In recent years, South African courts have begun framing tax relief measures as privileges rather than entitlements, significantly impacting the economic landscape. This article explores the implications of this shift, examining key court cases and their effects on compliance, investment, and the broader social contract.
Image: File
In the last five years, a quiet but important shift has taken place in our rebate and refund jurisprudence. South African courts – especially in the diesel and fuel levy space – have started to speak about these relief measures as if they were “privileges”, exceptional concessions that taxpayers must earn through near-perfect formal compliance. You see this very clearly in CSARS v Glencore Operations SA (Pty) Ltd, and then, more starkly, in the Constitutional Court’s recent Tholo Energy Services CC v CSARS judgment.
On one level, this is about interpretation: how tightly we read section 75, Note 6, and related provisions. But at a deeper level, it is about something more fundamental – how we understand the role of rebates and refunds in our economy, and whose interests our courts instinctively lean towards when the text runs out. In a country with stubbornly high unemployment, investment in ice, and corporates sitting on record cash piles, that instinct matters.
Relief is part of the design, not a favour
Excise duties and fuel levies are local consumption taxes. They are not meant to stick where goods are exported or used in qualifying non-road activities. If diesel never touches a South African public road, the economic logic is that the fuel levy should not apply. If excisable goods are not consumed in South Africa, the excise is not supposed to be part of the long-term cost base.
The same is true on the customs side. Drawbacks and production rebates exist so that our exporters and manufacturers are not saddled with import duties on inputs that feed into export production or clearly targeted industrial policy outcomes. The Automotive Production and Development Programme (APDP) is the poster child: duty rebates and production-linked credits are the main reason we still assemble vehicles in this country.
In all these cases, the rebate is not an act of generosity. It is the mechanism that makes the tax design coherent. If you accept that, then the debate should be about how best to police and evidence the underlying commercial reality – not whether the fiscus is doing taxpayers a favour.
How Glencore and Tholo changed the tone
Against that backdrop, what our appellate courts have done in the diesel and fuel space is subtle but significant. In CSARS v Glencore Operations SA (Pty) Ltd of August 2021, the Supreme Court of Appeal (SCA) had to decide whether Glencore’s diesel, used across its mining operations, fell within “primary production in mining” for purposes of the diesel refund scheme. The notes to Schedule 6 contain a detailed list of qualifying activities. Glencore argued that, read purposively, its integrated mining operations fit the bill.
The SCA wasn’t persuaded. It treated the list in Note 6(f)(iii) as exhaustive and effectively said: if an activity is not there in black and white, it is out, however integral to mining it may be in practice. There is very little appetite in the judgment for the idea that the refund is there to see that diesel used in real mining activity is relieved of levy, and that the list should be read with that economic purpose in mind.
Fast-forward to Tholo Energy Services CC v CSARS in January 2026 in the Constitutional Court. Here the issue was fuel levy refunds under the DAS regime, tied to exports and the requirement that fuel be obtained from a licensed manufacturing warehouse (a VM). Tholo’s underlying commercial story – fuel sourced from the local supply chain and destined for exports into the BLNE (Botswana, Lesotho, Namibia and Eswatini) region – looked, at least at a distance, like the sort of thing the scheme was aimed at.
The court’s answer was firm: every statutory requirement must be met strictly. The VM-premises link is not a technicality but a hard boundary. And, crucially for present purposes, the court embraced the language of “privilege” – these refunds are framed as special benefits that justify rigorous, almost sacramental, compliance.
Once you adopt that vocabulary, it becomes much easier to say “sorry, you miss one element and you are out”, even when the commercial use aligns with the policy rationale.
The pre-Glencore trajectory: High Court cases already pointing the way
Glencore did not fall out of a clear blue sky. By the time the SCA handed down judgment, the High Courts had already started to talk about diesel refunds in a way that narrowed the room for substantive compliance.
In Assmang Proprietary Limited v CSARS and Others (Gauteng High Court, 2023), the court explicitly referred to diesel refunds as “class privileges” aimed at specific primary sectors and emphasised that section 75 and Item 670.04 require strict compliance. The message was that this is not a general entitlement. If you want to share in the privilege, you must fit neatly within the statutory box and maintain records exactly as prescribed.
In Sandbaken Boerdery (Pty) Ltd v CSARS (Northern Cape High Court, 2025), the taxpayer tried to argue substantial compliance – that it had, in substance, met the purpose of the diesel refund provisions even if its paperwork was not perfect. The court shut that door, holding that the language of section 75 and Note 6 is clear and must be followed strictly, procurement-law notions of substantive compliance do not translate into this context.
Seen together, the pattern is clear: first, a line of High Court decisions talking about diesel refunds as class-based privileges and insisting on tight compliance. Glencore, in the SCA, cementing a literal, exhaustive reading of the qualifying activities. Finally, Tholo giving the “privilege” narrative constitutional oxygen.
Why “privilege” is the wrong frame
The “privilege” label is problematic for two reasons.
The first is conceptual. If excise and fuel levies are truly meant to tax domestic consumption and certain types of road usage, then relieving non-domestic or non-road use is not a kindness. It is an adjustment that restores the intended incidence of the tax. The same logic underpins export-related duty relief and schemes like APDP.
The second is economic. In an environment like ours, the competition stakes are not academic. When a rebate or refund attaches to one player and not another, the non-qualifying firm carries a structurally higher tax cost. That feeds directly into pricing, margins and the ability to invest.
Consider two operators whose diesel usage is, in substance, the same. One has the benefit of deep advisory support, long-standing systems and a compliance team that has had years to fine-tune records to match the South African Revenue Service’s (Sars) interpretation. The other is newer, leaner, still formalising its systems. If the first secures refunds and the second is knocked out on technicalities – despite an identical commercial reality – we are not just sorting compliant from non-compliant, we are entrenching an uneven playing field.
This is precisely why I find the “privilege” language troubling. It smuggles in an assumption that withholding relief is costless, as if the only issue at stake is protection of revenue. What it obscures is the other side of the social contract – that the state is meant to create conditions in which competitive, employment-creating businesses can survive and grow.
Substantive compliance and a country on the edge
None of this is to say that controls do not matter. They do. South Africa has lived through non-compliance and all manner of tax abuse. Sars is right to worry about leakage, carousel fraud, paper exports and sham “primary production” structures. The question is not whether we need controls, but how we calibrate them.
Our constitutional jurisprudence is capable of recognising substantive compliance when the purpose of a requirement is met and the deviation is not material. We have seen this in other areas of law. Yet, in the diesel and fuel levy space, courts are signalling that such flexibility is off the table.
In Sandbaken, the court explicitly declined to import substantive-compliance thinking into the refund context. In Assmang, both the High Court and SCA leaned into the language of strict eligibility and meticulous records. In Glencore and Tholo, the idea that one might look at the underlying use and ask, “has the economic rationale really been satisfied here?” barely features.
That stance might be defensible in a static, high-trust, high-growth economy. It looks different in South Africa in 2026. Official unemployment remains stuck around the one-in-three mark on a narrow definition, and much higher on an expanded definition including discouraged workers. Long-term unemployment dominates the profile.
The Organisation for Economic Co-operation and Development has flagged weak investment and low potential growth as central medium-term risks.
In that setting, a tax and judicial environment that says “if there is any doubt, we will resolve it in favour of the fiscus, and we will treat relief as a precarious privilege” is not neutral. It feeds directly into the decision whether to build the next plant here, to expand a fleet, to invest in a new shaft, or to simply sit on cash and wait.
Sars audits, distrust and the risk of a “deny first” culture
Layer onto this the lived experience of many taxpayers with Sars’ current audit and refund practices. Over recent years practitioners have consistently reported more aggressive verification of refunds, increasing use of “risk review” flags to delay payments, and, in some instances, system failures and internal processes that expose taxpayers to both cash-flow strain and procedural unfairness.
The Tax Ombud has had to intervene in cases of unlawfully delayed value-added tax (VAT) refunds. Tax firms write frequently about diesel refund disputes where SARS appears to reach for technicalities to justify non-payment. Small medium enterprises and corporates alike complain about documentation expectations that feel disconnected from how modern businesses actually operate.
Drop a string of appellate judgments into that environment – judgments that affirm narrow readings, strict compliance and “privilege” language – and it is not hard to see how the internal message inside Sars might be heard as:
- take the tightest possible view of eligibility;
- if anything is not exactly as the Act, rules or interpretation note envisages, deny; and
- the courts have our back.
No Commissioner or minister will ever say this explicitly. But systems respond to incentives and signals. When denying or delaying refunds is read as cautious, fiscally responsible behaviour, and when courts repeatedly back that approach, the natural risk for officials is to err in that direction. In a fiscally constrained environment, with the Minister of Finance under pressure to fund infrastructure and social spending without raising headline rates, the temptation to lean on refunds as a soft lever is obvious.
The cumulative effect is corrosive. It erodes trust. It encourages businesses to discount the value of relief in their investment models. It shifts working-capital pressure onto the private sector at precisely the wrong moment in the economic cycle.
The spillover to APDP and future schemes
The diesel and fuel levy space is the current theatre, but the logic will not stay neatly in that box. Take APDP. The entire programme rests on the credibility of a complex incentive architecture – Volume Assembly Localisation Allowance, Production Rebate Certificates and other mechanisms that reduce duty costs in proportion to local production and value addition. Automotive original equipment manufacturers and component manufacturers make multi-billion rand location and expansion decisions on the assumption that, if they invest and comply, the programme will function more or less as advertised.
If Sars, fortified by “privilege”-framed jurisprudence, starts to treat APDP relief like the diesel refund world – tight, unforgiving, with little room for commercial common sense – the risk profile of South Africa as an automotive location shifts. New entrants will factor that into their models. Existing players may think twice before committing to the next investment cycle here.
The same applies to any future green energy incentive, export processing relief or sector-specific rebate. The more our courts normalise the idea that relief is a delicate privilege, revocable on technical missteps, the less persuasive South Africa’s incentive story becomes, no matter what appears in glossy policy documents.
Are the courts just being cautious – and on whose side should they err?
To be fair to the judiciary, the pressures they are responding to are real. There is the separation-of-powers concern: judges are understandably wary of reading relief schemes more expansively than their text allows, particularly in tax matters, where they have been cautioned for decades not to “legislate under the guise of interpretation”. There is a tangible fear of abuse: the last decade has left a very low tolerance for any fiscal mechanism that can be gamed. And hovering over everything is the macro-fiscal context – high debt, pressing expenditure needs, and an understandable reluctance to authorise outcomes that look like large cheques back to big corporates.
But these concerns do not answer the key question: in a state that describes itself as developmental, in an economy with high unemployment and an investment drought, on whose side should courts err when the law genuinely admits of more than one reading?
Right now, the pattern suggests that the error term falls on the side of the fiscus. If there is ambiguity in Note 6, it is resolved against the claimant. If records are imperfect but the underlying commercial use is clearly within the scheme, the defect is fatal. If there is a tension between economic purpose and control architecture, control wins.
I would argue that, at this point in our history, that instinct is misaligned with what the broader social contract requires. We need courts that are uncompromising with fraud and abuse, but that are also alive to the economic function of relief measures, and willing to adopt interpretations that preserve competitiveness and jobs where the text reasonably allows it.
That does not mean opening the floodgates. It means acknowledging that, in borderline cases, the costs of leaning too far towards fiscal protection can be measured in foregone investment and entrenched unemployment.
A different way: commercial literacy and co-designed rules
The solution is not to abandon strictness, but to become smarter about what we are strict about. Much of the current tension arises because the rules and evidential expectations in rebate regimes are out of sync with how business actually works. Supply chains are no longer simple. Enterprise resource planning (ERP) systems, outsourced logistics, consolidated procurement, and digital records are the norm. Yet many of our rules and Sars’ expectations seem anchored in an era of single-site operations and paper-based logbooks. There is a practical path out of this:
This may reduce the need for courts to reach for “privilege” language as a blunt instrument and make it easier for honest taxpayers to show that they are within the economic and legal design. It will also afford Sars a firmer, more realistic basis on which to say “yes” to those who qualify and “no” to those who do not, without defaulting to a deny-first posture.
Where do we go from here?
The line from Assmang, through Glencore to Tholo tells us something important about where our law is going: rebates are being treated as narrow carve-outs, and the notion of “privilege” is gaining constitutional respectability. In a vacuum, that might be a tidy, fiscally conservative stance. In South Africa, in 2026, it is much more than that.
It shapes how Sars audits, how businesses invest, how much trust exists in the system, and ultimately how well our tax and industrial policy instruments serve their stated purpose of supporting growth and employment.
My contention is simple: relief measures like diesel refunds, fuel levy rebates, and APDP-style incentives are not favours. They are part of the architecture of a tax and industrial system that is supposed to make South Africa a viable place to produce, employ, and compete. If we continue to treat them as precarious privileges, we should not be surprised when capital stays on the sidelines, and when the jobs we so desperately need do not materialise.
* Ramatlhodi is the partner and Africa indirect tax leader at Deloitte Africa.
PERSONAL FINANCE