It is interesting to note that there was a split judgment in this case, in other words, the judgment comprised both a majority and a dissenting judgment. While the majority judgment constitutes binding precedent, the split judgment illustrates the intricacies in the interpretation and application of tax legislation to complex, multi-party arrangements, making both judgments worth understanding.
The majority judgment per Majiedt J
The court held that the arrangement fell squarely within the GAAR provisions, in that its main purpose was to obtain a tax benefit, and it lacked commercial substance. The jurisdictional requirement for the GAAR is objective, it is sufficient that such an arrangement existed, regardless of whether the taxpayer personally secured the benefit. Under the previous version of the GAAR, the purpose test was a subjective test, i.e., the person being challenged under the GAAR had to have had the subjective purpose of tax avoidance. It has been a matter of ongoing debate over the years whether the current version of the GAAR has shifted the purpose test to that of an objective test. The court has put this debate to bed by decisively ruling in favour of an objective test.
As to the question of who is a “party” for purposes of the GAAR, the court adopted a broad, purposive interpretation of the word “party” under section 80L and contended that participation does not require knowledge of every step of the arrangement. It is enough that a person’s conduct forms part of the chain of transactions constituting the arrangement for the person to be a “party” as defined. The court rejected the argument that only those with full knowledge can be parties, as such a narrow view would undermine the GAAR’s purpose and enable wilful blindness. The statutory language used to define “party” as “any person” who “participates or takes part” in an arrangement indicates a focus on involvement, not mental state.
On dealing with the question of whether Absa received a “tax benefit”, the court found that Absa did in fact obtain a tax benefit within the meaning of the GAAR. Stripped of the arrangement’s avoidance features, Absa’s returns would have been taxable interest, not tax-exempt dividends. The GAAR empowers SARS to disregard artificial steps and re-characterise the substance of the arrangement, ensuring that the tax benefit is taxed in the hands of the ultimate recipient. The court held that the proper “but-for” test is whether, absent the avoidance features, a tax liability would have arisen for Absa.
The appeal was dismissed with costs, confirming SARS’ assessments.
The dissenting judgment per Rogers J
The dissenting judgment highlighted concerns, inter alia, around the reach of the GAAR, in the majority judgment.
Rogers J held that to be a “party” to an arrangement, a taxpayer must know about and intend for the arrangement to take place. In other words, one cannot participate in an arrangement that one does not know exists. The court distinguished between knowledge of the arrangement and knowledge of its tax consequences, accepting that the latter is not required.
Rogers J further held that, on the facts, the tax benefit occurred at the level of D1 Trust and PSIC4, not at Absa’s level, and that Absa only received an economic benefit, not a tax benefit. The proper counterfactual is to remove the impermissible feature (the Brazilian interest swap), which would have left Absa’s dividends exempt in any event. The court averred that SARS’ remedial powers should be directed at the party that obtained the tax benefit. It was further noted that, in other jurisdictions, only the party obtaining the tax benefit is targeted.
In essence, the minority judgment reflects a concern of judicial overreach by the majority.
