MFN dangers: The (potential) unravelling of tax treaty policy
MFN clauses in tax treaties – a useful negotiation tool or an accident waiting to happen?
The dangers of most favoured nation (MFN) clauses have been demonstrated recently in tax cases concerning dividends between South Africa and the Netherlands. One case, heard in the Netherlands, concerned a dividend flow from the Netherlands to South Africa and the other, heard in South Africa, concerned dividend flows from South Africa to the Netherlands. Uniquely, these two cases concerned the same article and MFN clause.
By way of brief background: South Africa had changed its dividend policy to move from a secondary tax on companies (levying tax on resident companies on net dividends paid) to a withholding tax on dividends (that is, a tax on the shareholder receiving the dividend). While the intention had been to bring this change in swiftly, it was soon established that a number of South African tax treaties provided for a nil withholding tax on qualifying interests. The repeal of secondary tax on companies and the introduction of dividends tax was delayed until protocols to these treaties could be concluded.
Naturally, a change to a treaty is a negotiation. Those States moving from nil rates to a South African proposed minimum of 5% would require some concessions in return. In this regard, the Netherlands requested the insertion of an MFN clause (which had been based on past Dutch treaty practice). This MFN clause provided that should South Africa conclude a more favourable rate (than the new 5% withholding) in the future, then such reduced rate would also apply for the Netherlands. This seems to be a logically worded MFN provision ensuring that the Netherlands would not be sacrificing tax revenue only to find another State negotiating a different position with South Africa later. Subsequent to this protocol between South Africa and the Netherlands, South Africa concluded a protocol with Sweden which provided that if any agreement provided for a more favourable dividend withholding rate, then such rate would be applicable to Swedish recipients of South African dividends. At the time this protocol was concluded, there were a number of treaties for which protocols had not yet been concluded which still provided for a nil rate. Most of these were subsequently concluded. However, the protocol with Kuwait was never ratified in Kuwait. As a result, the nil dividend withholding remained applicable. On the assumption that the Kuwaiti protocol would be ratified in Kuwait, the dividend tax was introduced.
It was soon proven in Binding Private Rulings (see BPR 267 and 276) that Swedish shareholders could still obtain the nil dividend withholding through the MFN clause concluded by the protocol as the MFN clause referred to any agreement (and not any agreement in the future). It was therefore argued in the two cases that the Netherlands or South Africa (depending on the dividend direction) should also be entitled to the nil rate as Sweden, through a subsequent agreement (the protocol) had secured a more favourable rate (the Kuwaiti nil rate) and therefore the Netherlands MFN clause applied. Both jurisdictions courts agreed.
So where is the problem? If these were equally developed nations, this may not be of major concern. However, the South Africa–Netherlands relationship is between a developing and developed nation where South Africa receives more investment from the Netherlands than it makes. Thus, the conclusion that the MFN provided for a nil rate would yield significant tax loss to the South African Government.
Yet it would seem that this problem has the potential to expand. Firstly, it does not seem that Kuwait will ratify the protocol, securing the Swedish and Dutch nil rates through the MFN clauses. Secondly, the Ireland-South Africa treaty also includes an MFN clause providing that if any future agreement provides for a lower rate (than the 5% in that treaty), then ‘South Africa shall immediately inform the Government of Ireland in writing through the diplomatic channel and shall enter into negotiations with the Government of Ireland with a view to providing comparable treatment as may be provided for the third State’. These are, of course only negotiations and may end with the rate remaining at 5%. However, if Ireland is successful in the negotiation, this will increase the total States impacted to three (3). Further, if Ireland is successful, the MFN clause in the South Africa-United Kingdom treaty is then triggered which, like the South African treaty with Ireland, requires South Africa to enter into negotiations to provide for comparable treatment.
The above represents a clear unravelling of a change in domestic tax policy to provide for a tax on shareholders on the dividends received and further an unravelling of the protocols concluded in which concessions would have had to be made to achieve the 5% withholding (increased from nil). South Africa may be further pressed to provide for further concessions in the treaties with Ireland and the United Kingdom to try to secure its 5% withholding rate. These would be concessions that it did not foresee and would be tantamount to a further erosion of the South African tax base.
So where did it all go wrong? The adoption of MFN clauses as presented by these nations without a thorough testing of their impact on the tax treaty network appears to be the failing that has unravelled this scheme. Further, the enactment and bringing into force of the domestic withholding tax before ensuring that all affected treaties had been ratified in the other Contracting States represents a further fatal blow to the tax base in South Africa.
Is there a solution? South Africa finds itself in a vice. It is unlikely (politically) that the treaty with Kuwait will be terminated for failure to ratify the protocol. The Netherlands and Sweden have no reason to relinquish the, now, nil rate and Ireland and the United Kingdom may be hard-pressed to justify why they would not pursue a nil rate that other States are able to obtain. So, what now? Does South Africa try to renegotiate with Sweden, the Netherlands, Ireland, and the United Kingdom? Can it rely on domestic general anti-avoidance to pursue any scheme used by companies to restructure to these jurisdictions to receive the nil rate? Does South Africa wait to ratify the multilateral instrument (MLI) in the hope that the principle purpose test (PPT) will resolve the matter (when also read with the preamble)?
Finally, the interaction between MFN clauses and the future application of PPT may yield interesting debates going forward. In the current climate of concluding treaties to not only eliminate double taxation, but also ‘without creating opportunities for non-taxation or reduced taxation through tax evasion or avoidance’, it is unclear whether the issue will be resolved through other means. What is the future of the MFN clause? Is it creating an opportunity for reduced taxation? Can invoking the MFN provision be considered ‘avoidance’?
These are issues to be faced in future treaty negotiations which must be undertaken with due care and to ensure that countries, particularly developing countries, do not unnecessarily (or inadvertently) erode their tax bases. This post supplied a single example of the unravelling of a policy. Certainly more are out there!