Can the courts become SARS’ safe harbour for broken assessments?

A respectful critique of CSARS v Meiring Citrus (Pty) Ltd

I find myself in the slightly awkward position of agreeing, at least broadly, with the ultimate tax outcome in a judgment while disagreeing quite strongly with how the Court got there.

In Commissioner for the South African Revenue Service v Meiring Citrus (Pty) Ltd¹, the Western Cape High Court overturned a Tax Court judgment that had found in favour of the taxpayer. The High Court held that the bulk of what had been described as an insurance premium was not deductible and that SARS was entitled to reopen a prescribed assessment.

On the commercial substance of the arrangement, the High Court may well be right. The difficulty is that SARS did not properly plead much of the case on which the High Court ultimately relied. That raises a broader and, in my view, more important question than the tax treatment of one unusual insurance product:

Can the courts become SARS’ safe harbour for broken assessments — a place where an insufficiently investigated, inadequately reasoned or poorly pleaded assessment can be repaired after the taxpayer has already met the case SARS chose to make?

The background

Meiring Citrus carried on a citrus farming business and faced the usual risks associated with exporting fruit, including crop disease and rejected consignments. Historically, it had absorbed those risks from its own reserves.

Shortly before the end of its 2017 year of assessment, the taxpayer entered into what was marketed as a structured self-insurance product. It paid R10 million and received cover with an indemnity limit of R12 million. Of the R10 million, R400,000 was an underwriting charge. The remaining R9.6 million was credited to an “experience account”. Claims were effectively funded from that account, the positive balance earned notional interest, and the taxpayer could recover the remaining balance on cancellation after giving 30 days’ notice. The policy could also be pledged as security.

The taxpayer claimed the R10 million as an insurance expense. It did not declare R1,197.52 of notional interest credited to the experience account during the 2017 year.

SARS noticed the extraordinary increase in insurance expenditure and selected the return for verification. The taxpayer’s accountant described the product to SARS as “self-insurance” and supplied some supporting documents, but not the complete policy and experience-account statements. SARS nevertheless finalised the verification without adjustment. It only commenced the later audit after the ordinary three-year prescription period had expired.

SARS then disallowed R9.6 million of the deduction, included the notional interest and imposed a 10% understatement penalty.

The Tax Court found that SARS had not established a valid basis for lifting prescription. It also held, on the merits, that the payment constituted expenditure actually incurred because there had been at least a movement of assets: cash had passed to the insurer and the taxpayer had acquired contractual rights in exchange. The Tax Court stressed that SARS had not pleaded that the agreement was a sham or simulated transaction and had confined its section 11(a) case to non-incurral, together with reliance on sections 23L and 23(e).

The High Court disagreed. It held that the arrangement was, in substance, an investment or deposit disguised as insurance; that only the R400,000 underwriting charge was a genuine premium; that the R9.6 million was not expenditure actually incurred; and, alternatively, that it was capital in nature. It also held that the misrepresentation and non-disclosure permitted SARS to reopen the prescribed assessment, including the entire taxable-income determination for 2017.

The uncomfortable part: I largely agree with the commercial conclusion

I do not think it would be fair for me now to pretend that I always regarded the R9.6 million as an ordinary deductible insurance premium.

In my previous commentary on the Tax Court judgment, I expressly raised the difficulty that the taxpayer had exchanged cash for another asset or bundle of contractual rights. I questioned whether the transaction involved a true loss of wealth or simply an asset swap. I also emphasised that SARS may have argued the wrong case: had it properly relied on substance over form, simulation, GAAR or another correctly pleaded basis, the result might have been different.

So my criticism of the High Court is not that its commercial instincts were obviously wrong. On the contrary, I agree that an interest-bearing amount that remains recoverable on short notice, can be pledged as security and is used to fund the taxpayer’s own losses looks far more like a self-insurance reserve or deposit than a conventional insurance premium.

My criticism is that a substantively attractive answer does not excuse a procedurally defective route to that answer.

A selective reading of Labat

The first difficulty lies in the High Court’s treatment of Commissioner for SARS v Labat Africa Ltd².

Labat says that expenditure requires a diminution of assets, even if only temporary, or at the very least a movement of assets. It also expressly says that the taxpayer need not ultimately be poorer because the counter-performance may have the same or even a greater value than what was expended.

The High Court quoted that passage, but then held that there was no expenditure because the taxpayer merely changed the form in which it held value. Its net asset position remained unchanged: it surrendered cash but acquired a recoverable interest-bearing right against Santam. The Court added that Labat requires a movement “by way of expenditure” — a genuine parting with value — and not merely any movement of assets.

There is force in distinguishing an actual purchase from a loan or deposit. Depositing R10 million into a bank account is not ordinarily treated as expenditure merely because ownership of the physical money passes to the bank. But the High Court’s reasoning goes further. It places substantial weight on the taxpayer’s unchanged net-asset position and the acquisition of an equivalent right, even though Labat expressly warns that equal or greater counter-performance does not prevent expenditure.

Almost every purchase of an asset involves an exchange of cash for something of value. If unchanged net worth means there was no expenditure, the capital-versus-revenue enquiry would often never arise: the acquisition of a machine, building or other capital asset would cease to be expenditure because the taxpayer received an asset in exchange. That cannot be the principle.

The High Court’s conclusion is therefore only convincing if the R9.6 million is first characterised as a true deposit or loan-like placement rather than consideration paid for contractual rights under an insurance arrangement. And that brings us directly to the pleading problem.

The Court decided the case SARS did not plead

The Tax Court did not overlook the strange commercial features of the product. It expressly recognised the deposit analogy and repeatedly recorded that SARS had not pleaded that the ostensible insurance arrangement was really something else. Indeed, SARS’ counsel confirmed that sham or simulation was not SARS’ case.

The Tax Court then made an observation that, in my view, says almost everything. When considering capital character, it reiterated that SARS’ case was not that the expenditure was “in truth no insurance premium”, but only that it was unconventional insurance. The Court decided the case “viewed from that perspective”.

That was not judicial naivety. It was procedural discipline. The judge was, in substance, telling SARS: perhaps there is another and better case here, but that is not the case you brought.

The High Court said the Tax Court had confused contractual interpretation with simulation. A court is, of course, not bound by the label the parties attach to their agreement. An instrument called a lease may in law be a sale. Objective legal characterisation does not always require the magic words “substance over form”.

But the High Court did not stop at neutral characterisation. It described the arrangement as an investment “disguised or simulated” as insurance, invoked NWK, found that the parties intended to clothe an investment transaction as insurance, and said the terminology was deliberately misleading to avoid tax. Those are not merely conclusions about the dictionary meaning of contractual clauses. They are findings about genuineness, purpose, intention and concealment.

That is precisely the type of case SARS had not pleaded and had, according to the Tax Court record, expressly disavowed.

The same difficulty arises with the alternative finding that the expenditure was capital in nature. SARS’ assessment case under section 11(a) had focused on whether expenditure had actually been incurred. The Tax Court found that capital character had not been identified as a ground of assessment or as an issue in the appeal.

The High Court nevertheless held that section 11(a) “as a whole” was in dispute and that, because section 102 placed the burden of proving the deduction on the taxpayer, Meiring Citrus had to prove every requirement of section 11(a), whether or not SARS developed that requirement as part of its case.

With respect, that reasoning risks emptying Rules 31 and 34 of much of their purpose.

Section 102 tells us who bears the risk of non-proof. It does not tell us what issues are before the Court. The issues are defined by the grounds of assessment and the pleadings. Rule 31 requires SARS to set out the material facts and legal grounds on which it relies. Rule 34 confines the appeal to the issues contained in the Rule 31, 32 and 33 statements. Rule 31(3) prevents SARS from introducing a ground that novates the whole factual or legal basis of the assessment³.

If SARS can plead merely that “the deduction fails under section 11(a)” and then rely on the taxpayer’s onus to introduce any possible defect later, the obligation to state material facts and legal grounds becomes almost decorative.

Pretoria East Motors and the impossible litigation model

The Supreme Court of Appeal’s judgment in Commissioner for SARS v Pretoria East Motors⁴ is critical. The Court accepted that the taxpayer bore the onus, but rejected an approach that left the taxpayer uncertain about what was truly in dispute and what evidence it had to produce. SARS must identify the underlying facts it disputes so that the taxpayer knows the case it must meet. Any other approach, the SCA warned, would make Tax Court litigation unmanageable.

That warning could have been written for this case.

A taxpayer called upon to prove that an expense was incurred prepares one kind of case. It produces the contract, evidence of the unconditional obligation, payment records and evidence of the rights acquired.

A taxpayer called upon to answer a capital case prepares another. It leads evidence about the purpose of the expenditure, the nature and duration of the advantage, and the role of the payment in the income-earning operations.

A taxpayer called upon to answer simulation or deliberate disguise prepares something different again. It may call the broker, underwriter, product designer, insurance-law experts, actuarial witnesses and witnesses dealing with regulatory treatment, purpose and intention.

The fact that some documents overlap does not make these the same case.

The practical consequence of the High Court’s approach is especially troubling. Taxpayers would have to imagine every argument SARS might have made, but did not make, and then disprove all of them. A section 11(a) dispute about production of income could require the taxpayer to prove incurral, trade, revenue character and the absence of simulation — just in case SARS or the court later finds another weakness.

That is not litigation against a defined opponent’s case. It is litigation against every possible case the taxpayer can conceive on SARS’ behalf. It would make tax litigation prohibitively uncertain, expensive and, in many matters, practically impossible.

A court must apply the law correctly. But it is not SARS’ litigation department. It should not be required — or permitted — to reverse-engineer the case that might have saved an assessment after the taxpayer has met the case SARS actually pleaded.

Prescription: being wrong is not automatically misrepresentation

The same theme runs through the High Court’s prescription analysis.

The Court treated the characterisation of the R9.6 million as a deductible insurance premium as a misrepresentation because the complete contractual facts showed, in its view, that it was really a refundable interest-bearing deposit.

There is an important distinction here. A taxpayer may know that funds are refundable, that interest accrues and that claims reduce the account, yet still bona fide believe — particularly on advice and under documents issued by an insurer — that the arrangement constitutes insurance and that the premium is deductible.

Knowledge of the commercial features is not the same as knowledge that the legal characterisation is false.

The Tax Court’s finding was therefore important: whether an amount is deductible under section 11(a) is a legal conclusion. The fact that SARS later makes an adjustment, or that a court later adopts a different legal characterisation, does not by itself prove that the taxpayer misrepresented a fact.

The High Court also relied heavily on the incomplete information supplied during the 2018 verification. It rejected the accountant’s claim that he did not know about the experience account and regarded the non-production of the full policy and statements as deliberate concealment.

Conduct during a pre-prescription verification can, in principle, be relevant. The statutory language is not necessarily confined to statements made in the return. But the case must be pleaded that way, and the conduct must be causative. Here the Tax Court found that SARS had located its pleaded misrepresentation principally in the return, not in the later verification response. The High Court appears to have used the verification conduct to strengthen a weaker return-based case.

More fundamentally, knowledge of the experience account still does not establish knowledge that the deduction was legally impermissible. Nor can a later failure to supply documents retrospectively turn a legally arguable position in the return into a false factual statement.

The interest was material — but what did it reopen?

On one aspect, I think the High Court’s reasoning is stronger than the Tax Court’s.

The Tax Court treated the R1,197.52 of omitted interest as immaterial largely because of its tiny amount. The High Court correctly focused on the materiality of the fact rather than simply the quantum. The existence of interest was relevant because it revealed that the supposed premium remained credited for the taxpayer’s benefit. Even a small interest entry could therefore have prompted questions about the nature of the experience account.

But materiality is not the same as scope, and it is not the same as causation.

The omission could justify reopening the interest determination. It could also, in principle, justify reopening the premium deduction if SARS pleaded and proved that disclosure of the interest would probably have led it to investigate the account, obtain the complete policy, adopt the deposit characterisation and disallow the premium within the three-year period.

The Tax Court expressly accepted that cross-item causation is legally possible⁵. Its point was that SARS must plead it, lead evidence and prove the link. One instance of misconduct is not an “open sesame” to reconsider the entire assessment.

The High Court went much further. It held that the interest and premium were inextricably linked and accepted that, once an assessment is reopened because one item was under-assessed, SARS may correct all other components to ensure that the full amount of tax is assessed.

That proposition is extremely wide. It would mean that the non-disclosure of income item A could permit SARS to revisit an unrelated wear-and-tear allowance, legal expense, stock valuation or capital allowance in the same year, even though the omission of item A had nothing to do with those determinations.

Section 99(2) says prescription does not apply “to the extent that” the failure to assess the full amount of tax was due to the relevant misconduct. Those words call for a causally limited enquiry, not the wholesale reopening of a tax year.

The item-specific approach also fits uneasily with the assessment jurisprudence. First South African Holdings explains that an assessment is a determination of one or more matters. HR Computek rejects the idea of a globular objection to a single undifferentiated assessment amount⁶. Taxpayers must identify and challenge particular assessed amounts. It would be a remarkable asymmetry if the law insists on strict itemisation when the taxpayer objects, but treats the assessment as an indivisible whole when SARS wishes to escape prescription.

Could scrutinise is not would have assessed

Causation is the final difficulty.

The High Court regarded it as obvious that SARS could only scrutinise facts disclosed to it. That is true, but it is not the complete statutory test.

The question is not whether fuller disclosure could have enabled SARS to ask more questions. It is whether, but for the misrepresentation or non-disclosure, SARS would probably have issued the particular additional assessment within three years.

That distinction matters on these facts. SARS noticed the enormous increase in insurance expenditure. It initiated a verification. It was told that the arrangement was “self-insurance”, that R10 million purchased cover of R12 million, that the underwriting charge was 4%, and that the cover was short-term. It then closed the verification without adjustment or any further request.

Those facts do not prove that complete disclosure would have made no difference. But they make causation anything but self-evident. Was the non-assessment caused by the omitted documents and interest, or by a SARS official’s decision not to pursue an obviously unusual transaction?

In Spur Group⁷, SARS led detailed evidence explaining its internal triggers and what would have happened had the return been completed correctly. In Meiring Citrus, SARS called no equivalent witness. The Tax Court held that SARS had therefore failed to discharge its onus. The High Court replaced that missing evidence with the proposition that SARS could have scrutinised the contract if it had received it.

“Could have scrutinised” does not necessarily prove “would probably have assessed within three years”.

The real lesson

It is possible to believe all of the following at the same time:

• The R9.6 million probably should not have been deductible.
• The arrangement looked in substance like a recoverable self-insurance reserve or investment.
• The taxpayer may have supplied incomplete information during verification.
• The omitted interest may have been a material fact despite its small amount.
• And yet SARS may still have failed to raise and litigate a legally sustainable additional assessment before the Court.

Prescription exists precisely because a wrong assessment can become final. Pleadings exist precisely because a taxpayer must know the case it is called upon to meet. The taxpayer’s onus exists within that defined dispute; it does not authorise SARS or the Court to introduce every other possible ground on which the taxpayer might lose.

The High Court may have found the right commercial answer. But courts are not safe harbours into which broken assessments may be sailed for repairs.

Where SARS has assessed on case A, pleaded case A and the taxpayer has met case A, the Court should be slow to uphold the assessment because it can see that SARS might have succeeded on case B. Otherwise the assessment process, Rule 31, prescription and the taxpayer’s right to a fair hearing become little more than procedural scenery.

A broken assessment does not cease to be broken merely because a court can see how SARS could have fixed it.