay little or no tax in Australia.

The aim of these new provisions was twofold: (1) to overcome arrangements limiting income attributable to an entity’s PE in Australia and (2) to widen the ‘dominant purpose’ test in relation to tax avoidance and include foreign tax within the scope of arrangements that are subject to the provisions. In order to achieve this, the foreign entity falling under the MAAL is taxed as if it had supplied goods or provided services through the Australian PE. This means that it is subject to Australian tax on notional profits attributable to the deemed PE as well as withholding taxes where royalty or interest expenses are attributable to the deemed PE. To fall within the scope of the MAAL, the following conditions should be simultaneously met:

• a foreign entity makes a supply to an Australian customer of the foreign entity;

• activities are undertaken in Australia directly in connection with the supply;

• some or all of those activities are undertaken by an Australian entity that is an associate of - or is commercially dependent on - the foreign entity;

• the foreign entity derives income from the supply;

• some or all of that income is not attributable to an Australian PE of the foreign entity;

• it would be concluded - having regard to certain specific matters - that one or more persons entered into the scheme for a principal purpose of (or for more than one principal purpose that includes a purpose of) enabling a taxpayer to obtain a tax benefit or both to obtain a tax benefit and to reduce a taxpayer’s liability to foreign tax; and

• the foreign entity is a “significant global entity”.

As is evident from the above list, the MAAL has a rather limited scope and specifically targets significant global entities, i.e., those with group income exceeding AUD 1 billion per year. Moreover, the Australian party must be associated with or commercially dependent on the foreign entity. However, the term “commerS. 180cially dependent” is not precisely defined for the MAAL purposes. The Explanatory Memorandum to the MAAL states that this term has a broader meaning than legal dependency and should be determined on the basis of reality and actual situations.

Another crucial component of the MAAL is that, unlike some other anti-avoidance rules that require tax avoidance to be the primary or dominant purpose of an arrangement, the MAAL has a lower threshold. It is sufficient if obtaining a tax benefit is just one of the principal purposes of the arrangement. This broader scope makes it easier for the Australian Taxation Office (ATO) to apply the MAAL to arrangements that might not be purely tax driven.

Once a notional PE is deemed to exist, the taxable income should be determined. To do so, the actual role of the deemed PE must first be assessed considering all relevant facts and circumstances. The profit attributable to it is then calculated using the arm’s length principle, ensuring that only the appropriate amount of income excluding any exempt income under Australian law is taxed. Deductible expenses, such as royalties and interest, are considered in calculating taxable profits, but this may lead to withholding tax liabilities that increase revenue for Australia.

4.3. Compatibility and interaction with Action 7

In the Explanatory Memorandum to the MAAL, it is stated that the concept of a notional PE is consistent with the Action 7 because it operates as a general safeguard. However, the MAAL does not cover all avoidance techniques addressed by Action 7. Specifically, it does not directly address the artificial splitting of contracts which is the focus of MLI Article 14 because of its narrow scope and focus on supply. Moreover, the MAAL does not include provisions related to the redefinition of the auxiliary and preparatory activities exemption for all entities which is another key area of concern in Action 7. As for the latter, however, it should be noted that Australia has taken other unilateral measures outside of the MAAL to address these concerns. For instance, Australia's approach to taxing digital services and low value imported goods through the goods and services tax (GST) regime, often referred to as the “Netflix Tax” and “Amazon Tax”, indirectly targets activities that might otherwise be classified as auxiliary or preparatory. Although not directly changing the PE classification, the “Netflix Tax” S. 181and “Amazon Tax” rules ensure that foreign companies engaging in substantial economic activities within Australia are required to collect and remit local GST. This gives a taxing right to Australia even though the foreign entity has no physical activity other than, for example, delivery and storage and so, otherwise, no taxable presence in Australia.

The measure when there are similarities between Action 7 and the MAAL is related to the redesigned dependent agent provision. Both initiatives revise the concept of independent agents to address scenarios where agents that seem independent are, in fact, closely connected with the foreign enterprise they represent. As discussed earlier, Action 7 introduces the phrase “acting exclusively or almost exclusively on behalf of one or more enterprises to which it is closely related” which plays a crucial role in determining whether an agent can still be qualified as independent. The same “acting exclusively” condition is contained in the MAAL. If a legally independent entity acts almost exclusively for the foreign entity, it is considered to be commercially dependent, and the MAAL applies to the arrangements in question.

However, while there is alignment in how both Action 7 and the MAAL address dependent agents, the approaches and their implications differ in scope. Action 7 broadly amends the concept of the agency PE and offers its new definition in bilateral tax treaties through the MLI. In contrast, the MAAL does not redefine the dependent agent PE in domestic tax law but instead introduces the concept of the notional PE that is applicable only in specific circumstances and to a limited scope of companies, i.e., significant global entities.

At the same time, the MAAL allows a broader interpretation of what qualifies as a dependent agent. The ATO’s examples of the MAAL applications highlight the critical role of the Australian entity in concluding sales contracts when determining the principal purpose of obtaining a tax benefit. However, while Action 7 focuses on the habitual conclusion of contracts or playing the principal role in doing so in the source state, the MAAL takes a more expansive approach. It considers the specific functions, assets, and risks assumed by the associated or commercially dependent Australian person indicating its principal role thereby resulting in the conclusion of a contract. However, the concept of the notional PE does not provide for any auxiliary and preparatory activity exemption and does not literally state the habitually acting power of the dependent agent. Consequently, this results in a broader concept of a dependent agent regarding the notional PE.

S. 182Potentially because of this, the relationship between the MAAL and Action 7 reflects a deliberate choice by Australia to rely on its domestic SAAR and bilateral negotiations rather than adopting Article 12 of the MLI. By doing so, it allows Australia to address specific forms of tax avoidance domestically in a more flexible way. This is especially relevant because the notional PE is not defined in tax treaties, and a domestic MAAL could supersede them. As a result, this creates a high level of legal uncertainties for MNEs falling within the scope of these rules.

5. GAARs as a tool to prevent artificial avoidance of PE

5.1. Introduction

This chapter will examine the role of different GAARs in preventing the artificial avoidance of a PE. GAARs in tax treaties and domestic GAARs will be analysed. It will also examine experiences of selected countries to analyse if and how they address the avoidance of a PE with GAARs.

5.2. GAARs general overview

Unlike SAARs that target specific types of tax avoidance techniques with detailed provisions, GAARs are deliberately broad. This generality allows them to address a wide range of tax avoidance structures making them a dynamic tool in the hands of tax authorities. Although GAARs significantly vary in form and language, there are some similarities between all of them.

GAARs generally require three primary elements: (i) the taxpayer’s purpose or intent (subjective element); (ii) a tax scheme, arrangement, or transaction; and (iii) deriving a tax benefit or advantage contrary to the object and purpose of the law (objective element).

The common ground is that motives or purposes pertain to the reasons behind or desired outcomes for carrying out the arrangement. Since only humans can possess such purposes, these tests are often associated with subjectivity by probing into the specific taxpayer’s thoughts. Nonetheless, it is feasible to make this assessment more objective, for instance, by evaluating the taxpayer’s purposes based on external circumstances surrounding the transaction and downplaying S. 183the taxpayer’s declared motives and/or by considering what an objectively rational actor would have contemplated in the same scenario.

The subjective element connects the other two main components of the GAARs, specifically the scheme, arrangement, or transaction and the deriving tax benefit/advantage. In general, GAARs apply to schemes, transactions, arrangements, acts, or courses of action that provide a tax benefit in situations not in accordance with the object and purpose of the applicable law. For this reason, they are described as “unacceptable”, “impermissible”, “illegitimate”, “aggressive”, or “unjustified”. Their tax avoidance nature is recognized by means of tests. Although the formulation of these tests can vary, they typically involve assessing whether an arrangement lacks substance or non-tax effects by either making minimal changes to the original situation or by replicating the substance of another form or transaction that would not have resulted in the tax benefit.

Based on the above features, it is evident that, although being broad, GAARs can also be deployed to address abusive PE avoidance structures. Therefore, in the following sections, domestic and treaty based GAARs will be further investigated in light of prevention of PE avoidance.

5.3. Domestic GAARs and their role in preventing avoidance of a PE

5.3.1. General Overview

Although GAARs are relatively new in the tax laws of some countries, they have been a fundamental component of other tax systems for more than a century. Domestic GAARs can broadly be categorized into two types: judicial GAARs and statutory GAARs, although it is challenging to establish superiority in their effectiveness.

Judicial GAARs are developed and applied through case law rather than being explicitly codified in the tax legislation. In jurisdictions where judicial GAARs prevail, the courts play a major role in interpreting and applying them based on established legal doctrines. These include, among others, the “substance-over-form” and “business purpose” doctrines that can be particularly relevant in the context of PE avoidance. The “substance-over-form” doctrine aims to tax transactions based on their economic reality rather than legal structure. This is particularly favoured in common law countries, though the interpretation of “subS. 184stance” can differ significantly between jurisdictions. Similarly, the “business purpose” doctrine disregards transactions for tax purposes if their primary objective is tax avoidance or reduction, asserting that a transaction is recognized for tax purposes only if it was undertaken with legitimate business intentions.

Statutory domestic GAARs, on the other hand, are explicitly codified in a country’s tax legislation. Often, judicial anti-avoidance doctrines like “substance-over-form” are developed first and, later, they are formalized into statutory GAARs to provide greater legal certainty.

The usual legal consequence of applying GAARs is the denial of the tax benefit and recharacterization of the transaction as if the abuse never occurred with GAARs generally applying only for tax purposes.

5.3.2. Domestic examples

5.3.2.1. Spain

Statutory GAARs are part of a long-lasting tradition in Spain. They originated in the 1963 General Tax Code that introduced the anti-avoidance provision that was rarely applied due to its subjective nature and the availability of simpler alternatives. In 2003, due to major tax reform, a new General Tax Code was introduced establishing the current GAAR under Article 15. This provision addresses situations when a taxpayer avoids tax obligations through clearly artificial or improper arrangements that have no genuine economic purpose beyond achieving tax savings. The rule applies to all taxation areas, including direct and indirect taxes. It covers purely domestic situations as well as cross-border scenarios.

Despite the broad scope and intent of the statutory GAAR, when it comes to avoidance of a PE, Spanish tax authorities and courts have developed an additional measure that can be considered a form of Spanish judicial anti-avoidance rule. The origins of this measure can be found, first, in two rulings of the Spanish DirecS. 185torate General for Taxation (DGT). One conclusion was that, when a foreign entity structures its operations in Spain through subsidiaries, such as converting a distributor into a commissionaire and a manufacturer into a contract manufacturer, these arrangements may still constitute a PE if they collectively form a “complex operative settlement”. This happens if a detailed analysis shows that the commissionaire and contract manufacturer assume functions or risks beyond their contractual terms whether independently or as part of the broader group and even involving third-party services. The DGT supported this interpretation by referencing paragraph 27.1 of the Commentaries on Article 5 of the 2003 OECD MC.

Relying on the DGT's interpretation, Spanish courts have used the concept of the “complex operative settlement” as a judicial tool to look beyond the formal legal structures of business arrangements and assess the economic substance of the activities conducted within Spain. This approach is evident in the following landmark cases.

The first one is the Borax case where the Spanish subsidiary of a UK company was transformed from a full-fledged distributor with a broad scope of business activities into a contract manufacturer and service provider. The products (minerals) were imported by a parent company, warehoused and processed by the subsidiary, and then distributed to clients whereas the subsidiary also acted as a “sales channel”. The Spanish tax authorities argued that there is a significant overlap between the activities of the parent company and its Spanish subsidiary making it difficult to distinguish between the two and concluded that this highly integrated activity constituted a “fixed place of business” under Article 5(1) of the Spain-UK Tax Treaty. The National Court supported this position. When interpreting Article 5(1) of the tax treaty, it extended a concept of a “fixed place of business” to include “complex operative settlements”: the parent company subcontracted the subsidiary to carry out its activity that was not preparatory or auxiliary, and the economic cycle was closed in Spain, i.e. production, sales, and profit generation all occurred there. The Spanish Supreme Court largely agreed with the National Court but added further reasoning. It determined that the contract between the parties was not just a warehousing agreement but involved significant transformation from raw materials into products. As a result, according to the Spanish Supreme Court, it did not qualify for the exemptions provided under Article 5 of the Spain-UK Tax Treaty. The Supreme Court also concluded that the Spanish subsidiary was a dependent agent because the service agreement excluded any possibility of independence - there must be adherence with the terms set by the parent. Finally, the Supreme Court recognized that the activities carried out in S. 186Spain remained unchanged after the restructuring in 1996 with the only difference being the legal ownership of the minerals that shifted from a local subsidiary to a UK parent.

Another piece of evidence of the “complex operative settlements” doctrine application is the Roche case. The case involved Roche, a Swiss multinational pharmaceutical company, and its Spanish subsidiary. Similar to the Borax case, prior to reorganization, the latter was a full-fledged manufacturer and distributor while, after its local entity became a contract manufacturer and a commissionaire based on respective contracts with its parent company. The position of the Spanish tax authorities, in this case, was very similar to the one they took in the previous one - there was no auxiliary or preparatory activity, and a fixed place is present in Spain through which the main activity is performed. However, a key distinction in this case was that the local subsidiary was also considered a dependent agent PE of its parent company. According to the tax authorities, both criteria under Article 5 of the Spain-Switzerland Tax Treaty for establishing a PE were satisfied. The National Court accepted the tax authorities' arguments. When interpreting the dependent agent clause, it established that a dependent agent PE could exist not only when a person has the authority to conclude contracts on behalf of the foreign principal but also when considering the nature of its activities; that person “involves the foreign principal in the national market”. In this case, the local subsidiary was not just processing orders for its parent but also promoted its products, reinforcing the idea that Roche was present in the Spanish market. The lack of economic independence of the local entity, as it worked exclusively for its parent company under strict control, further supported the establishment of a PE in Spain. This position and reasoning were also supported by the Spanish Supreme Court. It attributed to the deemed PE not only the profits from the manufacturing activities but also those from marketing and sales conducted in Spain.

The overall approach of the National and Supreme Courts is largely consistent with the position taken in the Borax case. At the same time, the Roche case further developed and expanded the “complex operative settlement” doctrine by introducing the concept of the “industrial dependent agent” (entity deeply integrated and controlled by the parent) which extended the traditional concept of a dependent agent under Article 5(5) of the OECD MC. Thanks to these doctrines, Spanish tax authorities received a judicial GAAR allowing them to look beyond legal arrangements to the actual substance of business activities ensuring that foreign companies do not avoid a PE through artificial or fragmented structures.

S. 187It is worth mentioning that Spain also opted for the revision of the PE definition offered by the MLI, particularly regarding agency PEs and anti-fragmentation rules. However, considering that, for the MLI, to effectively modify a treaty, a “matching position” is required and only around 30-40 % of Spain’s treaty network is or will be updated. That being said, it is likely that Spanish tax authorities and courts will continue to rely on their judicial GAAR when addressing PE avoidance, especially with countries’ treaties that were not affected by the respective MLI provisions.

5.3.2.2. Italy

The development of a GAAR in Italy reflects the country’s ongoing efforts to address tax avoidance through statutory and judicial measures. Historically, Italy’s approach to combating tax avoidance was fragmented with early attempts being reactive and limited to specific types of transactions. Until the end of the 1980s, tax legislation did not include any tax avoidance provisions with a general effect on income tax.

The introduction of the GAAR began with Article 10 of Law No. 408 of 29 December 1990. This provision allowed disregarding

As can be concluded from the wording, this attempt at codifying the GAAR was limited in scope and primarily targeted corporate reorganizations. However, the need for a more general provision became apparent leading to the introduction of Article 37-bis in 1997 that expanded the application of anti-avoidance rules to a broader range of arrangements beyond reorganizations.

The Italian Supreme Court played an essential role in the evolution of the GAAR by recognizing the existence of an unwritten anti-avoidance principle even before the introduction of Art. 37-bis. It often focused on the misuse of legal instruments to obtain tax benefits without valid commercial reasons. This judicial approach was further reinforced by legislative developments culminating in adopting the formalized statutory GAAR in December 2014 that replaced Article 37-bis S. 188with Article 10-bis of the Statute on Taxpayer Rights. This provision allows tax authorities to disregard business transactions that lack economic substance and are primarily intended to achieve undue tax advantages, ensuring that such transactions have no legal effect for tax purposes.

Possibly because of the existence of this domestic GAAR, Italy did not deem it necessary to adopt the MLI’s rule targeting contract splitting (Article 14). These domestic measures play a crucial role in preventing the artificial avoidance of a PE in Italy which is also evidenced by local jurisprudence.

One of the landmark cases in this field is the Philip Morris case. In this case, the Italian Supreme Court ruled that a German company had established a PE in Italy due to its significant involvement in the activities of its Italian subsidiary. The Supreme Court found that, under the distribution agreement, the German parent had the authority to approve the terms and conditions of sales made by the subsidiary and actively participated in negotiations on behalf of the subsidiary. This level of control indicated that the subsidiary lacked the autonomy to conduct its business independently leading the Supreme Court to conclude that the German parent had a taxable presence in Italy. The Supreme Court’s application of Italy’s domestic statutory GAAR, specifically Article 37-bis, was critical in reaching this conclusion. It allowed the recharacterization of transactions that, while being legally compliant, were structured to avoid the creation of a PE and thus minimize tax liabilities. Like in the previously discussed Spanish cases, the Supreme Court looked beyond the existing distribution agreement, assessed the actual economic relationship between the parent company and its subsidiary, and concluded that the subsidiary was not independent but effectively acted as an extension of the parent’s business operations in Italy. However, unlike the Spanish examples, instead of judicial GAARs in the Philip Morris case, a statutory GAAR was applied.

5.4. PPT as a treaty GAAR and its role in preventingavoidance of a PE

Historically, the primary focus of tax treaties was to eliminate double taxation without explicitly addressing tax avoidance. At the same time, some bilateral treaties already included clauses preventing abuse. However, those were not wideS. 189spread and harmonized. The first attempt to do that is evident when, in 2003, the OECD MC Commentaries were amended for the first time to deal with the problem of treaty abuse.

First of all, it was recognized that prevention of abuse may have a domestic or treaty foundation. The 2003 OECD MC Commentaries then confirmed that domestic anti-avoidance rules, like those based on “substance-over-form”, “economic substance” principles, and other GAARs are compatible with treaty obligations. Then, the 2003 OECD MC Commentaries pointed out that, according to Article 31 of the VCLT, tax treaties should be interpreted based on their goals and in good faith. This led to the understanding that the application of tax treaties is subject to an unwritten rule against abuse.

Second, it introduced a “guiding principle” stating that the benefits of the tax treaty should not be available when the primary purpose for entering certain transactions or arrangements was to secure a more favourable tax position and obtaining that more favourable treatment in these circumstances would be contrary to the object and purpose of the relevant provisions. This change laid down a foundation for the later introduction of the GAAR into the OECD MC.

Although the “guiding principle” in the OECD MC Commentaries provided a helpful anti-abuse framework for tax treaty interpretation, it was not mandatory to follow since it lacked binding legal force. Because it was implemented in the Commentaries rather than in the text of OECD MC, its practical impact was limited, and application heavily depended on the desire of tax authorities and courts to follow the proposed interpretation.

In response to this, in 2015, the OECD released the Action 6 Final Report, “Preventing the Granting of Treaty Benefits in Inappropriate Circumstances” that included changes to the text of the OECD MC by introducing rules preventing treaty abuse. Such rules included the PPT. Under the PPT rule, if one of the principal purposes of transactions or arrangements is to obtain treaty benefits, these benefits would be denied unless it is established that granting these benefits would be in accordance with the object and purpose of the provisions of the treaty.

As expected from the GAAR, the PPT has both subjective and objective elements for determining whether treaty benefits should be granted. The subjective eleS. 190ment focuses on whether it is “reasonable to conclude” that the main purpose of a transaction was to obtain tax benefits. The objective element, which has a higher threshold, requires proof that granting the tax benefit aligns with the object and purpose of the treaty and relevant provisions. Even if the subjective criterion is met, benefits will not be denied if the objective element is satisfied. It is of no relevance for the interpretation of the PPT that the objective element is formulated as an exception because whether a certain element of a provision is formulated as an exception or an additional requirement solely depends on the legal drafting technique used.

As a result, the PPT has merely a signalling function and serves as a reminder that the object and purpose of tax treaty provisions should always guide interpretation, especially when one of the primary purposes of the taxpayer is to obtain a benefit. However, this does not mean that the object and purpose of the provision are irrelevant in other situations. Therefore, the legal effect of the PPT is limited. The provision merely provides guidance for the interpretation of other provisions but should not be read as an exception to those other provisions or as a separate legal basis for not granting tax treaty benefits.

Like with Action 7, the implementation of the Action 6 into bilateral treaties was facilitated through the MLI. However, unlike anti-PE avoidance SAARs, they were included in one of the four BEPS minimum standards. Countries were required to implement these measures, although a certain degree of flexibility was still provided.

Most countries that signed up for the MLI have opted to include the PPT in their covered tax treaties on a stand-alone basis. Even though the jurisdictions were allowed to opt out of the PPT in covered tax agreements that already included the anti-treaty-shopping measures, they have all adopted the PPT. Some countries preferred the PPT because applying it to treaties did not represent technical or administrative difficulties due to their previous experience with anti-avoidance provisions both domestically and on a tax treaty level. For some, the implemenS. 191tation of the PPT was a continuation of an integral legislative anti-abuse process that was taking place domestically, frequently in the framework of compliance with the relevant EU directives (e.g., the ATAD).

5.5. Compatibility and interaction of the GAARs with the Action 7

As discussed in the previous sections based on examples of Australia, Spain, and Italy, prior to Action 7, some countries already had tools to combat PE avoidance via broader interpretations of its definitions by domestic courts and GAARs. In the case that such countries have also adopted Action 7, respective SAARs operate alongside existing domestic general rules and the PPT. This raises the question about their compatibility and interaction.

A first example of interaction between an anti-PE avoidance SAAR and the PPT is provided by Action 7 itself. It is suggested that the PPT can be used as an alternative to the SAAR addressing contract splitting arrangements (MLI’s Article 14). Considering that, unlike the SAARs of Action 7, the PPT has been included in the minimum standard, and most countries that signed up for the MLI reserved the right not to apply Article 14 almost unanimously claiming that they will use the PPT instead.

For instance, Austria opted for the PPT instead of the anti-contract splitting SAAR, arguing that the latter is regarded to be too complicated to apply in practice which would lead to issues of interpretation and can be combated with domestic anti-abuse provisions. The same position was taken by Germany. In the view of Germans, Article 14 does not add any significant tool besides what is already offered by the PPT. In the UK, the government also reserved against special measures by recognizing that the standalone anti-contract splitting rule is not necessary when the PPT is adopted. These examples show that the relationship between this specific SAAR and the PPT can be seen as one of functional S. 192redundancy or overlapping coverage. The PPT along with domestic GAARs create a situation where the SAAR of the Article 14 is seen as unnecessary.

When it comes to other SAARs like Articles 12 and 13, the relationship is more nuanced. If a specific SAAR addresses a particular type of tax avoidance, can the PPT and domestic GAAR also be applied to the same situation, or does the existence of the SAAR effectively override them? This question is crucial because it impacts legal certainty for taxpayers and the administrative practices of tax authorities.

One approach to resolving the potential overlap is the application of the “lex specialis derogat legi generali” principle. Following it, if a specific factual situation falls within the scope of the application of the SAAR, the same situation cannot be tested under the GAAR. The application of the latter should remain of a subsidiary nature in the relationship with the former. When applying this principle to the Action 7, its provisions are at the top of the hierarchy when addressing PE avoidance. Then follows the PPT for broader treaty-related avoidance. Domestic GAARs serve as a more general catch-all for avoidance strategies that fall outside the specific contexts addressed by the treaty provisions.

However, the application of the “lex specialis” principle requires a clear delineation of the scope of the SAAR. In fact, this principle only applies when a normative conflict arises while all of the factual scenarios covered within the special norms are also covered by the broader general rules. Thus, the scope of the SAAR has to completely overlap with that of the GAAR so that both rules are applicable in a given factual scenario, and the conflict is resolved by giving prevalence to the special rule. If the scopes of provisions do not completely overlap, then the “lex specialis” principle cannot be used to accomplish this.

Consider the situation when a taxpayer uses the safe harbour provided by Action 7 SAAR or when the SAAR does not apply to certain transactions. Could the GAAR still be applied to such cases? The challenge lies in determining when it is appropriate to deploy it and scrutinize the specific arrangement. Some scholars argue, for instance, that the opening words of the PPT - “Notwithstanding the other provisions of this convention” - give the PPT priority, allowing it to be exercised when a transaction passes the formal tests of SAARs. Others suggest that these opening words do not refer to other SAARs but focus exclusively on the distributive and potential relief rules that the taxpayer seeks to exploit through abuS. 193sive behaviour. In any case, the presence of the SAAR and its non-application to a taxpayer’s structure is not irrelevant when applying the PPT. In situations when the SAAR specifically addresses the abuse in question, its non-application could imply that the taxpayer’s transaction does not constitute the specific type of abuse that the SAAR was designed to prevent. If the SAAR is the more specific rule and does not apply, it may indicate that the arrangement is not abusive under the strict terms of the SAAR. Even if the PPT could technically be applied, doing so would contradict the SAAR’s clear indicators of when a transaction is abusive. The purpose of the SAAR, among others, is to provide legal certainty, and applying the PPT in these cases would undermine this objective.

6. Conclusions

This thesis started by exploring the evolving PE concept through BEPS Action 7 that was designed to address the artificial avoidance of a PE in the source states. Through a detailed analysis of its implementation via the MLI, it becomes clear that the success of this initiative alone has been somewhat limited. The quantitative analysis of adoption rates across different countries indicated that the application of Action 7 measures has not been universally accepted considering reservations from various countries.

In response to the partial success of Action 7, this thesis has also examined examples of alternative domestic anti-avoidance measures that could address PE avoidance. These included Australia’s MAAL and domestic statutory and judicial GAARs in Spain and Italy that were already established when the Action 7 Final Report was released by the OECD in 2015. Moreover, Action 6 and its implementation have been looked at in the context of its role in preventing artificial avoidance of a PE.

This thesis then focused on the interaction between various anti-abuse measures and Action 7 to reveal the complex dynamics and hierarchy between them. When assessing the compatibility of the MAAL in Australia, it has been concluded that it not only complements Action 7 but can also override international tax treaties by creating a concept of a “notional PE” that does not exist there. Similarly, the interpretation of PE treaty provisions by Spanish and Italian courts using domestic GAARs offers a potential for a broader threshold than what Action 7 might suggest. When it comes to the PPT, it has also been analysed in this work how the treaty GAAR was used as an alternative to some Action 7 provisions (especially Article 14) and how it interacts with the other measures. This work also looked into the complex architecture of different anti-avoidance measures and the interS. 194play between them, especially in cases when they co-exist and may apply simultaneously to prevent avoidance of a PE.

In conclusion, while Action 7 is an essential effort in the global fight against artificial avoidance of a PE, its limited adoption by different jurisdictions suggests that domestic anti-abuse rules like SAARs and GAARs as well as the PPT play a crucial role in combating this type of abuse. On the one hand, these measures provide more flexibility and close the gaps that are not directly addressed by the tax treaties. On the other hand, they work against the harmonization of the international tax landscape and create legal uncertainties for taxpayers.