A dramatic change in international tax law is underway, one which since 2016 is altering the dense network of over 4,100 bilateral tax treaties (double taxation and information exchange agreements) through a complex international treaty – the OECD’s Multilateral Convention to Implement Tax Treaty Related Measures to Prevent Base Erosion and Profit Shifting (MLI) – that changes the terms of the bilateral treaties without “amending” each individually.
Although adoption of the MLI requires ratification by participating countries, the changes to existing bilateral treaties facilitated by the MLI do not. As the OECD explains, the MLI “sits alongside” existing treaties rather than replacing them.[i] The result is a significant change in international tax law. Not only does the MLI alter treaty provisions that are applied to millions of cross-border financial transactions, it also represents a fundamental move away from the autonomy of countries to create international tax law. Crucially, these dramatic changes are not made as the result of a global consensus on principles of international taxation but in order to coerce developing economies into an OECD-designed programme whose purpose is to protect OECD members’ tax systems rather than advance broader interests. Allowing a few of the world’s wealthiest economies to dictate the global financial infrastructure is a wrong turn for global governance.
Tax Treaty Networks
Although there were scattered double tax agreements prior to World War II, today’s dense global tax treaty network is a recent phenomenon. As late as 1970, fewer than 300 treaties comprised the global tax treaty network. The United Kingdom and France concluded treaties with their former colonies and were core of the largest networks (reflected as betweenness centrality in network terms).[ii] Runners up were Norway, Denmark, Switzerland, and Japan. See Figure 1. By 2000, the global treaty network had evolved into a significantly different configuration, with a dense core of jurisdictions interlinked by multiple treaties and a periphery of mostly developing countries with just a few treaty ties, primarily to one or more core jurisdictions. See Figure 2.
Double tax treaties originally had the goal of preventing double taxation. They did so by allocating the right to tax covered transactions amongst the signatories. As new treaties expanded the global treaty network, differences in these treaties provided /created opportunities for taxpayers to reduce or even to eliminate their tax obligations. A well-known example was the successful exploitation by Curaçao of the US-Netherlands tax treaty (which had been extended to the Dutch Caribbean territories) to allow US firms to access the Eurodollar market without paying US withholding tax on interest payments by routing the transaction through a Curaçao subsidiary.[iii] To stop what anti-tax-competition campaigners labelled “treaty shopping,” the United States cancelled the treaty’s extension to the Dutch Caribbean territories. This brute force approach worked to rein in a small jurisdiction like Curaçao (although the abrupt cancellation caused a crisis in the corporate bond market) but it was inadequate to address more complex financial strategies involving larger jurisdictions with whom a tax treaty was desirable.
Tax authorities struggled to cope with the growing sophistication of tax avoidance strategies that took advantage of specific treaty provisions and were not limited strategies involving ‘tax havens’. For example, by its 1997 edition, Milton Grundy’s definitive Offshore Business Centres: A World Survey featured extensive discussion of routing interest, royalties, and dividends through otherwise high tax “stepping stone” jurisdictions to save on taxes through treaty networks. To combat this “treaty abuse,” high tax jurisdictions (especially OECD members) increased the complexity of their tax laws to restrict the availability of transactions the tax authorities viewed as illegitimate. They further added domestic anti-abuse provisions and incorporated them into their tax treaties.
The growing complexity of tax law rendered it unintelligible to most non-specialists (and, we suspect, to many specialists as well).[iv] The anti-abuse provisions created rule of law and democratic accountability problems with no clear guidance on what is legal and what is not. Both moves are costly to the global economy by increasing uncertainty surrounding transactions. And as the ever-expanding reach of these provisions demonstrates, disallowing transactions ex post via an added layer of domestic law was not sufficient for tax authorities to eliminate transactions which fit the literal terms of tax treaties but of which they disapproved.
Cutting The Gordian Knot
By 2010, the global double tax agreement network had grown to more than 3,800 bilateral double tax and information exchange treaties. The oldest of these treaties still in force were between the United Kingdom and four of its now former colonies (St. Kitts and Nevis, Sierra Leone, Belize, and Montserrat, all signed on December 19, 1947). France and Germany, countries closely identified with the OECD’s push for limiting tax competition, had treaty networks of 149 and 116 treaties respectively. Even for them, renegotiating their entire treaty networks to incorporate limitation-of-benefits, information-exchange, and provisions to facilitate the restriction of treaty benefits to approved uses would be a costly and complicated undertaking. Once the OECD launched its Base Erosion and Profit Shifting (BEPS) initiative, which sought to impose a rich-country (and overwhelmingly Eurocentric) set of tax rules on the global economy, the existing hodge-podge of treaty provisions was even more problematic for the goal of forcing global tax into a uniform, OECD-dictated framework of complex rules.
Regularly rewriting the current global treaty network of 3,333 double taxation agreements (and 784 tax information exchange agreements)[v] to take into account the evolving BEPS provisions intended to force recognition of permanent establishments, control “abuse”, etc. would be a challenge even for countries with well-resourced ministries and a practically impossible burden for most developing economies. The OECD’s solution was the MLI. While the drafting of the MLI has an inclusive veneer with participation by 100 countries, this dramatically overstates non-OECD members’ involvement. The most important substantive decisions were made earlier in crafting the BEPS provisions, turning the drafting of the MLI itself into a mostly ministerial activity.[vi] Further, subsequent commentary has emphasised the considerable extent to which the “soft law” of the BEPS final reports serves as the key interpretive source for the MLI, reinforcing the idea that the MLI is a mere translator of the substantive decisions made in the crafting of the OECD’s BEPS project.[vii]
Once a country ratifies the MLI, the MLI’s text on key tax provisions modifies – without requiring renegotiation or ratification of the existing tax treaties covered. Although the OECD and other MLI-proponents stress that the MLI does not “directly amend” treaty text, even the OECD concedes that the MLI “will be applied alongside existing tax treaties, modifying their application in order to implement the BEPS measures”.[viii] The MLI thus cuts through the Gordian knot of the existing treaty network to, de facto if not de jure, replace the relevant existing treaty text with the MLI provisions. For example, Article 7(2) of the MLI allows treaties to add only the MLI principal purpose test to treaties without one or replace an existing provision with the MLI provision.[ix] It thus effectively will harmonise international tax law over time. Jurisdictions that adopt the MLI can designate to which of their treaties it applies and make reservations about some, but not all, MLI provisions. If both parties to a tax treaty designate the treaty as covered (and have made identical decisions with respect to a particular provision for the optional ones), the MLI applies to the treaty.[x]
Using OECD data on adoption of the MLI, we have calculated some simple statistics giving a broad sense of its impact thus far. As Table 1 shows, this impact is significant.
Table 1 - MLI Adoption by Treaty Pairs
Number of Treaty Pairs
MLI adopter – MLI adopter treaties applying the MLI
MLI adopter – MLI adopter treaties not applying the MLI
MLI adopter – Non-MLI adopter treaties
To put these numbers into context, we calculate that 78.74 per cent of treaties between MLI adopters, and 61.47 per cent of MLI adopters’ total treaties, have the MLI applied. Since most major economies (save only the United States) have adopted the MLI, the per cent of the world economy covered by the MLI is even larger than these numbers suggest.
Isn’t this a good thing? Wouldn’t globally harmonised international tax law make international transactions simpler? We think not, for three reasons. First, while the MLI is a clever means of circumventing the burden of amending thousands of bilateral agreements, making an end run around the process of negotiating treaties is not inherently good. Treaties differ for many reasons, including the treaty partners’ accommodation of features unique to particular relationships. Quashing diversity in treaty arrangements is not a step to be taken lightly.
Second, there is an important gap in democratic accountability caused by this manoeuvre. It is undoubtedly inconvenient to have to negotiate an individual treaty and undergo the legislative scrutiny often required by ratification but that inconvenience is the price of democratic legitimacy and assuring domestic stakeholder buy-in. Particularly given the ‘Hotel California’ design of the MLI – one can check in but it is extremely difficult, if not impossible, to check out – this is a high price.
Third, paying that price might be worthwhile if it was in pursuit of an important, widely shared goal but here it is a price paid to coerce developing economies into an OECD-designed program to protect OECD members. Whether implementing the OECD’s BEPS project is a desirable end state for international tax law or not, enabling a small number of rich countries to coerce the rest of the world into accepting their rewiring of the world’s treaty network without substantive involvement of the vast majority of the world’s jurisdictions sets a disturbing precedent.
The authors gratefully acknowledge the research and programming assistance of Jakub Bartoszewski and Shubham Mahendra Mundada and the support of the Mays Innovation Research Center at Texas A&M University.
[i] See OECD, MLI Frequently Asked Questions available at https://www.oecd.org/tax/treaties/MLI-frequently-asked-questions.pdf. The MLI uses a novel form of treaty-making that does not technically replace existing agreements but alters their application by concluding a multilateral treaty to “update” bilateral ones. As a result, it significantly complicates treaty interpretation since those seeking guidance must consult both the treaty and the MLI.
[ii] Betweenness centrality measures “the extent to which a node lies on paths between other nodes” and is calculated as:
where is 1 if node i lies on the shortest path from s to t and 0 if it does not or there is no such path. Mark Newman, Networks (2nd ed.) 173-74 (Oxford: Oxford University Press, 2018).
[iii] Craig M. Boise & Andrew P. Morriss, Change, Dependency, and Regime Plasticity in Offshore Financial Intermediation: The Saga of the Netherlands Antilles, 45 Tex. Int’l L. J. 377, 409-414 (2009) available at https://tinyurl.com/6wd42duv
[iv] For example, consider this description of the rewriting of one set of provisions in the 2007 British Income Tax Act from Giles Clark’s definitive treatise on offshore tax planning:
“It has to be said that the rewriting of the transfer of assets code in the ITA 2007 has compounded rather than diminished the difficulties of construction and thus the uncertainty. In part this is because the rewriting process, by unpicking and reassembling the legislation, has made explicit ambiguities and inconsistencies which previously were latent. But also, the very process of unpicking and reassembly has resulted in subtle changes which, with legislation as complex as this, can have the result of changing meaning or rendering it uncertain”.
Giles Clarke, Offshore Tax Planning 930-31 (26th ed. ) (London: Lexis Nexis/Tolley, 2019). Clarke identifies dozens more instances where the complexity of tax law is dramatically increased by anti-avoidance measures.
[v] We calculated these numbers using the IBFD treaty database.
[vi] The MLI’s draftsmanship leaves quite a bit to be desired. The chapters in The OECD Multilateral Instrument for Tax Treaties: Analysis and Effects (Michael Lang, Pasquale Pistone, Alexander Rust, Josef Schuch, & Claus Staringer, eds.) (Wolters Kluwer, 2018), the result of a conference held in 2017 at the Vienna University of Economics and Business, identify multiple points at which the MLI’s language is unclear and opaque. There are also structural issues. For example, as Josef Schuch and Jean-Phillipe Van West note in their chapter in the above work, Authentic Languages and Official Translations of the Multilateral Instrument and Covered Tax Agreements, the MLI is only in French and English, making its application to treaties in other languages complicated at best. Schuch and Van West estimate this affects less than 10 per cent of tax treaties; still, it is hard to know whether this linguistic narrow-mindedness on the part of the OECD reflects arrogance or laziness or a combination of the two.
[vii] See, e.g., Andreas Langer, The Legal Relevance of the Minimum Standard in the OECD/BEPS Project, in Lang, et al., supra note 6.
[viii] OECD, Explanatory Statement to the Multilateral Convention to Implement Tax Treaty Related Measures to Implement Base Erosion and Profit Shifting, available at https://tinyurl.com/6wd42duv
[ix] Sriram Govind & Pasquale Pistone, The Relationship Between Tax Treaties and the Multilateral Instrument in Lang, et al., supra note 6.
[x] This is excellent news for lawyers and tax consultants, as determining the relevant law governing an international transaction now requires careful reading of both the treaty text and the MLI to determine the relevant language. While some countries have published unofficial consolidated texts taking the MLI’s application into account, these are not binding and an independent analysis of the impact of the MLI on any particular treaty will be required in each case.
Andrew P. Morriss
Andrew Morriss is Professor of the Bush School of Government & Public Service and School of Law at Texas A&M University. Prior to this position, he was the Dean of the Texas A&M School of Innovation, the Dean of the Texas A&M School of Law, the D. Paul Jones & Charlene A. Jones Chairholder in Law at the University of Alabama, the Ross & Helen Workman Professor of Law at the University of Illinois, and the Galen J. Roush Chair in Law at Case Western Reserve University.
Charlotte Ku is Professor and Director of Global Programs at Texas A&M University School of Law. Her specialties include international law and global governance. She has previously served as Professor and Assistant Dean for Graduate and International Legal Studies at University of Illinois College of Law, Acting Director of Lauterpacht Research Centre for International Law at University of Cambridge, and Executive Director and Executive Vice President at American Society of International Law.