Absa Bank Ltd and Another v Commissioner for SARS
The Constitutional Court’s judgment in Absa Bank Ltd and Another v Commissioner for the South African Revenue Service is one of the most significant South African tax decisions in recent years, particularly in relation to the interpretation and operation of the General Anti-Avoidance Rules (“GAAR”) in Part IIA of the Income Tax Act 58 of 1962 (“the Act”).
This is not merely because SARS succeeded but because the majority judgment materially expands the practical reach of the GAAR in three respects:
- First, by adopting an expansive and objective conception of who may constitute a “party” to an avoidance arrangement;
- Second, by endorsing a robust, substance-orientated conception of “tax benefit”; and
- Third, by confirming that the “sole or main purpose” enquiry under section 80A is objective rather than subjective.
At a factual level, the case concerned Absa Bank Ltd’s (“Absa”) investment of approximately ZAR1.9 billion in preference shares issued by PSIC3, within a broader structured finance arrangement associated with the Macquarie Group. SARS’ case was, in essence, that the structure was designed so that income which, in economic substance, was interest generated through a chain of downstream entities and Brazilian government bond transactions, was transformed into returns ultimately received by Absa in the form of exempt dividends. On SARS’ version, the interposition of various entities and transactional steps served to alter the juridical form of the return without altering its essential economic character. SARS accordingly invoked the GAAR and sought to re-characterise the dividend income in Absa’s hands as taxable interest.
Two principal issues came before the Constitutional Court. The first was whether Absa could properly be regarded as a “party” to the relevant avoidance arrangement under section 80L of the Act, notwithstanding its contention that it lacked knowledge of the downstream features of the structure. The second was whether Absa had obtained a “tax benefit” as contemplated in the GAAR, given that the amounts it received were, in formal legal terms, dividends on preference shares.
The majority judgment
The majority judgment, authored by Majiedt J, resolved both issues against Absa.
Whether Absa was a “party” to the arrangement
On the first issue, the majority adopted a deliberately broad and functional interpretation of the concept of being a “party” to an arrangement. The court held that participation for purposes of Part IIA is not dependent on proof that the taxpayer subjectively knew every element, step, or mechanism of the overall structure. Instead, the enquiry is objective: one asks what the taxpayer did, what role it performed, and whether, viewed in context, it formed part of the arrangement that produced the impugned tax result.
This is an important interpretive move. It shifts the focus away from subjective states of mind and towards objective involvement in a composite arrangement. On that approach, a taxpayer may be a participant in an avoidance arrangement even if it did not know the full architecture of the arrangement or the precise means by which the downstream entities generated or redirected the economic return. The majority was plainly concerned that a narrower interpretation would materially undermine the efficacy of the GAAR by permitting sophisticated taxpayers to insulate themselves through compartmentalisation of knowledge or by structuring transactions so that ultimate beneficiaries remained formally ignorant of critical downstream features. In the majority’s view, the statute cannot sensibly be interpreted in a manner that rewards deliberate ignorance or wilful blindness.
The practical implication of this aspect of the judgment is considerable. It means that in complex structured finance and cross-border arrangements, a taxpayer cannot necessarily avoid the reach of the GAAR merely by showing that it was not informed of all relevant steps in the wider transactional design. Objective participation in the arrangement may suffice.
The concept of a “tax benefit”
On the second issue, the majority also found that Absa had obtained a tax benefit. Here the court endorsed an analytically rigorous approach to section 80A and related provisions. The correct enquiry is not confined to the immediate legal form of what the taxpayer received, but requires the court to identify and strip out the tax-avoidant features of the arrangement and then ask what remains.
The majority formulated the test in “but for” terms, holding that the correct enquiry is whether, but for the tax-avoidant features and the “dressing up” of the transaction, a tax liability would have arisen. Applying that approach, the majority concluded that once the avoidance features were stripped away, the return received by Absa was properly revealed as interest rather than a dividend. On that footing, Absa had indeed obtained a tax benefit in that the arrangement had caused what was, in substance, taxable interest to assume the form of exempt dividend income.
This aspect of the judgment is particularly significant because it confirms that the “tax benefit” enquiry under the GAAR is not exhausted by asking how the receipt is classified under ordinary charging provisions if one takes the transactional form at face value. Rather, the GAAR permits a deeper enquiry into whether the legal form of the taxpayer’s receipt is itself the product of avoidance steps designed to secure a more favourable fiscal outcome. The majority’s reasoning is therefore strongly aligned with the anti-avoidance logic of Part IIA: If the arrangement’s structuring features produce a mismatch between economic substance and tax character, those features may be disregarded in determining whether a tax benefit has been obtained.
The majority further held that section 80B empowered SARS to determine the tax consequences of an impermissible avoidance arrangement in relation to “any party” to that arrangement, including Absa as the ultimate economic beneficiary. SARS was therefore not confined to pursuing intermediate entities in the structure, particularly where those entities may have had little commercial substance or limited practical significance apart from their role in the arrangement. This reinforces the breadth of the remedial discretion available to SARS under the GAAR and confirms that the fiscus may proceed against the party who ultimately enjoyed the fiscal advantage, rather than only against entities closer to the transactional mechanics.
The “sole or main purpose” requirement
The Constitutional Court was also required to consider the jurisdictional threshold in section 80A, namely whether there existed an “impermissible avoidance arrangement”. Section 80A requires, among other things, an arrangement that results in a tax benefit and whose sole or main purpose is to obtain that tax benefit, coupled with the requisite abnormality or misuse features specified in the section.
The court held that the “sole or main purpose” enquiry is objective. This is a critical holding. The majority reasoned that section 80G strongly supports this interpretation because it places the burden on the party obtaining the tax benefit to show that, reasonably considered in the light of the relevant facts and circumstances, the obtaining of a tax benefit was not the sole or main purpose of the arrangement. That language, the court held, is directed not at the taxpayer’s asserted subjective intention but at an objective evaluation of the surrounding facts.
The majority treated this as a marked departure from the old section 103(1) regime, under which the courts had generally approached purpose as a subjective enquiry into the taxpayer’s state of mind. Under Part IIA, by contrast, the emphasis is on objective commercial and factual indicators. The relevant question is not simply what the taxpayer says it intended, but what the arrangement, viewed reasonably and in context, reveals as its prevailing purpose. This substantially strengthens SARS’ position in GAAR disputes, because it reduces the forensic value of self-serving testimony as to commercial motive and increases the significance of structural features, pricing, risk allocation, circularity of funds, lack of commercial substance, and the mismatch between economic outcome and legal form.
The dissenting judgment
Rogers J delivered the sole dissent. Although dissenting, his judgment is careful, substantial, and likely to remain influential in future GAAR litigation, particularly where questions of fairness, legal certainty, and the limits of objective attribution arise.
The “party” issue
On whether Absa was a “party” to the arrangement, Rogers J took a materially narrower view than the majority. In his analysis, one cannot sensibly be said to be a party to an arrangement of which one is unaware. Knowledge, in his view, is inherent in the ordinary meaning of participation. He illustrated this by analogy: a person who gives another a lift to a place where that person later commits murder is not, without more, a party to the murder. The point of the analogy was that causal or factual involvement in a chain of events is not the same as participatory involvement in the relevant arrangement.
For Rogers J, the concept of an “arrangement” ordinarily connotes at least some measure of understanding, consensus, or meeting of minds among those said to be parties to it. He expressed concern that the majority’s interpretation stretched the notion of participation too far, thereby introducing unfairness and legal uncertainty. On that approach, taxpayers could be exposed to serious fiscal consequences in relation to arrangements whose existence, design, or salient features were unknown to them. He also recognised the commercial reality that in institutional finance and structured transactions, counterparties are not always apprised of all downstream elements of a larger structure, and that such non-disclosure is not inherently suspicious. While he accepted that wilful blindness could in an appropriate case supply the necessary link, he pointed out that SARS had not advanced its case on that footing.
The “tax benefit” issue
On the tax benefit issue, Rogers J drew a sharp distinction between a tax benefit and an economic advantage. In his view, Absa received dividends on preference shares, and the Act ordinarily treats such dividends as exempt. The legal character of Absa’s receipt was therefore determined by the instrument it held, not by the downstream transactions undertaken by other entities in the structure. On that reasoning, the mere fact that the wider arrangement may have generated an economic advantage for Absa did not mean that Absa itself obtained the relevant “tax benefit” for purposes of the GAAR.
Rogers J was particularly troubled by the majority’s readiness to disregard the legal form of the instrument in Absa’s hands by reference to transactions further down the chain of which Absa said it was unaware. In his view, even if one notionally removed the Brazilian bond and swap transactions, that would not convert Absa’s dividend income into interest, because Absa’s entitlement remained one arising under preference shares. He therefore considered that the tax benefit, if any, was more naturally located elsewhere in the structure, among the entities directly involved in the impugned transactions.
That led him to a broader concern of principle. In his view, SARS was effectively seeking to tax a taxpayer who may have enjoyed an economic upside from a wider arrangement, but who had not itself obtained the tax benefit that the GAAR is directed at. He regarded that as problematic both doctrinally and in terms of fairness. He also observed that SARS had potentially available remedies against the entities more directly implicated in the generation of the alleged tax advantage.
Broader significance
The judgment is a landmark in several respects.
First, it materially enlarges the practical scope of the GAAR by confirming that taxpayer knowledge is not necessarily required before a person may be treated as a “party” to an avoidance arrangement. In future disputes, the battleground is likely to shift toward objective indicators of participation, functional role, and economic integration into the impugned structure.
Secondly, the court’s treatment of “tax benefit” strengthens SARS’ ability to attack arrangements in which legal form obscures economic substance. The majority’s formulation of the “but for” enquiry is likely to feature prominently in future cases involving hybrid instruments, preference share structures, funding arrangements, and cross-border transactions where the tax character of a receipt is said to have been engineered through intermediate steps.
Thirdly, the judgment decisively confirms that the “sole or main purpose” requirement in section 80A is objective. That is perhaps the most systemically important aspect of the case. It marks a clear departure from jurisprudence under the old section 103(1) and confirms that modern South African GAAR analysis is structurally closer to a substance-and-purpose enquiry grounded in objective facts than to a subjective motive test.
Finally, the decision underscores the breadth of SARS’ remedial powers under section 80B. SARS is not required to pursue only those entities at the centre of the transactional machinery. It may proceed against any party to the arrangement, including the party that ultimately enjoys the economic and fiscal result.
Final Comments
Absa Bank Ltd and Another v C:SARS is a foundational judgment in contemporary South African tax law. The majority has given the GAAR a robust and purposive interpretation that materially enhances its effectiveness against sophisticated tax planning. In doing so, the court has broadened the meaning of participation, endorsed a substance-focused understanding of tax benefit, and confirmed the objective nature of the purpose enquiry under section 80A. At the same time, Rogers J’s dissent offers an important counterweight grounded in legality, doctrinal precision, and fairness. For that reason, although the majority now states the law, the dissent is likely to remain highly relevant in future cases testing the outer limits of the GAAR.
