On 22 April 2026, the Constitutional Court handed down judgment in Absa Bank Ltd and Another v Commissioner for the South African Revenue Service [2026] ZACC 15 (“ABSA”).
It is the first time our highest court has interpreted the General Anti-Avoidance Rules (“GAAR”) in sections 80A to 80L of the Income Tax Act, 58 of 1962 (“the Act”). In broad terms, those rules allow the South African Revenue Service (“SARS”) to counter an ‘impermissible tax avoidance arrangement’ (as defined in the GAAR) by looking beyond its legal form and taxing the transaction on an alternate basis.
Barely ten weeks later, on 3 July 2026, the Tax Court applied that judgment for the first time in Company AF (Pty) Ltd and Others v C:SARS (IT 76725; IT 76750–76755) (the “RASS” case).
This article explains, in plain terms, what has changed, why it matters, and what the decisions mean in practice. Our view, having considered the judgments carefully, is that the emerging test is fundamentally commercial rather than mechanical: the tax analysis now starts with the real commercial purpose, risk and economics of the arrangement.
A quick refresher: what Is the GAAR and why does it exist?
Almost every business transaction has tax consequences. Some transactions can be implemented in different ways and there is nothing wrong with choosing the most efficient option if it achieves the same commercial outcome. Our courts have confirmed this principle for decades. The GAAR exists for the rare cases where a transaction has been engineered, in whole or in part, to manufacture a tax outcome that has no real connection to what is actually happening commercially.
Where the GAAR applies, SARS can effectively look through the structure and tax the transaction as if it had been done the “normal” way.
Until recently, however, there was limited guidance on how the current version of the GAAR, often described as the modern GAAR and in force since 2006, would be applied in practice. Earlier anti-avoidance provisions had been considered by the courts, but the modern GAAR itself had not been properly tested.
That has now changed three times in under a year: first in September 2025, when the Tax Court decided Mr Taxpayer G v Commissioner for the South African Revenue Service (IT 24502) [2025] ZATC 12 (“Taxpayer G”), then definitively in April 2026 with Absa and, most recently in July 2026, when the Tax Court applied Absa in RASS.
Absa in a nutshell
Absa had invested around R1.9 billion in preference shares issued by a special-purpose company and received dividends on those shares. In the ordinary course, dividends are tax-free in the hands of a South African company. What Absa contended it had no knowledge of was that, downstream, the money it had invested flowed through a chain of entities and ultimately into Brazilian government bonds, generating what was, in substance, interest income.
That interest income was then, as the judgment put it, “swopped” and delivered back to Absa as a tax-free dividend. To complete the picture, the bank’s return was guaranteed regardless of what happened to the underlying investments.
SARS applied the GAAR and taxed the dividends as interest. The matter was heard by the Constitutional Court and Absa lost, with one judge (Rogers J) dissenting in a thoughtful judgment that we think will continue to matter in future cases.
What has ABSA changed?
In our view, the most important shift is the test for whether an arrangement was mainly entered into or carried out, to obtain a tax benefit. The inquiry is no longer driven primarily by what the taxpayer says it intended. It is now an objective assessment of what the arrangement, viewed as a whole and against the relevant facts, shows.
The Constitutional Court has now confirmed that this is an objective test. The question is not only what the taxpayer says its motive was, but what the arrangement itself shows when viewed as a whole: the structure, documents, risk allocation and economics.
Two further points from the judgment are worth understanding:
- A wider range of participants can be caught. The Court rejected the argument that a person must know every step of an arrangement before being treated as part of it. If a taxpayer’s role and contractual rights are structurally tied to the tax outcome, as Absa’s were, limited knowledge of the mechanics elsewhere in the structure will not necessarily be enough.
- The tax benefit is tested by stripping away the artificial features. The Court said the comparison is between the tax position under the arrangement and the tax position that would have existed without those artificial features. In Absa’s case, once those features were stripped away, the return was taxable interest income.
Our view: the commercials must lead the tax
The Constitutional Court did not say that tax-efficient structuring is no longer permissible. Quite the opposite, the Court expressly reaffirmed a principle that has been part of our law since the 1990s, that is, a taxpayer is fully entitled to choose the most tax-efficient way of achieving a genuine commercial result. That is the choice principle articulated in Commissioner for Inland Revenue v Conhage (Pty) Ltd (formerly Tycon (Pty) Ltd) 1999 (4) SA 1149 (SCA) and endorsed in several later judgments. In our view, nothing about that has changed.
What Absa really does is sharpen the lens through which a genuine commercial result is assessed. Under the older, more subjective approach, a taxpayer could get a long way by asserting that a transaction was commercially motivated. Under the new approach, that assertion must be supported by facts that exist independently of the tax outcome.
A transaction with a real, multi-faceted commercial story, one that would make sense to a businessperson even without the tax benefit, is not the kind of case Absa was concerned with. In our view, it should not be troubled by it.
The Tax Court in RASS
We did not have to wait long to see what Absa means in practice. The RASS case concerned seven companies that disposed of their shares in another company. A straight sale of the shares would have triggered significant capital gains tax.
Instead, their advisers designed a four-step structure, namely, the target company declared a dividend of R274.6 million to the sellers on loan account; the buyer subscribed for new shares in the target company to the tune of R280 million; the company used that subscription price to settle the dividend loans; and the sellers then sold their original, now almost worthless, shares to the buyer for a nominal amount.
Applying Absa, the Tax Court held that the GAAR applied. Judged objectively, on the documents, the sequencing and the economics, not on what the witnesses said they intended, the arrangement was reasonably to be regarded as having the tax benefit as its main purpose. The dividend declared on loan account and subsequent subscription served no commercial purpose that a straightforward sale would not have done; their only real function was the tax saving. Stripped of those features, the transaction was simply a sale, and it was taxed as one.
The taxpayers relied heavily on the long-standing choice principle from Conhage and the court expressly accepted that the principle is alive, is part of our law, and was not weakened by Absa. A taxpayer remains fully entitled to arrange its affairs to attract less tax and, where the same commercial result can be achieved in different ways, to choose the route that attracts the least tax.
What the court clarified is what the principle protects a choice between genuine routes to the same commercial result, not the insertion of additional, commercially unnecessary steps whose only job is to reduce tax.
There is one more lesson in RASS. Although the taxpayers lost on the GAAR argument, the 75% understatement penalties were overturned. The court drew a clear distinction between the GAAR analysis and the penalty analysis, which looks at how the taxpayers behaved. These taxpayers had obtained a written opinion from specialist advisers directed at their own facts, had disclosed the arrangement to SARS before any audit, and genuinely believed the structure was lawful. On those facts, they had not behaved unreasonably, even though the structure ultimately failed. A transaction-specific opinion and upfront disclosure, in other words, may remain powerful protection against penalties where the law is unsettled.
What this means in practice
If the test is now objective, the facts must exist and be documented when the transaction is implemented, not reconstructed later. In practice:
- Parties must document the commercial rationale early and the contemporaneous records explaining why a transaction is structured as it is can be critical evidence of commercial purpose.
- Genuine commercial risk must sit with the party purportedly bearing it.
- The more independent, verifiable commercial reasons supporting a transaction, the stronger its position.
- Obtain transaction-specific advice.
- Do not fear legitimate restructurings. Reorganisations and similar transactions remain valid provided they serve a genuine commercial objective.
Ultimately, the focus is on objective commercial reality. The stronger the evidence of purpose, substance and risk, the stronger the defence.
Where to from here?
In our view, transactions with a genuine commercial purpose, supported by sound documentation, remain safe. Those driven primarily by tax outcomes are now more vulnerable to challenge.
Absa and RASS do not make tax-efficient structuring impermissible. They do, however, reinforce a simple question: would the transaction still make commercial sense without the tax benefit?
If you are planning, implementing, or have recently completed a significant transaction, now is a good time to review the commercial rationale and supporting documentation. The law has not become more hostile to legitimate commercial arrangements, but the importance of demonstrating commercial substance is critical.
This alert is general in nature and does not constitute tax or legal advice.
