Abstract
Recent public concerns and publicity about the extent of tax avoidance by some of the largest and most respected transnational corporations have prompted numerous parliamentary inquiries and intergovernmental initiatives. Among the questions raised during hearings in the UK House of Lords on 11 June 2013 were whether such avoidance could be more effectively prevented either by a reduction in the complexity of the rules, or by a more aggressive interpretation of those rules by tax authorities. These questions raised issues concerning the complexity and interpretation of law, which received interestingly different responses from the academics to whom they were posed, and they merit closer examination, which is the aim of this article. The article begins by discussing some types of complexity, then considers how these are rooted in the indeterminacy of legal language and finally discusses the sociology and politics of the interpretive communities concerned. The main part applies this analysis to the historical development of international tax rules, seen in the context of these current pressures and initiatives for reform of international corporate taxation.
Types of complexity
International tax coordination is governed by tax treaties, expressed in rules which must be described as plain and simple, at least by the normal standards of legalese. The central treaty provision governing taxation of a transnational corporate group 2 states:
Where: (a) an enterprise of a Contracting State participates directly or indirectly in the management, control or capital of an enterprise of the other Contracting State, or (b) the same persons participate directly or indirectly in the management, control or capital of an enterprise of a Contracting State and an enterprise of the other Contracting State, and in either case conditions are made or imposed between the two enterprises in their commercial or financial relations which differ from those which would be made between independent enterprises, then any profits which would, but for those conditions, have accrued to one of the enterprises, but, by reason of those conditions, have not so accrued, may be included in the profits of that enterprise and taxed accordingly.
Tax treaty texts are generally incorporated directly into domestic law 3 and in effect create a special tax regime for the taxpayers to whom they apply. Since essentially this same provision has been used in the more than 3000 bilateral tax treaties now in force, it constitutes a clear and simple legal rule that applies globally, although not universally. It is a remarkable, if under-appreciated, example of a ‘hard’ rule of international law, binding states as well as having direct effect for individual legal subjects. Of course, they are impersonal subjects (corporations), and the rule gives them rights, not duties, since it restricts states’ power to tax. Corporations can, and frequently do, bring cases in national courts challenging national tax measures that they consider contrary to a treaty. Even as powerful a state as the United States found itself significantly constrained in carrying through its major tax reforms of 1986, as they related to transnational corporations (TNCs) (Picciotto, 1992: 325). This creation of legally binding rights for corporations is a sharp contrast with the various attempts to create ‘codes of conduct’ with obligations for TNCs, which generally take the form of non-binding ‘soft law’ (Picciotto, 2011, Chapter 4).
However, the text of the treaties provides only the skeleton of international tax rules. Flesh is put on these bones by additional authoritative documents. First, the model treaties are accompanied by commentaries, which are drawn up by the bodies responsible for drafting the model treaty texts, 4 and are intended to guide their interpretation. Under generally accepted principles of treaty interpretation in international law, as well as under most domestic laws, these commentaries can be referred to by courts and others as an aid to understanding the meanings of the treaty articles. In addition, those same bodies issue reports and other documents that are also considered authoritative.
The Organization for Economic Cooperation and Development (OECD) model treaty commentary on Article 9(1) amounts to less than a couple of pages. However, it refers to the OECD’s Transfer Pricing Guidelines for Multinational Enterprises and Tax Administrations, describing them as ‘internationally agreed principles [which provide] guidelines for the application of the arm’s length principle of which the Article is the authoritative statement’. Those guidelines, began as a report on transfer pricing in 1979, were renamed guidelines when revised in 1995, have been periodically revised since then and now extend to some 370 pages (OECD, 2010). The commentary also refers to the report on thin capitalization issued by the Committee on Fiscal Affairs (CFA) in 1986, which is one of a number of such reports (now 24) that have for some time been included in the compilation described as the ‘full-length version’ of the OECD model treaty (OECD, 2012). These additional reports are therefore also likely to be considered as authoritative aids to interpretation of the treaty provisions. 5
These considerable additions to the basic tax treaty provision may be described as producing complexity through elaboration. The plain and simple basic texts of the tax treaties have become overlaid with increasingly elaborate supplementary provisions, ostensibly aimed at clarifying how they should be applied. The result is a labyrinth of provisions, requiring skilled navigation. An important aspect of this complexity is that the relationship between the various types of provisions is not clear. In principle, only the treaty texts are ‘hard’ law, the commentaries, guidelines and other documents are merely supplementary aids to interpretation. However, the specificity of the latter means that they are the provisions applied in practice. Nevertheless, their language is discursive, and they refer to concrete situations, providing many examples, rather than being expressed in the abstract terms of legal rules. Hence, although these texts are not expressed in abstruse or technical language, specialist skills are involved in understanding the relationships between them and deploying them effectively.
This may be contrasted with what can be described as complexity through attempted precision. This is the main characteristic of UK legislative drafting. An apposite illustration is provided, indeed, by the provisions of Taxation (International and Other Provisions) Act (TIOPA) that aim to spell out in much greater detail the ‘arm’s length principle’ (ALP) of Article 9 of the model treaties. They are contained in part 4 of TIOPA, which has no less than 45 sections, opening with the following:
147. Tax calculations to be based on arm’s length, not actual, provision For the purposes of this section ‘the basic pre-condition’ is that –
provision (‘the actual provision’) has been made or imposed as between any two persons (‘the affected persons’) by means of a transaction or series of transactions, the participation condition is met (see section 148), the actual provision is not within subsection (7) (oil transactions), and the actual provision differs from the provision (‘the arm’s length provision’) which would have been made as between independent enterprises. Subsection (3) applies if –
the basic precondition is met, and the actual provision confers a potential advantage in relation to United Kingdom taxation on one of the affected persons. The profits and losses of the potentially advantaged person are to be calculated for tax purposes as if the arm’s length provision had been made or imposed instead of the actual provision. Subsection (5) applies if –
the basic pre-condition is met, and the actual provision confers a potential advantage in relation to United Kingdom taxation (whether or not the same advantage) on each of the affected persons. The profits and losses of each of the affected persons are to be calculated for tax purposes as if the arm’s length provision had been made or imposed instead of the actual provision. Subsections (3) and (5) have effect subject to –
section 165 (exemption for dormant companies), section 166 (exemption for small and medium-sized enterprises), section 213 (this Part generally does not affect calculation of capital allowances), section 214 (this Part generally does not affect calculation of chargeable gains), section 447(5) and (6) of CTA 2009 (this Part generally does not affect how exchange gains or losses from loan relationships are accounted for), and section 694(8) and (9) of CTA 2009 (this Part generally does not affect how exchange gains or losses from derivative contracts are accounted for). The actual provision is within this subsection if it is made or imposed by means of any transaction or deemed transaction in the case of which the price or consideration is determined in accordance with any of sections 225F to 225J of ITTOIA 2005 or any of sections 281 to 285 of CTA 2010 (transactions and deemed transactions involving oil treated as made at market value).
It should be noted that these provisions do not affect the treaty text, they merely complement it, by specifying the circumstances in which the ALP (the ‘provision’ in section 147(1)(d)) applies in UK tax law. However, further sections in part 4 of TIOPA add additional detailed provisions amplifying what ‘arm’s length’ means in specific circumstances, especially in financial transactions between related entities. In that way they also add complexity through elaboration. Deciding on the tax treatment of a transaction now entails consideration not only of the treaty rule, and in addition the commentaries, the guidelines and other reports, but also of these domestic legal rules as well as the interaction of all these norms.
This form of complexity deploys logical linguistics to attempt to nail down precise applications for rules, in contrast to the use of broad general principles. It is characteristic of Anglo-Saxon legislative drafting, carried probably to the greatest extreme in the United Kingdom. This seems to be due to the dominance of techniques that have become firmly embedded in the parliamentary draftsman’s office, despite hopes that European (EU) influence might result in the use of at least a mixture of general principles and detailed rules (Avery Jones, 1996). As with elaboration, this also creates a labyrinth of provisions, though one requiring different skills to navigate, essentially an ability to follow abstract logic, something like computer programming. In contrast with the OECD’s guidelines and reports, these legal rules are expressed in abstract terms, with few specific examples. Nevertheless, their aim seems to be to provide precision, through ‘bright line’ rules rather than general principles, such as the treaty rules.
This raises the question, why is the law not limited to the basic tax treaty rules? Since these are the texts that matter, why add to them? Consideration of this question has the following two aspects: regarding first the nature of legal language and second, the social and political dynamics of enforcement and compliance with tax law, in this case, particularly international tax. These will be discussed in the next two sections
Indeterminacy and Legal Formalism
The first part of the answer to the question above is that although the tax treaty rules are relatively clear and simple, they are not precise. They are expressed as general principles, which could be interpreted to apply very differently in any specific case. This raises the central issue of the indeterminacy of the meaning of legal rules. As a starting point, Lewis Carroll’s well-known aphorism is worth restating: ‘When I use a word’, Humpty Dumpty said, in a rather scornful tone, ‘it means just what I choose it to mean – neither more nor less’. ‘The question is’, said Alice, ‘whether you can make words mean so many different things’. ‘The question is’, said Humpty Dumpty, ‘which is to be master – that’s all’.
Alice Through the Looking Glass.
The first is at the general level of language. As linguistic philosophy has stressed, linguistic signifiers (words) do not have an intrinsic meaning. Meaning depends on the linguistic context. Since language is social, this also means its social context and that meaning is constructed through social interaction. Meaning can therefore be fluid and changeable. In some social contexts, participants may value clear understandings, so prefer to establish fixed and stable meanings. In others, ambiguity may help to maintain harmony, despite underlying disagreement or conflict. Sometimes, such conflict may be played out linguistically, through contests either over the meaning of certain words or phrases or over how to express a particular concept. There is clearly much more that could be said about linguistic indeterminacy, but we must move on to next two levels, which are specific to the nature of legal rules.
The second level of indeterminacy flows from the characteristics of liberal legality. In liberal polities, with at least a formal separation of legislative, adjudicative and executive powers, law mainly takes the form of general rules articulated in advance, leaving individuals free to make their own decisions within the guidance offered by the law. This clearly involves many tensions or contradictions, especially between the degree of abstractness as against specificity with which laws are formulated and the need to interpret those abstract principles when applying them to specific cases.
However, liberal legality can be supplemented by administrative procedures, such as prior rulings. Tax authorities may make use of these, although generally with caution. In international taxation, indeed in relation to the ALP itself, many countries have introduced an Advance Price Agreement (APA) procedure. A firm that submits full and documented details of its planned inter-affiliate transactions and proposed transfer-pricing methodology can get advance approval for a specified period of years, subject to the terms of the agreement. However, whilst this may alleviate the potential problems of conflicts over interpretation resulting from a broad principle such as the ALP, it cannot replace the rule of law in a system that is fundamentally liberal. Administrative authorities can only tie themselves down by giving rulings in relation to very specific circumstances and after careful consideration. APAs therefore usually require the company to provide extensive information and documentation and subsequent audits to verify compliance with their terms. Thus, although they do have the advantage of providing clarity, they are relatively little used by TNCs, and their main supporters are the large accountancy firms, for which they provide good business. One of the few academic studies suggests that, interestingly, APAs seem to be resorted to especially when firms seek to use a non-standard transfer pricing methodology (Rogers and Oats, 2013). There are also potential problems of legitimacy, if potentially contentious issues are dealt with in advance by private and ad hoc deals, especially if they involve significant sums. Indeed, some countries have used APAs to agree arrangements providing advantages to foreign firms, which are now being challenged as ‘sweetheart deals’, notably in the proceedings brought by the European Commission in 2014 against Ireland in respect of an APA granted to Apple Computer in 1991 (European Commission, 2014) as well as Luxembourg’s secret APAs that received wide publicity in November 2014.
The problem of legal indeterminacy is particularly acute for rules applying to economic activity. This is due to the disjuncture between legal form and economic purpose since different legal forms can be devised to achieve the same or a similar economic purpose. This is exacerbated in areas that have a direct negative economic impact on regulated persons, such as taxation. Since a tax imposes a direct cost burden on a defined type of economic activity, an immediate incentive is created for those who might be affected by a rule, to devise a legal form for their activity that complies with the letter of the law whilst achieving an economic purpose that the legislator did not envisage, or intend, or perhaps even desired to prohibit. This is all the more so for activities that are organized and planned, as compared to those that are a routine or spontaneous part of daily life. Also, as is pointed out by Gribnau in this collection, the use of taxes as a policy instrument by governments has fostered a generally instrumental attitude to compliance by all citizens, which may encourage a widespread acceptance of tax planning or avoidance.
In the context of a social process geared towards achieving compliance with formal norms, the indeterminacy of general legal principles raises a range of issues about the design of regulatory systems. In particular, broad general principles leave great scope for interpretation and so for uncertainty. Wide room for disagreement between those responsible for enforcement (regulators) and those to whom the rules apply (regulatees) can lead to ineffectiveness, unless regulators are given a high degree of discretion but this can lead to arbitrary, unfair or partial decisions. Hence the attempt to draw up rules to try to provide precision by anticipating every contingency but which tend to produce complexity. A preference for this type of complexity derives or is justified by an instrumental and formalist view of law: ‘a narrow approach to legal control – the use of clearly defined, highly administrable rules, and an emphasis on uniformity, consistency, and predictability, on the legal form of transactions and relationships and on literal interpretation’. (McBarnet and Whelan, 1991: 849).
However, the formalist approach does not prevent avoidance, but shifts it to a new level, involving gameplaying and ‘creative compliance’, to use the phrase of McBarnet and Whelan (1991: 849). Addressing this, regulation theorists have developed the concept of ‘responsive regulation’, which suggests that regulators should apply sanctions selectively, whilst trying to build a relationship of mutual trust and understanding with regulatees to ensure harmonious understandings of what compliance requires (Ayres and Braithwaite, 1992; Braithwaite, 2001). This approach has been applied particularly in the tax field (Braithwaite, 2007; Gribnau, 2007), including in international taxation (Rawlings, 2007).
Certainly, there has been much emphasis by tax authorities in recent years, both national and international, on the importance, especially in relations with big business, of establishing relationships based on trust, aiming at ‘clarity through effective consultation and dialogue’ (HMRC, 2006: 5), sometimes described as an ‘enhanced relationship’ (OECD, 2008, Chapter 8; OECD, 2013a). However, increased use of such arrangements in the United Kingdom has led to considerable public controversy. Activist group UK Uncut’s legal challenge to a settlement reached with investment bankers Goldman Sachs by the then head of Her Majesty’s Revenue and Customs, against the advice of his lawyers, revealed ‘the deeper tensions entailed in supplementing a traditional command and control approach to tax regulation with one more dependent on relationships and co-operation’ (de Cogan, 2013: 552).
The third level of indeterminacy arises because legal precepts are not merely positivist statements of a general character but norms, so that interpreting their meaning when applying them to particular cases entails a normative judgement. Thus, deciding whether a particular instance fits within a general principle is not just a factual inquiry but entails a value judgement since laws are normative. The implication of this is that even the core meaning of a legal norm might depend on a shared view of the values or purposes that underlie it. Differing views about those values will result in different interpretations of the meaning of the norm, which may be equally potentially acceptable. This understanding leads to a different view of the nature of regulation and compliance from that of legal formalism and instrumentalism. It suggests that different interpretations may arise from different views about the purpose of the rule in question or the values it aims to promote. The interpretations of lawyers, especially judges, are also authoritative normative pronouncements, so they also involve social and political power.
The second and third layers of indeterminacy discussed here open up a sociological perspective, concerning the techniques and practices deployed by lawyers in the formulation and interpretation of legal texts. A fruitful approach was suggested by Pierre Bourdieu, who argued that this involves the appropriation of the ‘symbolic power which is potentially contained within the text’, which he sees as taking place through competitive struggles to ‘control’ the legal text (Bourdieu, 1987: 818). He also proposed a sociological answer to the question of how, in view of the indeterminacy of legal texts, the system can nevertheless function in a relatively stable manner. He suggested that coherence emerges partly through the social organization of the field, which produces mutual understandings based on ‘habitus’ and partly because, to succeed, competing interpretations must be presented ‘as the necessary result of a principled interpretation of unanimously accepted texts’ (1987: 818). In this sense, law operates to defuse social conflicts and depoliticize them, shifting political and economic conflicts on to the terrain of debates over the symbolic power of texts.
This explains the apparent paradox that, whilst lawyers spend much of their time disagreeing about the meaning of texts, they generally do so from an objectivist perspective. They usually prefer to deny that indeterminacy is inherent in legal rules and tend to attribute disagreements to bad drafting and lack of clarity in the texts, which are said to create ‘loopholes’ in the logical fabric of the law. Bourdieu’s perspective can also be integrated with the ‘interpretivist turn’ in sociolegal studies and critical approaches to law. These study the ways in which the indeterminacy of legal texts provides the space for the deployment of legal skills and techniques, the introduction of political preferences and social values, and ultimately, the ways in which law is deployed in and mediates struggles over power.
However, a purely sociological perspective may be limited, especially if it focuses only on competition between professionals such as lawyers, aiming to accumulate ‘social capital’ by asserting themselves as authoritative interpreters of texts. Whilst this is certainly an important element, there should also be consideration of the political and economic context in which these struggles take place, as well as of the substance of the issues being debated through the mediation of competing interpretations. The next section aims to integrate these approaches into a re-examination of the international tax rules discussed in the first section of this article, in a broader sociological as well as political context.
International Tax Politics, Technocracy and Complexity
The (relatively) clear and simple tax treaty provisions discussed at the beginning of this article originated from debates early in the last century, concerning the coordination of national taxation of income or profits as applied to international business. The first model tax treaties resulted mainly from discussions and negotiations conducted through the League of Nations, particularly at a conference in 1928. The model provisions were used as a basis for negotiating bilateral treaties from the 1930s, spreading more quickly from the 1950s especially among OECD countries, which developed model treaty articles from the late 1950s, first formalized as a model treaty in 1963. However, the process of elaboration of more complex provisions is much more recent, originating in the 1960s.
It has occurred essentially as a reaction to the tax planning strategies of TNCs, which were themselves developed as a response to the basic treaty rules. As this section will explain, this process resulted from the flawed formulation in Article 9, which in effect encouraged firms to develop ever more complex corporate structures, leading tax authorities to react in turn by formulating increasingly complex regulations purportedly based on Article 9. The origins of this vicious circle lie in the relationship between politics and technocracy, which can be traced back to the beginnings of the system, but persist today. Any hopes for effective reform, which could deliver a simpler and more effective system, depend on understanding and transcending this relationship.
The Fundamental Flaw
The driver of this process has been the fundamental flaw in the basic treaty rule, contained in Article 9 (cited at the beginning of this article). Although TNCs are structured legally as corporate groups, sometimes nowadays with hundreds of affiliates, they operate as integrated enterprises under central management and control. Yet for the purpose of allocating tax rights between states, Article 9 states that they should be treated as if they were independent entities in each country. The tax treaty rules were aimed primarily at portfolio investment, which in the 1920s was the dominant form, that is, lending by investors to businesses abroad. Understandably, they gave the host country, where the business was located, the right to tax its profits, whilst the investor’s country of residence could tax the returns (interest, dividends) on the capital invested.
The ‘separate entity’ principle in Article 9 was an adaptation of this basic rule for the case of foreign direct investment by multinational corporate groups (TNCs). Revenue authorities understood that TNCs posed a special problem since the business profits of their component affiliates would depend on the pricing of intra-group transactions and the financial structure of the firm. Hence, Article 9 enabled national tax authorities to adjust the accounts of related entities to combat possible distortions or prevent ‘diversion’ of profits. The various methods that were used to deal with the special case of TNCs were examined by a study for the League of Nations Fiscal Committee in 1932–1933, coordinated by the US representative, Mitchell B. Carroll (Picciotto, 1992, Chapter 1, 2013a).
Carroll reported that most tax authorities began from the accounts of the local entity, especially if it was a separately incorporated subsidiary. However, they generally had to examine those accounts carefully and had powers to adjust them as necessary. Local establishments of foreign companies were frequently assessed on an ‘empirical’ or ‘fractional’ basis. Empirical methods entailed estimating the income ‘by comparing the given enterprise with similar enterprises, or taking into account turnover, assets and other readily ascertainable factors’; whilst ‘[b]y fractional apportionment is meant the determination of the income of one establishment of an enterprise by dividing total net income in the ratio of certain factors – for example, assets, turnover, pay-roll or a fixed percentage’ (Carroll, 1933: 46).
In the case of corporate groups, Carroll considered that the parent had a legal obligation to deal with an affiliate as if it were a separate legal person. As long as this was respected, tax authorities could assess the affiliate on the basis of the separate accounts, scrutinizing them to verify whether they were indeed at arm’s length. However, where such verification was difficult due to extensive interlocking transactions, they resorted to empirical or fractional methods. If they considered that there was a ‘diversion of profits’ such that adjusting accounts was not possible, countries had various methods for ‘reaching the parent’. Some, such as the United Kingdom, could treat a subsidiary as a dependent agent, and charge the parent through its subsidiary as agent, usually on the basis of a percentage of turnover. Others, notably Germany and Spain, could assess a parent company directly if they formed an ‘organic unity’. Hence, in addition to the separate entity principle in Article 9, the formulary apportionment method was also explicitly accepted in Article 7(4) of the model tax treaty. Although Article 7(4) only applies to Permanent Establishments (PEs), it was understood then, and continues to be widely accepted, that a subsidiary that is a dependent agent of its parent can be treated as a PE (Avi-Yonah and Tinhaga, 2014). Indeed, the direct control usually exercised by the parent in a multinational corporate group means that most subsidiaries are in practice dependent (Le Gall, 2007). This provision was dropped by the OECD model in 2010, but remains in the United Nations (UN) model, and is still present in the vast majority of actual bilateral treaties, including some recently concluded (Avi-Yonah and Tinhaga, 2014).
The national reports for Carroll’s study showed that those tax authorities that started from separate accounts were clearly well aware of the limitations of such accounts and also that they could not be checked solely by reference to alternative ‘normal’ pricing. For example, the British Inland Revenue stated that separate accounts could only be used as the basis in about half their cases and then only because there was the threat in reserve of an assessment based on percentage of turnover (League of Nations, 1932: 191). The principle of fractional apportionment also had clear advantages, as argued in the Spanish report, and shown by the experience of federal systems. Since it entailed treating the accounts of a globally integrated enterprise as a single unity, it avoided the many problems and administrative difficulties involved in attempting to construct arm’s length relationships where none existed.
However, the consensus among the tax specialists was that a unitary approach would be difficult if not impossible to adopt for political reasons, since they considered that it would require international agreement on (i) tax accounting principles for assessment and (ii) a common allocation formula. The ALP offered a more pragmatic and practical approach, which seemed to them easier to operate in a network of bilateral treaties. At that time, it was understood that the ALP did not establish a clear or precise measure but at best a general principle. Indeed, the German report accepted that fractional apportionment was superior in principle and would in practice be used in the many cases where separate assessment was not feasible (League of Nations, 1932: 122). However, adoption of the ALP merely converted the problem, from a decision on the principles of general apportionment by formula to negotiation of specific ad hoc apportionments, by adjustment of transfer prices to ensure a fair profit split. The German report stressed that this would in practice require close cooperation between tax authorities, from which more general principles could perhaps emerge. Unfortunately, this did not occur, and international tax rules have continued to be based on the fundamental flaw of the separate entity principle.
Responding to the Flaw
The basic treaty rules do not apportion the taxable profits (tax base) of a TNC but allocate rights to tax, that is, the host country of the subsidiary should tax its business profits, whilst the parent company’s state of residence has the primary right to tax investment returns (e.g. dividends or interest on loans). However, the rapid expansion of TNCs from the 1950s onwards, especially from the United States, was largely funded from retained earnings. This was partly due to controls on capital movements, which only began to be relaxed in the late 1950s, but there were also tax advantages since the foreign earnings were generally taxed at a lower rate than that applicable in the Unites States, often due to incentives and tax holidays offered by countries wishing to attract investment. If foreign profits were retained for reinvestment rather than returned to the US parent, they would continue to benefit from these relatively low tax burdens.
This sparked a debate in the United States about ‘tax deferral’ (Barlow and Wender, 1955), leading to legislation by the Kennedy administration in 1962 to tax the retained earnings of certain ‘controlled foreign corporations’ (CFCs), if they constituted ‘active business income’. This issue is today again central in US international tax policy debates. Press reports and parliamentary inquiries have spotlighted the accumulation of some US$2 trillion by US firms ‘offshore’ and stark anomalies such as Apple choosing to borrow US$1.7 billion to help fund a special dividend, rather than tap into its US$145 billion of cash reserves held offshore (Drucker, 2010; Editorial 2013; US Senate, 2013).
In economic terms, the tax treaties aim to ensure tax neutrality, that is, host country taxation of business profits ensures equal treatment between foreign-owned and local businesses, whilst residence country taxation of returns on investment ensures that investors’ income is taxed the same whether it derives from domestic or foreign investment. The ability of TNCs to finance their international expansion through retained foreign earnings upsets this balance. More fundamentally, economic theory tells us that the profits of such firms are largely due to their ability to integrate and coordinate operations, which provides synergy and economies of scale and scope (Coase, 1988; Williamson, 1985). Hence, profitability depends on the integrated whole and cannot be attributed to different parts, so the residence–source split makes no sense for TNCs.
The tax treaty rules are further undermined when TNCs also reduce the scope of source taxation by making deductions from their subsidiaries’ business profits for payments such as interest, fees for central management and royalties for intellectual property rights. If retained abroad rather than repatriated to the parent, this income is not taxed anywhere and can be described as ‘stateless’ (Kleinbard, 2011) or ‘homeless’ (Wells and Lowell, 2011). TNCs typically use variations of a ‘stepping stone’ structure, deploying conduit companies in countries with tax treaties to avoid source taxation and base companies in low- or zero-tax countries to accumulate retained earnings (Corporate Reform Collective, 2014, Chaper 1). Thus, the problem of avoidance affects both source and residence country taxation.
The United States also referred the issue to the OECD in 1962, which set up a Working Party on Avoidance by Abuse of Tax Conventions. 6 Its investigations, secret until recently, opened up a range of issues that went to the heart of the treaty principles. The same questions are again on the agenda today since the issue of tax avoidance by TNCs has been taken up by the G20 group of world leaders. In the Tax Annex of the St Petersburg Declaration of 2013, they have mandated a potentially far-reaching review of international tax rules (G20, 2013). Using clear language, the Declaration states that tax rules should ‘ensure that profits are taxed where economic activities occur and value is created’. But the detailed work has been entrusted to the OECD technicians, the main architects of the system, through their project on ‘base erosion and profit shifting’ (BEPS).
In the 1960s, the discussions on combating avoidance were largely inconclusive and no significant changes were made when revised versions of the model treaty were released in 1963 and 1977. In relation to CFCs, the CFA eventually in 1986 approved two reports on the use of base and conduit companies, 7 which set out the problems, but essentially concluded that no generally applicable solutions could be agreed. Switzerland in particular argued that since tax treaties define and allocate rights to tax, a residence country cannot use anti-avoidance rules to tax the undistributed income of foreign affiliate validly incorporated in a treaty–partner state, unless the treaty specifically recognizes the applicability of those rules (OECD, 2012 R(5), paras 95–97). Other states did not go so far, but the Committee suggested that CFC rules should be guided by an international consensus, to be formulated by the CFA (2012, para. 48). This has not been done formally, but the section on Improper Use of the Convention in the commentary to Article 1 of the model convention was expanded in 1992, to include discussion of counteracting measures such as CFC rules, and in effect provides criteria to validate such measures.
The other major issue was transfer pricing, again raised by the United States, in a letter to the CFA in 1964 from Stanley Surrey, who initiated many international tax reforms in this period. As with CFCs, Surrey hoped that at least some general rules could be agreed, if only to deal with the risk of inconsistent transfer price adjustments by states. The United States adopted detailed transfer pricing regulations in 1968 but initially found little support among its OECD partners for the adoption of any international rules. The general opinion was that Article 9 was sufficiently clear, and there was little support even to extend the commentary. The United States continued to argue for the formulation of guidelines, which it suggested should be non-binding. The eventual result was the report on Transfer Pricing and Multinational Enterprises of 1979, which substantially adopted the approach in the 1968 US regulations, followed by reports on three specific issues (1984) and another on thin capitalization (1987). In 1995, these were consolidated into what were finally formally described as Transfer Pricing Guidelines.
The precipitating event for the formulation of the guidelines was the conflict that had arisen from the adoption of a new approach by the United States. When the Congress debated both deferral and transfer pricing in the 1960s, it considered the possibility of an alternative approach, based on formulary apportionment of the total profits of a TNC, along the lines of the system long used by US states (Durst and Culbertson, 2003). The Treasury instead opted for the CFC and transfer pricing regulations but concerns soon resurfaced that these reforms had been ineffective. These reached a peak with a report in 1981 (US-GAO, 1981), which showed that the attempt in the 1968 transfer pricing regulations to provide clear and precise rules based on the ALP had failed, and in a large number of cases adjustments had to be made by imprecise ‘other’ methods, accepted only as a fallback under the Regulations.
The problem centred especially on allocation of profits from ‘intangibles’, that is, technology and other knowledge-based assets, which are central to the competitive advantages of TNCs, and were becoming increasingly important in the world economy. In response, the Tax Reform Act of 1986 included the first substantial amendment since its enactment in 1928 to the basic statutory rule allowing adjustment of accounts (section 482), specifying that for the transfer of an intangible the income attributable ‘shall be commensurate with the income attributable to the intangible’. The issue was evaluated in a report by the US Treasury (1988). It showed that the problem of intangibles is central to the issue of intra-firm relationships and is closely related to charges for central services and other general overhead costs, which are often ‘bundled’ in intangibles pricing.
It proposed a new criterion, the ‘arm’s length return’ method. This was carefully designed to appear compatible with the ALP but essentially used analysis of functions performed to attribute a normal rate of profit. It quickly provoked an international storm and was notably denounced by the chair of the Taxation Commission of the International Chamber of Commerce, Wolfgang Ritter, as ‘nothing but another form of unitary taxation’ (cited in Picciotto, 1992: 221). The CFA produced a more careful and measured evaluation, which emphasized the need for international coordination, acknowledged the difficulties of transaction-based methods and pointed to the dangers of the ‘comparable profits’ criterion (OECD, 1993). It recommended rather the use of a ‘profit-split approach’, dividing the aggregated profits of the actual associated enterprises and urged that this method should be ‘subject to a more exhaustive examination’ to make it less arbitrary (1993: 24).
The outcome, in the 1995 guidelines, was the addition of two new approved methods for transfer pricing adjustment, that is, the ‘transactional net margin method’, applying a functional analysis before attributing a normal profit and the profit-split method. This acceptance of profit split marked a significant shift towards what could be described as a unitary approach. However, the guidelines resolutely rejected this inference, describing both the new methods as ‘transactional profits methods’ and including a section resolutely rejecting ‘global formulary apportionment’. In 2010, the CFA once again returned to the vexed question of intangibles transfer pricing, and the issue was also included in the BEPS Action Plan (OECD, 2013b).
General Principles and Technical Elaboration
The result of these debates and conflicts has been an uneasy balance between upholding the clear and simple language of general principles in the tax treaties and issuing more detailed elaborations of the methods and norms applied in practice. In 1992, the model treaty became ‘ambulatory’, that is, published in loose-leaf form to allow regular revisions, and in 1997 the additional reports began to be published as appendices, by 2012 amounting to 24. However, no major changes have been made to the basic treaty texts, although some of the modifications have been significant, such as the expansion of the exchange of information provisions in Article 26 and the modification of Article 7 and removal of Article 7(4) (mentioned in ‘The Fundamental Flaw’ section). The commentary has been significantly expanded, but the major innovation has been adding the Appendix of Reports. Although these have not been described as guidelines, as for transfer pricing, they are nevertheless apparently intended to be considered authoritative and, as mentioned in ‘Types of Complexity’ section, are treated as such in some countries’ domestic law.
In parallel with the increased elaboration of norms by the OECD, national legislation has also expanded. On some topics it has become increasingly complex, as regards both elaboration and technical abstruseness. This seems to be especially the case where the OECD has not articulated agreed norms. A significant case is CFC rules in relation to which, as mentioned above, opinion was very divided and only a broad consensus could be reached. In practice, countries that have introduced CFC rules (around half of the OECD states, plus a few others) have followed similar general principles, although their details vary widely. Indeed, these are generally among the most abstruse and technically complex areas of tax law.
Essentially, such regimes rest on three tests, namely, (i) ownership or control (to identify an ultimate parent), (ii) passive income and (iii) low tax. All have become increasingly difficult to apply: the first, as TNCs have become more decentralized and regionalized, and many have adopted multi-tier structures; the second, with the growing importance of services and other activities that can be ‘virtual’ (sometimes described as the shift to the digital economy) and the third, with the growth of preferential and low-tax regimes offering ‘production havens’ (e.g. Ireland). These difficulties seem also to have contributed to the technical complexity of legislation. For example, Australia’s CFC legislation 8 runs to over 150 sections and some 52,000 words of primary legislation. The US CFC rules (subpart F) have been described as ‘too complex to be susceptible of summary’ and interact with the foreign tax credit rules, the detailed application of which is ‘even more cruelly byzantine than are the subpart F rules’ (Kleinbard, 2011: 26). Concealed behind this complexity is a system that now ensures that substantially all the foreign income of US TNCs can be effectively exempt from all taxation (Fleming et al., 2009). This stark anomaly is hard for the US authorities to challenge or rectify because US TNCs generally ensure that they do pay US tax on their activities in the United States and obliging them to pay the high US rate of 36% on their worldwide profits would damage their competitiveness. 9
The tension between the general principles and the more detailed rules can be seen perhaps most clearly in relation to transfer pricing, where the OECD has elaborated what it terms guidelines. Since 1995 these have established five approved methods for transfer price adjustments. The first, the ‘comparable uncontrolled price’, requires the adjustment of prices of inter-affiliate transactions based on identifying a ‘comparable’ transaction between unrelated parties. This lies at the heart of the controversy about the ALP since the practical experience is that true comparables can generally not be found, for the very good reason that TNCs generally produce distinctive products or services, and can do so at lower prices due to their advantages of economies of scale and scope. The difficulty of identifying appropriate comparables was the main finding of the 1981 report for the US Congress (US-GAO, 1981) and is a repeated refrain among practitioners especially speaking in private.
As regulation of transfer pricing has spread, particularly to developing countries, they have also confirmed the difficulty of finding suitable comparables. Since the search also requires time and valuable skills, leading countries have adopted alternative methods. Thus, Brazil has a system based on legislated fixed profit margins, whilst China makes adjustments to take account of ‘location specific advantages’, frequently falls back on the profit-split method, and considers that in some circumstances ‘a global formulary approach should be a realistic and appropriate option’. 10 In 2013, the US official responsible for negotiations over conflicting transfer price adjustments complained, in a meeting for practitioners, of his Indian counterpart’s preference for policy over rules, and for profit split rather than the more transactional methods (Parillo and Trivedi, 2013).
Rather than trying to deal with such points of conflict, the BEPS Action Plan seems set on further elaboration of the guidelines, rendering them even more murky. Notably, the report issued in September 2014 on intangibles (OECD, 2014a) is obscure and in some respects contradictory. It still includes the by now ritualistic reaffirmations of the ALP and comparability analysis. It nevertheless casts doubt on the use of ‘one-sided methods’ (para. 6.58), and affirms that ‘[i]t should be recognized that the identification of reliable comparables in many cases involving intangibles may be difficult or impossible’ (para. 6.143). This points clearly to greater use of profit split but this is not stated explicitly (Picciotto, 2013b). Nor is there any attempt to try to systematize the profit-split method, to make it less arbitrary, as was suggested in the 1993 report on the proposed US regulations, and the additional proposals published in December 2014 make only small and cautious moves in this direction.
A similar trend towards complexity can be seen in many of the reports emerging after the first year of the Action Plan. These include elegant but complex proposals to deal with ‘hybrid mismatch arrangements’ (exploiting the different treatment between countries of the legal form of either entities or transactions, or both) and a sophisticated but also complex approach for applying an ‘economic substance’ test to tax incentives offered by countries, such as the ‘patent box’, considered to be ‘harmful tax practices’. Yet, as shown in a critique of these proposals produced by an activist group, simpler and more effective solutions could be found based on treating TNCs as unitary firms, in line with economic and business reality (BEPS Monitoring Group, 2014).
The patch-up approach adopted to reform of the rules will inevitably increase uncertainty and conflict, both between states and enterprises and among states in asserting tax claims. This can also already be seen, notably in the proposals announced by the UK government for a diverted profits tax. The draft legislation, extending to some 30 pages, adopts a combination of detailed rules worded in convoluted style, with general principles, to invalidate arrangements, which it is ‘reasonable to assume’, are designed to avoid tax. Although considerable skill has been applied to trying to ensure that they are compatible with both EU and international tax law, the provisions both go beyond and in some respects conflict with the proposals emerging from the OECD consensus (Picciotto, 2015).
Complexity, Consensus and Conflict
What can be seen from this process is the great difficulty that the OECD tax specialists have had in maintaining a consensus on the applicable rules and even more in expressing such a consensus clearly. The clear and simple language of the treaty articles can command consensus because they are susceptible of widely differing interpretations. Their capacity to have a fairly stable and settled meaning depends on the cohesiveness of the members of the ‘interpretative community’ to whom they are addressed, for example, habitus to use Bourdieu’s term. In the international tax arena, this began to break down as TNCs developed tax planning strategies taking advantage of the separate entity principle, tearing open a fissure from the fundamentally flawed treaty rule. This revealed the inappropriateness of the rule to deliver an acceptable division of taxing rights between residence and source countries.
In such a context of conflict, it is understandable that the reaction should be to elaborate additional rules. A more effective response would have been to remedy the flaw, by revising the basic principle. There was some support in the 1980s for a revision that would have moved away from the separate entity principle and towards a unitary approach to TNCs (Langbein, 1986) but also significant opposition. Even today, although the OECD has been mandated to ‘ensure that profits are taxed where economic activities occur and value is created’ (G20, 2013), a move towards ‘formulary apportionment’ has been ruled out. In the past the result, as outlined above, was the formulation of the transfer pricing guidelines to include five accepted methods, including profit split. Yet this elaboration was also obfuscation. An apparent consensus was maintained, by the insistence that all these methods comply with the ALP, and indeed this mantra continues to be repeated today, with a fervency that betrays a lack of confidence. The likely result of the current BEPS project will be further elaboration and obfuscation.
The form taken by the elaboration has also been complex, due to the lack of clarity of the status of the additional rules and the nature of their interactions. It seems that when there has been insufficient consensus to amend a treaty article, the commentary has been expanded. More often the preference has been to produce ‘reports’, although appending them to the ‘full version’ of the model treaty gives them some authority and some are also referenced in the commentary. Only for transfer pricing, and in response to a major conflict, have guidelines been formulated. As discussed in ‘Types of Complexity’ section, this type of elaboration leaves scope for skilful navigation and hence helps to maintain at least a loose cohesion among interpretations.
Cohesion can also be maintained by limiting the membership of the interpretative community and trying to ensure that they are like-minded. One method, of course, is simply to keep deliberations secret, which was the route chosen in the 1960s and 1970s. However, this may be hard to maintain without losing legitimacy, especially for rules intended to have a global scope, and which have considerable importance. This perhaps explains the shift to technical complexity, which effectively limits access to the interpretative community to those able to invest in learning the arcane terminology and linguistic techniques familiar to that group. This can be self-reinforcing, that is, if the language is one developed and refined by a closed group, entry can be made difficult and expulsion can be used as a sanction against those who attempt to open up the possibility of different interpretations. Thus, today, a practitioner who dares to question the ALP or commend the possibility of a unitary approach or formulary apportionment is likely to experience a dearth of clients and of invitations to professional forums.
It has been argued that effective tax law can best be achieved by a combination of general principles and more detailed specific rules, expressed as subsidiary or illustrative of the overarching principles (Braithwaite, 2003). This approach has much to commend it, from a technical perspective. However, as with much of the work on responsive regulation, it tends to overlook or underestimate the important power dynamics involved in struggles over interpretation. If the meaning of compliance is constructed in ‘regulatory conversations’, to use Julia Black’s phrase, much depends on who is allowed to speak, treated with respect, or accepted as authoritative.
Furthermore, the stress on ‘cooperative compliance’ may result in the pattern familiar to sociolegal studies of ‘regulatory capture’, when regulators surrender their coercive enforcement powers in favour of consensual solutions. Gregory Rawlings identified such an outcome as resulting from the earlier OECD initiative against tax avoidance, under the banner of ‘harmful tax practices’, which was (and is still being) pursued through a Global Forum on Transparency and Exchange of Information for Tax Purposes. Aimed at dismantling the ‘offshore secrecy system, it became a laborious effort to improve bilateral exchange of tax information and the Forum enlisted as members all the tax havens and secrecy jurisdictions to participate in the process of standard setting and ‘peer review’ of compliance, in effect bringing foxes in to help guard the hen coop (Rawlings, 2007). 11 Similarly, de Cogan (2015) points to the dangers that cooperative compliance can become embedded in a network of professional relationships constituting a closed interpretative community. This is certainly characteristic of the international tax arena, where the ‘revolving door’ is rife. For example, in 2011, the international law firm Baker McKenzie recruited to its Paris office both Caroline Silberztein (after 10 years as head of the OECD’s transfer pricing unit) and Mary Bennett (head of its tax treaty unit, who had also previously been at Baker McKenzie) (Stewart, 2011); whilst Jeffrey Owens, who built the OECD’s tax work over four decades from a handful to well over 100 professionals, on retirement in 2012 joined Ernst & Young (Owens, 2012).
In recent years, in contrast to the secretive deliberations of two decades ago, the OECD has chosen transparency and consultation in the process of norm formulation and interpretation. Yet this small opening has been counterbalanced by the shrouding of that process in the opacity of regulatory complexity and arcane technical language. It remains to be seen whether the social and economic pressures that have resulted in the politicization of the issue of corporate tax avoidance will break open the carapace of complexity constructed by the international tax technocrats.