Why so little case law on ss 41–47
Sweep the law reports and a pattern emerges: in the reported decisions located for this review (SAFLII and the SARS judgment lists, searched to 18 July 2026), none interprets the substantive requirements of s 42, s 43, s 45 (including its six-year degrouping charge), s 46 or s 47 — reporting changes over time, so treat that as a dated observation rather than a guarantee. The corporate rules are largely self-assessed, definition-driven, and — where uncertainty exists — resolved through SARS advance rulings rather than litigation (for example BPR 388 on the degrouping rule after successive s 45 transfers, and BPR 408 on consecutive s 42 transactions). Disputes surface at the edges: was the arrangement real (simulation)? was its purpose permissible (GAAR)? was value stripped on exit (para 43A and its predecessors)? That is where the judgments live — and why this page spends most of its time there. The cases below are a selection, each read for its actual holding; historical company-tax decisions, procedural GAAR decisions and direct rollover-eligibility decisions are not interchangeable authorities.
The one direct case: Pienaar Brothers (s 44)
Why it matters today: it is the standing proof that structures relying on drafting gaps in the corporate rules carry legislative risk even after implementation. National Treasury announces; Parliament backdates; the courts have allowed it.
The simulation line: Randles Brothers to Sasol Oil
South African law has asked “is this transaction real?” since long before the modern GAAR. The foundation is Commissioner of Customs and Excise v Randles Brothers & Hudson Ltd 1941 AD 369: a disguised transaction is taxed as what it really is — the question being whether the parties actually intended their agreements to operate according to their terms.
The modern restatement came in CSARS v NWK Ltd [2010] ZASCA 168, where a maize-loan structure with offsetting, illusory delivery obligations was held simulated:
NWK alarmed practitioners — read widely, it seemed to condemn any tax-motivated structure. Roshcon (Pty) Ltd v Anchor Auto Body Builders CC [2014] ZASCA 40 settled the reading: NWK did not change the law; a transaction is simulated only if the parties did not intend it to have its stated legal effect, judged on all the circumstances — a tax motive alone does not make it a sham. Sasol Oil (Pty) Ltd v CSARS [2018] ZASCA 153 then applied the calibrated test to a group supply-chain restructure: the majority found the back-to-back crude-oil sale agreements genuine and implemented as written, and the taxpayer won.
The working rule from the trilogy: structure freely, but mean it. Documents must match conduct — deliveries, payments, board approvals, registers. A roll-over implemented on paper while the parties behave as if nothing changed is exactly what the simulation doctrine exists to unwind.
Founders Hill and Labat: two structural principles
The GAAR line: Absa, Erasmus and the dividend schemes
The modern battleground. Four strands:
The Absa saga (2021–2026)
SARS applied the GAAR to Absa’s participation in a preference-share structure that converted taxable interest into exempt dividends. After a procedural odyssey — High Court review set aside the GAAR notices (2021), SCA reversed on jurisdiction (2023), Constitutional Court restored High Court jurisdiction (United Manganese [2025] ZACC 2) — the Constitutional Court decided the merits of that review challenge in Absa Bank Ltd v CSARS [2026] ZACC 15, dismissing Absa’s appeal 9–1 (Majiedt J for the majority; Rogers J dissenting):
For restructures, Absa means the GAAR net is wide: every participant in a funded reorganisation — including the bank — must ask what the structure as a whole achieves, not only what its own leg looks like. Two limits keep the case in proportion. The court was deciding a review: whether SARS’s GAAR assessments had to be set aside for two alleged errors of law (who counts as a “party”, and whether Absa obtained a “tax benefit”). It was not finally fixing every element and amount of the tax eventually payable. And being a participant is only one requirement — the other GAAR tests must still be met.
Erasmus: GAAR procedure
Mr Taxpayer G: the scheme architect
Taxpayer Arrow: tax upheld, penalties and interest set aside
Company AF: the corporate rules meet the GAAR
Company AF (Pty) Ltd and Others v CSARS [2026] ZATC 6 is the closest a court has come to the heart of this hub. Shareholders had earlier moved their operating-company shares into personal investment companies using s 42 asset-for-share roll-overs — unremarkable, and not attacked. The trouble was the 2017 exit: a four-step composite in which the target declared R274.7 million in pre-sale dividends, funded by the purchaser’s R280.3 million share subscription, after which the old shares were sold for R1,000 — the classic strip, executed before the modern para 43A applied.
The Tax Court applied the GAAR to the composite arrangement, finding it abnormal, without commercial substance and a misuse of the s 10(1)(k)(i) dividend exemption; the “dividends” were taxed as what they economically were — sale proceeds. The court dissected the arrangement step by step:
The outcome had several parts, and each matters. The assessments were upheld on the merits, but the court confirmed the tax is CGT on the gain — proceeds less the shareholders’ base cost, being their s 42 roll-over cost plus later qualifying expenditure, not nil. The understatement penalties were remitted — the taxpayers had taken specific professional advice and disclosed the arrangement under the reportable-arrangement rules before any audit. The provisional-tax penalty was set aside and remitted to SARS for reconsideration. The s 89quat interest was confirmed and could not be remitted once the GAAR was invoked. And, following Erasmus, SARS’s attempt to re-plead its GAAR case was struck. Advice, disclosure and procedure all counted — on penalties, not on the tax or the interest. See Dividend Stripping & the 18-Month Web for the statutory rule that now covers this ground.
Thistle Trust: trusts, tiers and capital gains
For anyone restructuring around a family trust — the audience of this hub — Thistle is the caution against assuming capital gains flow tax-transparently through stacked trusts. Where a restructure’s gains will be vested through more than one trust, model the CGT at each tier before implementing. Keep the holding within its limits: it concerns capital gains under paragraph 80(2), not the separate s 25B rules for ordinary trust income. And the Court was interpreting what the statute already said — it did not create a new restriction in 2024.
Lance Dickson: penalties and the onus
Lance Dickson Construction CC v CSARS [2023] ZAWCHC 12; 84 SATC 209 is not a roll-over case — it concerned CGT timing on property sales and the 25% understatement penalty — but it is sometimes mislabelled as a s 42 authority, and its real holding is valuable. Under s 102(2) of the Tax Administration Act, SARS bears the onus of proving the behaviour on which a penalty is based (“reasonable care not taken”, in that case). SARS failed to prove it, and the full bench would not let the Tax Court substitute a different, unpleaded basis:
If a restructure position is ever challenged, the penalty layer is its own battlefield with its own onus. Tax, penalties and interest are decided separately: in Company AF, Thistle and Taxpayer Arrow the tax was lost, yet the understatement penalties did not survive. Professional advice and honest disclosure can be important evidence at that stage — they do not guarantee that a penalty will be remitted.
What the case law means for your restructure
Pull the threads together and the practical guidance is short. Implement for real — registers, resolutions, deliveries and payments matching the documents (Randles Brothers, NWK, Roshcon, Sasol Oil). Have a commercial story for every step, because the GAAR examines how the steps interlock and catches participants who merely stepped consciously into the structure (Company AF, Absa) — remembering that simulation and the GAAR are distinct enquiries with distinct tests. Treat the exit as the danger zone — value extraction around a disposal is where the specific and general anti-avoidance rules converge (Company AF, Erasmus, Mr Taxpayer G). Keep the tax separate from penalties and interest — a taxpayer can lose on the tax and still succeed, in whole or in part, on penalties or interest (Taxpayer Arrow, Company AF). Do not build on drafting gaps (Pienaar Brothers). Model trust tiers separately (Thistle). And paper your advice — contemporaneous professional advice and full disclosure may support a penalty defence, though they do not guarantee remission.
Frequently asked questions
Barely — on the decisions located for this review (to 18 July 2026), Pienaar Brothers (2017) is the only judgment directly on a corporate-rules provision (s 44’s retrospective s 44(9A)), and Company AF (2026) is the closest engagement with a s 42-based structure — attacked at the exit under the GAAR, not on the s 42 requirements. The substantive tests in ss 42–47 remain, on that search, judicially untested; practice runs on the statutory text and SARS rulings.
No. Roshcon (2014) confirmed that simulation requires the parties not to intend their stated legal arrangements — a tax motive alone is not disqualifying — and Sasol Oil (2018) upheld a tax-efficient group restructure that was genuinely implemented. The line is crossed when documents and conduct diverge, or when steps exist that no one would take but for the tax result — that is NWK and GAAR territory.
Two from 2026 share the title. Absa v CSARS [2026] ZACC 15 (Constitutional Court) set the GAAR’s reach — a participant can be caught without knowing the full structure, provided the other GAAR requirements are also met. Company AF [2026] ZATC 6 (Tax Court) is the first full judgment on the pre-2017 dividend-strip exits and shows the GAAR doing para 43A’s work for older transactions, while also rewarding advice and disclosure at the penalty stage.
A binding private ruling binds SARS only for the applicant and the disclosed facts — it is not precedent. But in an area with almost no case law, published rulings are the best available evidence of SARS’s interpretation: BPR 388 (degrouping after successive s 45 transfers) and BPR 408 (consecutive s 42 transactions) are required reading for those fact patterns.
They decide what a dispute actually costs. Lance Dickson (2023) puts the onus on SARS to prove the penalty behaviour it pleads; in Thistle (2024) the penalties stayed set aside because SARS was refused leave to cross-appeal, and in Company AF (2026) the understatement penalties were remitted on the facts, with the taxpayers’ specific professional advice and reportable-arrangement disclosure counting in their favour. Advice and disclosure may support a penalty defence — the outcome still turns on the statutory penalty categories, the reasonableness of the position taken and the taxpayer’s conduct. The pattern remains: document the advice before you implement.
Before signing
Case-law lessons are general; roll-over relief depends on the exact parties, values, documents and order of steps in your transaction. Obtain transaction-specific tax and legal advice before agreements or resolutions are signed — a later correction may not restore the relief.