GAAR at a Crossroads – Part I: A Shift to Objectivity?

By Dr Hendri Herbst & Dewald Pieterse

Introduction

The South African fiscal landscape is currently at a jurisprudential crossroads which stands to reshape the foundations of tax liability within the country. The application and interpretation of the General Anti-Avoidance Rule (GAAR), codified in sections 80A–80L of the Income Tax Act 58 of 1962 (ITA), is undergoing its most rigorous judicial interrogation since its legislative enactment in its current form in 2006. Though not the only recent development, the most consequential concerns the forthcoming judgment in the matter of Absa Bank Limited and Another v C:SARS,[i] a case that has traversed the High Court and Supreme Court of Appeal (SCA) and is now poised for final determination on its merits before the Constitutional Court following its hearing on 23 September 2025. Furthermore, the Tax Court handed down a remarkably detailed judgment in Mr Taxpayer G v C:SARS[ii] in September 2025 which pronounced upon various aspects of the GAAR, departing from traditional interpretations of the requirements in many respects.

This article, the first in a three-part series, examines the genesis of the conflicting interpretations of the GAAR giving rise to these disputes. It outlines how these disputes arose from a conflict of visions pitting long-established interpretations of the GAAR against the ostensibly revised approach under the so-called ‘new GAAR’ enacted in 2006. The central question of disagreement is whether the GAAR has evolved from a regime tethered to the taxpayer’s subjective purpose (given, after all, the GAAR’s policy rationale of deterring intentional tax avoidance) into one governed by the objective structural logic and economic effects of an arrangement. While subsequent articles will delve into the details of the ‘tax benefit’ enquiry (Part II) and the ‘party error’ issue (Part III) currently under consideration by the Constitutional Court, this instalment explores the broader transition towards the ‘more objective’ approach which SARS claims the new GAAR to embody.

The choice principle and the Duke of Westminster

To understand how deep the now-attempted transition reaches, one must look to the so-called ‘choice principle’, a foundational doctrine of tax law derived from the English case IRC v Duke of Westminster.[iii] The choice principle posits that every taxpayer is entitled to order its affairs so that the tax attaching under the appropriate Acts is less than it otherwise would be. In other words, faced with a variety of means to achieve a specific end, a taxpayer is entitled to pursue the most tax-efficient means. Affirmed domestically in CIR v King,[iv] it has long been established in South Africa that genuine legal rights and obligations must be respected by courts, even if the primary motive behind a particular course of action among various options was the reduction of tax.

The subjective regime: the primacy of the taxpayer’s mind

Under the antecedent regime of the deleted section 103(1), the GAAR was characterised by its reliance on a subjective purpose test, namely, that the taxpayer’s sole or main purpose for entering into an abnormal arrangement was the achievement of a tax benefit. The primacy of the taxpayer’s subjective purpose was cemented in the seminal decision of SIR v Gallagher,[v] in which the then Appellate Division held that the test for tax avoidance was ‘undoubtedly a subjective one’, requiring the Commissioner to determine the specific intent of the taxpayer at the moment the transaction was implemented. The court emphasised that of prime importance in determining purpose was the evidence of the progenitor of the scheme as to why it was carried out.

This subjective approach was further reinforced in SIR v Geusteyn, Forsyth and Joubert,[vi] which involved partners in a consulting firm who transferred their business to a company with unlimited liability. The court accepted the taxpayers’ evidence that the primary purpose was to obtain commercial benefits, such as the ability to participate in larger consortiums, despite the substantial and foreseeable tax savings. The taxpayers’ ipse dixit (their sworn testimony regarding their intent) therefore carried considerable weight in the subjective purpose test.

The ‘high-water mark’ of this defence was CIR v Conhage (Pty) Ltd.[vii] In Conhage, the SCA held that because the taxpayer had a genuine commercial need for capital, the choice of a tax-efficient sale and leaseback vehicle (as opposed to a less tax-efficient loan) did not engage the jurisdiction of the GAAR. Hefer JA famously reaffirmed the choice principle in stating that ‘within the bounds of any anti-avoidance provisions… a taxpayer may minimise his tax liability by arranging his affairs in a suitable manner.’ To an extent, this created a paradox where a subjective commercially driven sole or main purpose could immunise an objectively unusual transaction, rendering the ‘abnormality’ requirement of the GAAR superfluous in the face of a commercially rationalised purpose.

From subjective purpose to objective effect?

Prior to the enactment of the new GAAR, successful invocation by SARS therefore often hinged on the ‘sole or main purpose’ test, namely, on whether an arrangement ‘was entered into or carried out solely or mainly for the purposes of the avoidance or the postponement of liability’. As we have noted, the purpose requirement of the GAAR had been tested since its inception by means of a subjective enquiry into the subjectively held purpose for which a particular taxpayer entered into the arrangement.

In contrast, the 2006 overhaul introduced section 80A, which defines an ‘avoidance arrangement’ as an ‘impermissible avoidance arrangement’ if its sole or main purpose was to obtain a tax benefit. SARS has attempted to leverage this subtle textual change to argue for a depersonalised enquiry: namely, that rather than interrogating the taxpayer’s subjective purpose, the focus has shifted to whether the arrangement’s primary effect, regardless of the taxpayer’s intention, is the production of a tax benefit.

Such a depersonalisation of the purpose test would mark a paradigm shift toward an overall objective approach. As explored in our October 2025 Tax Flash, Mind over Matter? Developments in the GAAR ‘purpose’ debate, however, we do not believe that the arguments in support of a departure from the subjective test withstand scrutiny. For one thing, such an argument is based on an incorrect (or, at best, incomplete) reading of the textual differences, as the wording of the new GAAR’s purpose test is virtually identical to that of the old GAAR as it read at the time when the then Appellate Division unequivocally found in favour of applying a subjective purpose test in Gallagher, Geusteyn, and Ovenstone v SIR.[viii]

For another, if the legislature had intended for the new GAAR to be a purely objective test, this could and should have been made clear in the wording of the section by the lawmaker. The failure to use explicit language mandating objectivity suggests that the traditional subjective standard remains the intended position. Furthermore, the SCA’s recent judgment in the Absa Bank[ix] matter states in no unclear terms that ‘[w]hat must be determined in every case is the subjective purpose of the taxpayer’.

Crucially, while this broader shift in GAAR interpretation is part of the current jurisprudential friction, it must be clarified that the ‘sole or main purpose’ requirement (and the subjective versus objective debate in relation to that requirement) is not specifically part of the dispute currently before the Constitutional Court in the Absa Bank matter. Absa’s review is directed solely against the so-called ‘party error’ and the ‘tax benefit error.’ Consequently, while the Constitutional Court’s forthcoming judgment will set the trajectory for future interpretations, the Court will not decisively settle the debate on the nature of the sole or main purpose requirement.

The ‘tax benefit’ enquiry

The most consequential divergence in interpretations for the purposes of the dispute in Absa Bank arguably lies in the approach to the GAAR’s ‘tax benefit’ requirement. Under the old GAAR, determining a ‘tax benefit’ was a taxpayer-centric enquiry which essentially asked: ‘but for this arrangement, would this specific taxpayer have incurred a higher tax liability, which the taxpayer has now stepped out of the way of?’.

In Absa Bank, however, SARS asserts that an arrangement-centric view must be taken. In a dramatic departure from the established approach, SARS contends that it is not necessary to prove that a specific taxpayer avoided a tax they personally owed; instead, it suffices that the arrangement was structured to produce a tax saving for someone, and that the taxpayer being assessed benefited financially. This ‘transferable tax benefit’ theory lies at the heart of the litigation. SARS’s position arguably conflates an economic benefit (receiving a higher return) with a tax benefit (avoiding a specific liability).

This reorientation is further complicated by the Mr Taxpayer G judgment, which (although non-binding) recalibrated the ‘but for’ test. The Tax Court rejected that the correct counterfactual under the but-for test is ‘no transaction at all’, holding instead that the correct inquiry is whether, but for the transaction being structured in a way to avoid tax, the party would have incurred a liability when compared to an ‘appropriate hypothesis’. While this article seeks only to point out that the conceptual shift toward objectivity advanced by SARS concerns the tax benefit enquiry as well, Part II will interrogate the potential implications of this development in further detail.

The ‘party error’ and knowledge

The shift towards objectivity is equally discernible in SARS’s approach to the ‘party’ issue. Absa argues that participation in an impermissible avoidance arrangement requires volition and conscious involvement in the affected steps. SARS, however, maintains that knowledge is irrelevant in determining whether a person is a ‘party’ to such an arrangement.

In SARS’s view, if a taxpayer provides funding or receives economic benefits from an arrangement, they are a ‘party’ to it for purposes of the GAAR, even in relation to steps of which they bear no knowledge. SARS argues that requiring proof of knowledge would encourage ‘deliberate ignorance’ or ‘wilful blindness,’ allowing participants to profit from tax avoidance while maintaining plausible deniability. This reveals once again a (radical) shift in approach toward focusing on objective outcomes rather than a taxpayer’s state of mind or personal knowledge. Such an approach, however, must necessarily be premised on the assumption that the purpose of the GAAR is not to defeat intentional tax avoidance, but tax avoidance per se, even if accidental – yet whether this premise can be supported with reference to any authority is highly doubtful.

The 2006 overhaul and the judicial interregnum

A striking aspect of the legal dispute in Absa Bank is the judicial vacuum that preceded it. Since the 2006 enactment of the new GAAR, no major cases related thereto had reached the higher courts for adjudication, resulting in considerable uncertainty as to the true nature of the new GAAR and how sections 80A–80L were to be interpreted.

Presumably, the use of alternative dispute resolution and settlements off record meant that many GAAR contentions were resolved behind closed doors, resulting in a dearth of  judgments. One may infer that SARS was careful not to litigate borderline cases but waited instead for a perfect test case – a particularly egregious scheme – in an attempt to ensure a strong precedent in its favour.

It would now seem that this long hiatus provided cover for SARS’s aggressively objective interpretation to remain untested, during which period the market continued to rely on long-standing precedents like Gallagher and Conhage,  assuming that the long-standing principles of the old GAAR, insofar as they had not been clearly changed by the new GAAR, still held firm. The sole or main purpose requirement aside, Absa Bank represents the inevitable collision between these two visions. Critics justifiably contend that applying a radically new interpretation retroactively to transactions entered into during this interregnum violates the principle of legal certainty. For taxpayers, such a belated shift in interpretation of the GAAR towards objectivity would be less a clarification of the law and more akin to the jaws of a long-obscured trap closing shut on those who structured affairs in good faith based on then-prevailing precedents.

Procedural thresholds and the business context

One might remark that the ostensible shift toward objectivity is balanced by recent judicial insistence on procedural precision. In BCJ v C:SARS,[x] the Tax Court ruled that for SARS to validly invoke the GAAR, it must expressly and separately address each of the four elements of section 80A in its reasoning. Specifically, the Court held that SARS must provide a ‘preliminary assessment’ explaining why an arrangement occurred within a ‘business context’ if the abnormality indicators related to a business context were being relied upon. The failure to furnish such reasoning, the court found, prevents a taxpayer from meaningfully objecting to SARS’s assessment.

Conclusion

The new GAAR is currently undergoing its most decisive ‘stress test’ yet. If SARS’s vision prevails, the GAAR will become a potent weapon where the objective effect of a transaction would govern irrespective of the subjective purpose or knowledge of taxpayers. If the courts reaffirm the traditional approach, it will restore the importance of volition and protect taxpayers’ affairs from being tainted unbeknownst to them by the actions of third parties.

As we await the Constitutional Court’s judgment, it is clear that the GAAR stands at a crossroads. Part II of this series will examine the mechanics of the ‘tax benefit’ enquiry in greater detail, analysing the direction of travel signalled by Mr Taxpayer G, and using the facts of Absa Bank to illustrate the competing interpretations of this requirement.

[i] Absa Bank Limited and Another v Commissioner for the South African Revenue Service (CCT Case No: 72/24).

[ii] Mr Taxpayer G v Commissioner for the South African Revenue Service (IT 24502) [2025] SATC 12 (unreported).

[iii] Inland Revenue Commissioners v Duke of Westminster [1936] AC 1.

[iv] Commissioner for Inland Revenue v King 1947 (2) SA 196 (A).

[v] Secretary for Inland Revenue v Gallagher 1978 (2) SA 463 (A).

[vi] Secretary for Inland Revenue v Geustyn, Forsyth & Joubert 1971 (3) SA 567 (A).

[vii] Commissioner for Inland Revenue v Conhage (Pty) Ltd 1999 (4) SA 1149 (SCA).

[viii] Ovenstone v Secretary for Inland Revenue 42 SATC 55.

[ix] Commissioner for the South African Revenue Service v Absa Bank Limited and Another (596/2021) [2023] ZASCA 125 (29 September 2023).

[x] BCJ v Commissioner for the South African Revenue Service (2024/8) [2025] ZATC 7 (23 May 2025).