At The Tax Institute’s March 2019 National Convention, in Hobart, Professor Graeme S Cooper at the Sydney University Law School, presented his paper, entitled: Application of the MLI (or Multi-Lateral Instrument).



1 Background

This paper is about the development and operation of the Multilateral Convention to Implement Tax Treaty Related Measures to Prevent Base Erosion and Profit Shifting (‘MLI’), a multilateral treaty signed by Australia in June 2017, and which began operation between Australia and 6 of its treaty partners on 1 January 2019.

While it is a multilateral instrument, its effect is to amend many of Australia’s bilateral tax treatieswhich makes the MLI probably the most complicated and annoying amending document ever created. It is complicated because the mechanisms used to amend the text are prolix and very imprecise. Sometimes existing text is exercised and replaced by new text; sometimes existing text and new text will have to be read together and their inconsistencies reconciled. And it is annoying because, for the most part, the changes being made by the MLI are trivial, often to the point of total irrelevance. The one exception to this is the provisions in Part IV, and Australia’s decision to take on board the policyof accepting mandatory and binding arbitration of unresolved cross-border disputes. This is a major change of policy.

The MLI is a curious instrument in international law terms. Unlike the OECD’s Model Convention on Income and on Capital it is not merely a template serving as the basis for other operative documents; it is a document with full force and effect in international law for those countries that have signed it. And while it is a multilateral instrument – as at February 2019 it had been signed by over 80 countries and several regions such as Guernsey, Jersey, Isle of Man and Hong Kong (China) – it is not really a multilateral instrument in the same sense as (say) the OECD’s Convention on Mutual Administrative Assistance in Tax Matters because its effect is to modify and amend other documents – it has no coherence as a stand-alone document because its principal effect is to introduce amendments into existing bilateral income tax treaties.

The design. That model is a key design feature of the MLI. The OECD wanted quick implementation to entrench the proposals with implications for tax treaties (rather than domestic law) that would arise out of the BEPS project. Modifying the OECD Model was easily achievable but that has no direct effect on existing treaties, only possibly on future treaties. And renegotiating each of the 3,000+ individual existing bilateral tax treaty was not a promising option. The BEPS Action Plan [OECD,Action Plan on Base Erosion and Profit Shifting (Paris, OECD, 2013) page 23] said some new paradigm was needed:

Changes to the OECD Model Tax Convention are not directly effective without amendments to bilateral tax treaties. If undertaken on a purely treaty-by-treaty basis, the sheer number of treaties in effect may make such a process very lengthy, the more so where countries embark on comprehensive renegotiations of their bilateral tax treaties. A multilateral instrument to amend bilateral treaties is a promising way forward in this respect (page 24)

But just what that multilateral instrument would look like was not self-evident. The goal of quick implementation might have been achieved by, say –

  • creating a new multilateral BEPS-compliant income tax treaty which willing countries could sign – the model would be multi-jurisdiction income tax treaties like the Nordic treaty; or
  • creating a new definitive “BEPS-only treaty” which basically operated as a stand-alone treaty setting out a fixed and comprehensive set of obligations emerging from some or all of the 14 other outputs of the BEPS project. The effect would be –
    • to override domestic law and create new international law obligations where none already existed in treaties (eg, to entrench the outputs from BEPS Action 3, 4 or 5), and/or
    • to complement, or supplant, inconsistent provisions in existing treaties of every country which signed on the basis of lex posterior.

The model would be something like the Convention on Mutual Administrative Assistance in Tax Matters which operates alongside (or instead of) existing administrative provisions in bilateral treaties (such as art 26 on exchange of information) and adds new obligations into domestic law (such as provisions dealing with mutual assistance in collection which are not included in most of our treaties).

The problem with the first model is that no-one would sign such a document; even though 95% of the text of all bilateral tax treaties is very much the same, countries are too wedded to the idea that they have negotiated bespoke deals with individual treaty partners, and they do not want to have these deals supplanted. The problem with the second model is similar: some countries did not want new obligations involving matters that were not fixing bits of treaties and, for provisions that were already in treaties, they apparently wanted choice, either to insist on retaining existing provisions for some orall of their existing treaties, or else to ‘cherry-pick’ the bits of the BEPS agenda they were willing to take on board and those they were not. In this regard, the decision to have provisions dealing with binding arbitration of tax disputes included in the document (Part VI of the MLI) was clearly unacceptable to several countries.

So the model for the MLI became a stand-alone instrument, to amend selected provisions of existing bilateral treaties, and with choices and options at almost every step of the process, made by individual countries choosing unilaterally. The effect of the MLI was thus contingent; just whether and how any existing treaty would be affected depends on just how these individual choices intersected.


About the author