On 2 October 2024, after an excruciating delay of eight months, the Constitutional Court finally delivered its judgment in the matter of
The Thistle Trust v C: SARS.[1] The judgment spans 59 pages and was a majority (6/8) decision with two dissenting judges. By contrast, the SCA's judgment was only 11 pages, which, if anything, illustrates the seriousness with which the apex court approached the matter in dismissing both the appeal and cross appeal.
The facts
The Thistle Trust is a resident discretionary trust and a beneficiary of 10 vesting trusts collectively known as the Zenprop Group (Zenprop). Zenprop is a property developer and property owner. In the course of its business, it frequently buys and sells properties.
In the 2014, 2015 and 2016 years of assessment the vesting trusts disposed of properties which realised capital gains which were passed on to the Thistle Trust, which in turn passed them on to its natural person beneficiaries.
SARS taxed the Thistle Trust on the capital gains, while the beneficiaries of Thistle took them into account in determining their aggregate capital gain or loss.
The issue the court had to decide was whether s 25B of the Income Tax Act (ITA), the common law conduit principle and para 80(2) of the Eighth Schedule to the ITA supported Thistle's contention that the capital gains should be taxed in the hands of the natural person beneficiaries, or whether the gains were taxable in the Thistle Trust, as contended by SARS.
SARS had imposed a 50% understatement penalty under s 223(1)(iii) of the Tax Administration Act 28 of 2011 (TAA) on the basis that Thistle had no reasonable grounds for tax position taken. Thistle argued that no penalties should have been imposed as this was a
bona fide inadvertent error, and hence excluded under s 222(1) of the TAA. In the SCA judgment[2] it was stated that SARS had conceded on the penalties, but SARS later disputed this and cross-appealed, contending that this was not a case of a
bona fide inadvertent error.
Before getting into the arguments, the one aspect that struck me was whether this case should even have been about capital gains. Normally, a property developer that 'frequently buys and sells properties' is dealing in trading stock and not capital assets. But that issue was not in dispute and no facts were presented which would enable me to express an opinion on it.
The majority decision
Chaskalson AJ wrote the majority judgment and after setting out the background to the case commenced with an examination of the origin of the conduit principle. He noted that it could be traced back to the Privy Council case of
Syme v Commissioner of Taxes, which was an Australian case.[3] The United Kingdom's Privy Council used to be the apex court for many Commonwealth countries, including South Africa up to 1950. The court noted that subsequent commonwealth cases emphasised two points regarding the conduit principle:
- First, it was used to identify the taxpayer who was liable to taxation on particular income, namely, the trustee or the beneficiary.
- Secondly, it was used to protect legislative choices in respect of the favourable or prejudicial income tax treatment of particular categories of income.
The conduit principle was first applied in South Africa in the Armstrong[4] and Rosen[5] cases to protect the legislative choice that dividends should retain their exempt status in the hands of beneficiaries. The court noted that this was not an issue in the present matter as trusts were taxed at twice the CGT rate as individuals, and in fact permitting a discretionary trust to attribute a capital gain to a beneficiary where it would be taxed at half the rate appeared to subvert the legislative intent.[6] Importantly, the court noted the following:[7]
'When a taxation statute addressed either of these issues directly, the case no longer became an exercise in applying the conduit principle. Instead, it became an exercise in giving effect to the direct legislative intention expressed in the statute.'
The court then noted that in 1991 the legislature had amended the definition of 'person' to include a trust and inserted s 25B.
It stated:[8]
'Since 1991, questions relating to the taxation of trusts and beneficiaries under the ITA have accordingly become questions of the interpretation of the relevant provisions of the ITA that deal directly with trusts and beneficiaries. Common law principles relating to the conduit principle may inform these questions of interpretation, particularly where the ITA does not expressly regulate the respective tax treatment of trusts and beneficiaries. However, the exercise remains primarily one of statutory interpretation.'
The court then rejected the argument that s 25B could apply to capital gains, stating that
'paragraph 80 addresses itself pertinently to the conduit principle and the liability for taxation on capital gains realised by the sale of assets by a trust. Therefore, it is the specific provision that applies'.
It noted[9] that if s 25B overrode para 80, it would be tautologous (saying the same thing twice in different words) and that there was a presumption against tautology in a statute.[10]
'Paragraph 80 must have been included in the Eighth Schedule for some purpose. It cannot be interpreted as though everything that it provides is to be rendered irrelevant because the pre-existing deeming provision in section 25B overrides paragraph 80.'
Before 2008, para 80(2) referred to 'where a capital gain arises in a trust'. It was amended by the Revenue Laws Amendment Act 60 of 2008 to refer to 'where a capital gain is determined in respect of the disposal of an asset by a trust'.
In 2007 a tax consultant emailed me regarding the issue of whether a gain could bounce through several trusts. His view was that it could not, and my colleague on the Legislative Research Section who had drafted para 80(2) agreed with him. When compiling the first issue of the
Comprehensive Guide to Capital Gains Tax, I reflected this view on the basis that 'arise' meant 'originate'. The capital gain thus originated in the trust that disposed of the asset and it was only that trust that was permitted to attribute the capital gain. Nevertheless, I thought that para 80(2) could have been clearer.
In the Revenue Laws Amendment Act 60 of 2008 several amendments were made to para 80. These included removing exempt bodies from attribution under para 80(1) and a small clarifying amendment to para 80(3). The opportunity was also taken to amend para 80)2) to more accurately reflect SARS's view that the gain could be attributed only by the trust that disposed of the asset.
The Explanatory Memorandum made it clear that the interpretation that a gain could flow through multiple discretionary trusts was not accepted and the purpose of the amendment was to clarify para 80(2).
The court noted that it was acceptable to consult the Explanatory Memorandum (EM) to ascertain the purpose of the 2008 amendment.[11] The judgment provides a useful summary of cases in which the courts have relied on Explanatory Memoranda to ascertain the purpose of legislation.
The only problem, of course, was that the EM did not actually spell out the policy reason why attribution was restricted to the trust that disposed of the asset. When pressed on the point, SARS's counsel declined to offer any suggestion as to what the purpose might be.
I do not recall the purpose of the amendment ever being discussed during the drafting process. If a policy reason was to be provided, this should have been disclosed in the initial EM in 2001, or if not, in the 2008 EM. In my view, the reason for the amendment in 2008 was simply to align para 80(2) with para 80(1). Paragraph 80(1) also permits the capital gain to be attributed only once by the trust that vested the asset in a resident beneficiary. One could of course speculate on policy reasons why para 80 opens the conduit pipe only for the immediate beneficiary. The constitutional court suggested it might have been to prevent a capital gain from flowing to another trust having a capital loss available for set-off.[12] Another reason might be to restrict opportunities for income splitting or to make it easier for SARS to audit a capital gain without going through an extensive tracing exercise. The Davis Tax Committee noted that[13]
'a substantial risk of non-disclosure arises when income is taxed in the hands of a taxpayer who is different to the taxpayer receiving it or to whom it accrues'.
SARS has been addressing the problem of non-disclosure by beneficiaries by introducing the IT3(t) return for trusts which will eventually lead to the beneficiary return being prepopulated.
In the result, the majority decision was that para 80(2) was unambiguous and clear after the 2008 amendment.[14]
The minority judgment
In the dissenting judgment, Bilchitz AJ stated:[15]
'The text, purpose, context and presumptions of statutory interpretation require construing the provision to give full effect to the conduit principle such that capital gains are taxed in the hands of the ultimate beneficiaries.'
In my view, the dissenting judgment failed to appreciate the effect of including a trust in the definition of 'person' in s 1 in 1991. Once a trust became a person for tax purposes, it had to comply with all the normal rules that apply to every other taxpayer. It was now the owner of its own assets and no longer a mere conduit pipe. Any common law conduit principle ceased to exist. It was precisely for this reason that s 25B was inserted into the ITA at the same time so that the conduit principle could continue. I have often been asked why para 80 differs from s 25B. The more pertinent question should be why s 25B differs from para 80, which was the later provision.
The intention when drafting the Eighth Schedule in 2000 was to mirror what happened in s 25B and s 7, subject to certain modifications. One of those differences was that para 80 did not make provision for capital gains to be attributed to non-resident beneficiaries, while s 25B contained no such restriction. Another difference was that s 25B permitted income to flow through multiple trusts, while para 80 allowed attribution only once.
In 2000 the drafting team, of which I formed part, was under a lot of pressure to finalise the CGT legislation before 1 April 2001 (the original effective date), which was later postponed to 1 October 2001, and a policy decision was taken not to interfere with s 25B. In the meanwhile, South Africa had introduced the residence basis of taxation in 2000, and this was one reason why para 80 did not provide for attribution to non-residents. It would be much harder to collect CGT from a non-resident.
Section 25B was finally amended in 2023[16] to prevent attribution to non-resident beneficiaries. I do not know why para 80(1) and (2) were drafted to block attribution through multiple discretionary trusts. There may well be sound reasons for the policy as suggested, and the policy is hardly absurd. This is simply a policy choice for government. As was stated in
Endumeni,[17]
'Judges must be alert to, and guard against, the temptation to substitute what they regard as reasonable, sensible or businesslike for the words actually used. To do so in regard to a statute or statutory instrument is to cross the divide between interpretation and legislation …'
Bilchitz AJ stated that legislation must be rational and non-arbitrary and be construed in a way that is consonant with a legitimate government purpose.
The judge then sought to highlight perceived ambiguities in para 80(2) which would justify the application of the
contra fiscum rule.
Paragraph 80(2) might not, at the time, have been 'a model of clear legal drafting' (to cite Bilchitz AJ), but it was not that bad that its intention could not be discerned. The 2008 amendment at least clarified that under both para 80(1) and (2) the capital gain could be attributed only once. I would take issue with the judge's comment that the 2008 amendment was intended to distinguish para 80(1) from para 80(2) because the former dealt with the vesting of an asset while the latter dealt with a disposal of an asset. The vesting of an asset is also a disposal under para 11(1)(d).
If SARS and National Treasury take anything away from the dissenting judgement, it is that legislation needs to be clearly explained in detail together with its policy rationale in the EM. This disclosure is not always easy as the full consequences, sometimes unanticipated, of legislation emerge only years later.
The cross appeal
The court refused to hear the cross appeal because the matter had not been properly argued in the lower courts and it did not want to be the court of first and last instance over the issue whether the taxpayer had committed a
bona fide inadvertent error.[18] The court observed that the penalties had not been considered by the SCA as SARS had apparently conceded the issue. It also found that SARS had no prospect of discharging the onus of proving that the taxpayer's behaviour fell within s 223(1)(ii) and (iii), namely, reasonable care not taken in completing return (ii) and no reasonable grounds for tax position taken (iii).
In my experience, SARS is frequently too zealous in its imposition of penalties. Trying to impose a 50% penalty on what amounted to a question of interpretation over a complex matter is just unreasonable. SARS might have been better served by imposing a 10% penalty for substantial understatement and then arguing that the error in interpretation was not inadvertent. Had the taxpayer wished to be free from all penalties, it should have sought a s 223(3) opinion from a registered tax practitioner. If taxpayers can escape penalties on interpretation issues involving substantial understatement based on a
bona fide inadvertent error, what is the purpose of s 223(3)?
Even so, it does seem harsh to impose a 10% penalty for a substantial understatement involving an interpretation issue when a taxpayer had reasonable grounds for their tax position taken. In this instance, the taxpayer had sought a senior counsel opinion. It is unclear to me why Thistle did not raise s 223(3) as a defence, but perhaps the senior counsel was not a registered tax practitioner.
Finally, it would be interesting to know whether the natural person beneficiaries would be able to have the tax they had paid refunded, particularly if their tax assessments had prescribed. Possibly, the exemption from prescription relating to the resolution of a dispute under s 99(2)(d) of the TAA could be used to avoid double taxation.
Conclusion
In my view the majority decision of Chaskalson AJ was the correct one based on the legislation. It is not always easy to determine the purpose of every piece of legislation. The Constitutional Court must be commended on its judgement, which dealt with a complex area of tax law. It illustrated the importance of having regard to the legislative history of a provision.
This article was first published in ASA December/January2025 - ASA Magazine : ASA December_January 2025
