---

FLYNOTES: TAX – Capital gains tax – Conduit principle – Disposal of properties – Trust receiving proceeds as beneficiary of group of vesting trusts – Distributing proceeds to natural persons – Contending that capital gains taxable in hands of ultimate beneficiaries – Conduit principle does not apply throughout multi-tier trust structure – Capital gains are taxable once distributed to second-tier trust – Income Tax Act 58 of 1962, s 25B and para 80(2) of Eighth Schedule.

CONSTITUTIONAL COURT OF SOUTH AFRICA

 

Case CCT 337/22

 

In the matter between:

 

THE THISTLE TRUST                                                                Applicant

 

and

 

COMMISSIONER FOR THE SOUTH AFRICAN

REVENUE SERVICE                                                               Respondent

 

 

Neutral citation:     The Thistle Trust v Commissioner for the South African Revenue Service [2024] ZACC 19

 

Coram:                  Bilchitz AJ, Chaskalson AJ, Madlanga J, Majiedt J, Mathopo J, Mhlantla J, Theron J and Tshiqi J

 

Judgments:            Chaskalson AJ (majority): [1] to [93]

                              Bilchitz AJ (dissenting): [94] to [141]

 

Heard on:              8 February 2024

 

Decided on:            2 October 2024

 

Summary:              Income Tax Act 58 of 1962 — Section 25B — Section 26A — conduit principle — capital gains tax — beneficiaries

 

Tax Administration Act 28 of 2011 — understatement penalties — bona fide inadvertent error

 

 

 

ORDER

 

 

 

On appeal from the Supreme Court of Appeal (hearing an appeal from the Tax Court of South Africa, Gauteng):

1.               The application for leave to appeal is granted.

2.               The appeal is dismissed.

3.               There is no order as to costs in the appeal.

4.               The conditional application for leave to cross appeal is dismissed.

5.               The respondent is ordered to pay the applicant’s costs in the cross appeal, including the costs of two counsel.

 

 

 

JUDGMENT

 

 

CHASKALSON AJ (Majiedt J, Mathopo J, Mhlantla J, Theron J and Tshiqi J concurring):

 

Introduction

[1]            The conduit principle in relation to the taxation of trusts and beneficiaries has been adopted by our courts from English law.  It governs how amounts distributed from a trust to its beneficiaries will be characterised for purposes of taxation.  The conduit principle treats a trust as a conduit for the transfer of taxable amounts into the hands of beneficiaries.  It provides for the amounts in question to be taxed on the basis that their nature, for the purposes of tax law, does not change in the process of distribution from the trust to the beneficiaries, and that they are ordinarily taxed in the hands of the true beneficial owner.

 

[2]         In this matter, this Court must decide how the conduit principle applies to the taxation of capital gains distributed to beneficiaries through multiple trusts in a tiered trust structure.  To do so, we must consider not only the conduit principle but also the interpretation of the relevant provisions of the Income Tax Act[1] (ITA), namely sections 25B and 26A of the ITA read with paragraph 80(2) of the Eighth Schedule (paragraph 80(2)).

 

[3]        Sections 25B and 26A are headed “Income of trusts and beneficiaries of trusts” and “Inclusion of taxable capital gain in taxable income”, respectively.  During the 2014 to 2016 tax years, which is the period relevant to this matter, these sections read as follows:

 

“25B

(1)        Any amount received by or accrued to or in favour of any person during any year of assessment in his or her capacity as the trustee of a trust, shall . . . to the extent to which that amount has been derived for the immediate or future benefit of any ascertained beneficiary who has a vested right to that amount during that year, be deemed to be an amount which has accrued to that beneficiary, and to the extent to which that amount is not so derived, be deemed to be an amount which has accrued to that trust.

(2)        Where a beneficiary has acquired a vested right to any amount referred to in subsection (1) in consequence of the exercise by the trustee of a discretion vested in him or her in terms of the relevant deed of trust, agreement or will of a deceased person, that amount shall for the purposes of that subsection be deemed to have been derived for the benefit of that beneficiary.

. . .

26A      There shall be included in the taxable income of a person for a year of assessment the taxable capital gain of that person for that year of assessment, as determined in terms of the Eighth Schedule.”

 

[4]            In relevant part, paragraph 80(2) read as follows during the 2014 to 2016 tax years:

 

“[W]here a capital gain is determined in respect of the disposal of an asset by a trust in a year of assessment during which a trust beneficiary . . . has a vested interest or acquires a vested interest (including an interest caused by the exercise of a discretion) in that capital gain but not in the asset, the disposal of which gave rise to the capital gain,

the whole or the portion of the capital gain so vested—

(a)        must be disregarded for the purpose of calculating the aggregate capital gain or aggregate capital loss of the trust; and

(b)        must be taken into account for the purpose of calculating the aggregate capital gain or aggregate capital loss of the beneficiary in whom the gain vests.”

 

Parties

[5]            The applicant is the Thistle Trust (Thistle).  Thistle is a registered inter vivos[2] discretionary trust and a South African tax resident.  Thistle is a beneficiary of 10 vesting trusts described 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.  The respondent is the Commissioner for the South African Revenue Service (SARS).

 

Factual background

[6]            In the 2014, 2015 and 2016 tax years, Zenprop disposed of assets and realised capital gains, the proceeds of which it distributed to Thistle.  Thistle, in turn, distributed the proceeds of those capital gains to the natural persons who were its beneficiaries.  The proceeds of the capital gains were all passed through the multi-tiered trust structure to the ultimate beneficiaries within the same tax years in which they were realised.  Acting on legal advice received, Zenprop and Thistle did not account for the capital gains in their tax returns for the 2014, 2015 and 2016 tax years.  They were advised that the relevant amounts were capital gains which, in terms of the common law conduit principle and the relevant provisions of the ITA, were taxable as capital gains in the hands of the ultimate beneficiaries.  The beneficiaries accounted for the capital gains in their tax returns and paid the capital gains tax for which they would have been liable in respect of these capital gains.

 

[7]            In the 2014 to 2016 tax years, the individual beneficiaries were liable for capital gains tax on only 33.3% of their respective net capital gains for each year of assessment.[3]  As an inter vivos trust, Thistle was liable for capital gains tax on 66.6% of its net capital gain for each year of assessment.[4]

 

[8]            SARS conducted a tax audit of Thistle.  It took the position that on a proper application of the ITA, liability for the capital gains realised by Zenprop had passed from Zenprop to Thistle as the direct beneficiary of Zenprop, but did not pass further from Thistle to its beneficiaries.  It accordingly held Thistle liable for capital gains tax in respect of the amount of the capital gains distributed to it by Zenprop.  On 21 September 2018, SARS raised additional assessments in which it claimed capital gains tax from Thistle for these amounts.  The additional assessments raised by SARS also imposed understatement penalties on Thistle in respect of the undeclared capital gains tax.

 

[9]            On behalf of Thistle, its attorneys objected to the additional assessments.  In the objection, Thistle’s attorneys stated—

 

“having regard to the provisions of section 25B of the ITA and paragraph 80(2) of the Eighth Schedule to the ITA . . . the capital gains . . . ought not to have been taxed as our client derived no taxable income in this regard, and such gains were properly taxable in the hands of our client’s beneficiaries under those provisions of the ITA mentioned.”

 

[10]        In addition to its primary objection to the additional assessment, Thistle also objected to the imposition of understatement penalties in the assessment.  Thistle contended that, even if the additional assessment was correct, its failure to account for these capital gains in its tax returns was a bona fide (good faith) inadvertent error within the meaning of section 222(1) of the Tax Administration Act[5] (TAA) and therefore could not give rise to understatement penalties.

 

[11]        SARS disallowed Thistle’s objection.  In March 2021, Thistle appealed to the Tax Court, challenging the additional assessments raised by SARS.

 

Litigation history

Tax Court

[12]        By the time of the hearing before the Tax Court, section 25B(1) of the ITA had been amended by section 28 of the Taxation Laws Amendment Act[6] (2020 Amendment Act) to read as follows:

 

“Taxation of trusts and beneficiaries of trusts

(1)        Any amount (other than an amount of a capital nature which is not included in gross income or an amount contemplated in paragraph 3B of the Second Schedule) received by or accrued to or in favour of any person during any year of assessment in his or her capacity as the trustee of a trust, shall, subject to the provisions of section 7, to the extent to which that amount has been derived for the immediate or future benefit of any ascertained beneficiary who has a vested right to that amount during that year, be deemed to be an amount which has accrued to that beneficiary, and to the extent to which that amount is not so derived, be deemed to be an amount which has accrued to that trust.

(2)        Where a beneficiary has acquired a vested right to any amount referred to in subsection (1) in consequence of the exercise by the trustee of a discretion vested in him or her in terms of the relevant deed of trust, agreement or will of a deceased person, that amount shall for the purposes of that subsection be deemed to have been derived for the benefit of that beneficiary.”  (Emphasis added.)

 

[13]        The Tax Court held that the amended wording of section 25B could not be read retrospectively to inform the proper interpretation of the section during the 2014 to 2016 tax years.  It emphasised the wide and unqualified meaning of the words “any amount” in subsections (1) and (2) of section 25B in its form during the 2014 to 2016 tax years.  It interpreted these words to include capital gains and accordingly held that section 25B applied to the taxation of the relevant capital gains.  Relying on section 25B and the Armstrong[7] and Rosen[8] decisions of the Appellate Division which introduced the conduit principle into South African law, the Tax Court held that the capital gains were not taxable in the hands of Thistle, but were taxable as capital gains in the hands of the beneficiaries.  Therefore, it upheld Thistle’s appeal.

 

Supreme Court of Appeal

[14]        SARS appealed to the Supreme Court of Appeal.  That Court upheld the appeal on the primary liability of Thistle for capital gains tax, but it dismissed SARS’ claim for understatement penalties.[9]

 

[15]        The Supreme Court of Appeal noted that, while Rosen had established that the conduit principle was of general application in tax law, Rosen had also cautioned that it ought only to be applied in appropriate circumstances.[10]  The Supreme Court of Appeal held that the facts of the present case did not present appropriate circumstances for the application of the conduit principle.[11]

 

[16]        Relying on its judgment in Milnerton Estates,[12] the Supreme Court of Appeal confirmed that the Eighth Schedule of the ITA was to be treated as providing a self contained method for determining matters relating to the capital gains that had to be included in a taxpayer’s taxable income.[13]  It pointed out that when section 25B was introduced in the Income Tax Act in 1991, capital gains tax did not exist in South Africa.  From this, it concluded that section 25B was not intended to apply to capital gains and that the reference to “any amount” in section 25B did not include taxable capital gains.  Flowing from this conclusion, it held that the treatment of Thistle’s tax liability was to be determined only in accordance with paragraph 80(2).

 

[17]        The Supreme Court of Appeal upheld SARS’ argument that the capital gains realised by the disposal of properties by Zenprop were taxable in the hands of Thistle and not in the hands of the ultimate beneficiaries.  This, so it held, flowed from the fact that Thistle had not itself disposed of any capital asset or determined any capital gain.  Thistle had only distributed moneys that vested in it from Zenprop as of right and in these circumstances the conduit principle did not apply in terms of paragraph 80(2).

 

[18]        Although the Supreme Court of Appeal found that SARS was correct to raise the additional assessment imposing capital gains tax on Thistle in respect of the 2014 to 2016 tax years, it held that Thistle could not be liable for understatement penalties.  In its judgment, it stated that SARS had conceded at the hearing that the understatement by Thistle was a bona fide inadvertent error.[14]  In terms of section 222 of the TAA, this precluded the imposition of any understatement penalties.

 

In this Court

[19]        Thistle applies for leave to appeal against the decision of the Supreme Court of Appeal.  SARS has filed a conditional counter-application for leave to appeal against the decision of the Supreme Court of Appeal in respect of understatement penalties.  The counter-application is conditional upon Thistle’s appeal failing.

 

Thistle’s submissions

Jurisdiction

[20]        Thistle originally submitted that this matter engages the jurisdiction of this Court by invoking both a constitutional issue and an arguable point of law of general public importance which ought to be considered by this Court.  The constitutional issue upon which Thistle relied was an issue of retrospectivity and its implications for the rule of law.  In this regard, it argued that the judgment and order of the Supreme Court of Appeal retrospectively applied the 2020 amendment to section 25B of the ITA to the tax dispute which concerned the 2014 to 2016 tax years.

 

[21]        In arguing its jurisdiction case at the hearing, however, Thistle relied less on the retrospectivity point and more on the argument that this case raises an arguable point of law of general public importance which ought to be considered by this Court.  The point of law concerns the proper interpretation of section 25B and paragraph 80(2) against the application of the common law conduit principle.  Thistle submits that this is a point of law of general public importance, because it will affect the capital gains tax liability of all trusts in tiered trust structures in respect of tax years prior to the amendment of section 25B of the ITA by the 2020 Amendment Act.

 

Merits

[22]        Thistle argues that liability for the capital gains tax lies with the individual beneficiaries in terms of the common law conduit principle, the provisions of section 25B of the ITA and the proper application of paragraph 80(2) of the ITA.

 

[23]        Thistle traces the history of the conduit principle since its introduction into South African law in 1938 and relies on Armstrong and Rosen.  It argues that the conduit principle is a rule of common law that applies to the taxation of trusts.  Therefore, it must not only inform the interpretation of the relevant provisions of the ITA but also apply to the taxation of the relevant capital gains, unless the ITA has clearly excluded or qualified such application.

 

[24]        Thistle contends that there is nothing in the ITA that excludes or qualifies the application of the conduit principle to the capital gains in this case.  It takes issue with the emphasis of the Supreme Court of Appeal on the fact that Thistle had not disposed of any asset itself and disputes that Thistle had not determined any capital gain.  Regarding the latter, it points to the wide meaning of “determined” as it is used in the Eighth Schedule and emphasises that the conduit principle means that the proceeds of the sale of an asset by a trust retain their character as capital gains after they have been distributed to the beneficiaries of that trust.

 

[25]        Apart from the conduit principle, Thistle relies on the deeming provision in section 25B of the ITA.[15]  It argues that in terms of section 25B the capital gain of Zenprop is deemed to be the capital gain of Thistle when it was distributed to Thistle and then deemed to be the capital gain of the beneficiaries when it was distributed further from Thistle to the beneficiaries.  It contends that even if section 25B was introduced into the ITA prior to capital gains tax, “any amount” in section 25B(1) and (2) must now be interpreted to include capital gains.  In this regard, Thistle emphasises not only the wide meaning of the words “any amount” but also the fact that section 26A of the ITA expressly includes taxable capital gains in the taxable income of a taxpayer.[16]

 

[26]        Thistle argues that the answer to the question in the present case is to be found in section 26A, read with section 25B.  It maintains that SARS is wrong to focus on paragraph 80(2), because section 26A is the taxing provision and the purpose of the Eighth Schedule is merely to quantify the amount of capital gains tax, and not to allocate liability to particular taxpayers for payment of that tax.

 

[27]        Finally, Thistle argues that even without section 25B, paragraph 80(2) entitles it not to be taxed on the relevant capital gains.  This is because paragraph 80(2) must be interpreted as an attempt to codify the conduit principle.  Thus, by application of the conduit principle, when a capital gain is attributed from Zenprop to its beneficiary, Thistle, which is itself a trust, the conduit is not blocked, but continues to allow that capital gain to be distributed from Thistle to its beneficiaries in whose hands it will be taxed as a capital gain.

 

[28]        Such an interpretation, Thistle argues, flows both from the application of the conduit principle and the wide meaning of “determined” in the Eighth Schedule.  Accordingly, when the conduit principle applies to the distribution of a capital gain from Zenprop to Thistle, a capital gain is determined in the tax accounts of Thistle.  In terms of the wording of paragraph 80(2)(a) and (b), when Thistle distributes the capital gain so determined to its beneficiaries, it must be disregarded for the purposes of calculating the aggregate capital gain or loss of Thistle, but must rather be taken into account in determining the aggregate capital gains or losses of the beneficiaries.

 

SARS’ submissions

Jurisdiction

[29]        SARS submits that leave to appeal should be refused because no constitutional issue is raised, the appeal is untenable on its merits and it is not in the interests of justice to grant leave to appeal.  It maintains that the Supreme Court of Appeal did not interpret the ITA retrospectively.

 

Merits

[30]        SARS submits that section 25B does not apply to capital gains, only to other income that is relevant for income tax purposes.  It emphasises that section 25B was introduced into the ITA at a time when capital gains tax did not exist in South Africa and accordingly could not, originally, have been intended to apply to capital gains.  Instead, section 26A and the Eighth Schedule to the ITA should be interpreted to make clear that all matters relating to the calculation of the taxable capital gain of a trust are to be determined in accordance with the Eighth Schedule.

 

[31]        SARS points out that paragraph 80(2) contains its own codification of the conduit principle which differs from that found in section 25B.  It argues that paragraph 80(2) makes clear that the conduit principle cannot operate beyond the first beneficiary trust in a multi tiered trust structure.  In support of this argument, it highlights the differences between the wording of paragraph 80(2) and section 25B.  It also relies on the explanatory memorandum to the Revenue Laws Amendment Bill of 2008 (2008 explanatory memorandum) which indicates that the purpose of the amendment to paragraph 80(2) by the Revenue Laws Amendment Act 60 of 2008 (2008 Amendment) was to ensure that a second level trust in a tiered trust structure could not avoid liability for capital gains tax on the proceeds of a capital gain it received from its vesting trust, by distributing the relevant amount to its beneficiaries.

 

Cross appeal

SARS’ submissions

[32]        In its conditional cross appeal, SARS denies that it made the concession attributed to it in the judgment of the Supreme Court of Appeal.  It accepts, however, that its counsel did not advance the argument before the Supreme Court of Appeal on the issue of whether Thistle’s failure to account for the capital gains it distributed to its beneficiaries amounts to a bona fide inadvertent error within the meaning of section 222 of the TAA.  In respect of that issue, SARS submits that Thistle did not have reasonable grounds for its reliance on its tax position.  As it intentionally adopted this position, its “error” cannot be described as a bona fide inadvertent error and it should be held liable for the understatement penalties.

 

[33]        SARS argues that Thistle’s understatement should be treated as one that is subject to penalties in terms of item (iii), alternatively item (ii) of the table in section 223(1) of the TAA.  These items respectively deal with the following cases—

(a)      “[n]o reasonable grounds for ‘tax position’ taken;” and

(b)      “[r]easonable care not taken in completing return.”

 

Thistle’s submissions

[34]        Thistle submits that it has not made any understatement and so there can be no understatement penalties.  In the alternative, it argues that even if the appeal fails, the cross appeal must be dismissed because any error in its original return falls within the category of “bona fide inadvertent error” in section 222 of the TAA, and accordingly, it is not an error which gives rise to any penalties.

 

Analysis and legal framework

Main application

Jurisdiction and leave to appeal

[35]        Thistle’s application for leave to appeal engages this Court’s general jurisdiction in terms of section 167(3)(b)(ii) of the Constitution as the application raises arguable points of law of general public importance.  The points of law raised concern the proper interpretation of section 25B and paragraph 80(2) and the application of the common law conduit principle.  As Thistle submits, these points of law are of general public importance because they will affect the capital gains tax liability of trusts in tiered trust structures in respect of all tax years up to 2021.[17]  They will also have implications for other trusts and their beneficiaries in cases that are affected by the application of the conduit principle.

 

[36]        Thistle’s proposed appeal will determine the outcome of a multi million rand tax dispute.  The issues that it raises are of general public importance and transcend the interests of the parties to the dispute.  They have implications for the tax liability of trusts and beneficiaries in countless other disputes.  The arguments Thistle raises are substantial.  These arguments were upheld by the Tax Court before its decision was overturned by the Supreme Court of Appeal.  With two competing decisions, it is accordingly clear that the interests of justice require leave to appeal to be granted.

 

Merits of the main application

The origins of the conduit principle and its incorporation into South African law

[37]        As stated above, the conduit principle has been adopted into our law from English common law.  Its origins are usually traced to the judgment of the Privy Council in Syme,[18] where the Privy Council considered the taxation of income derived from a newspaper business owned by a vesting trust whose trustees immediately distributed the profits of the newspaper business to the beneficiaries of the trust.  The relevant tax statute of the Australian State of Victoria taxed income “derived from personal exertion” at a lower rate than income derived from property.  It was common cause that, in the hands of trustees, the profits of the newspaper business would have been characterised as income “derived from personal exertion”.  The Privy Council had to decide whether this income lost its tax privileged status once it was distributed to the beneficiaries.

 

[38]        In Syme, Lord Sumner made the following statements which are generally understood to be the first formulation of the conduit principle:

 

“It does not follow when the appellant receives the cheque for his share . . . that the connection between his income and the newspaper business is lost.

. . .

What was the produce of personal exertion in the trustee’s hands till they part with it does not, in the instant of transfer, suffer a change, and become the produce of property and not of personal exertion, as it passes to the hands of the cestui que [(beneficiary)] trust.”[19]

 

[39]        Following Syme, various English,[20] Australian,[21] and Canadian[22] courts adopted similar approaches to the taxation of trustees and beneficiaries.

 

[40]        The first significant South African judgment to apply the conduit principle was Armstrong.  There, the Appellate Division held that dividends paid to a trust and distributed to the beneficiary of the trust did not lose their character as dividends through being distributed to the beneficiary.  Accordingly, they were not taxable income in the hands of the beneficiary because dividends were not, at the time, taxable income.  In reaching its conclusion, the Appellate Division invoked the judgment of the Privy Council in Syme and stated that the argument that the distributions deriving from dividends should be treated as taxable income in the hands of the beneficiaries did not accord with the scheme of the then applicable Income Tax Act.[23]  That scheme was—

 

“that income derived from companies should, in the hands of the true recipients of it, be free of the tax which has already been deducted at the source [i.e. through the company tax paid by the company declaring the dividends].  And the clear intention of the Act can only be effectively and generally carried out by exempting the person ultimately receiving such moneys.  In the simple case I am now examining, namely, that of a trio comprising a company, the intervening trustee, and the beneficiary it is manifest that in the truest sense the beneficiary derives his income from the company, for that income fluctuates with the fortunes of the company and the trustee can neither increase nor diminish it, he is a mere ‘conduit pipe’.”[24]

 

[41]        In Rosen, the Appellate Division held that Armstrong did not merely interpret the relevant provisions of the Income Tax Act.  Rather, it established the conduit principle as a common law principle applicable to the taxation of trusts and beneficiaries where appropriate, albeit one that was always subject to a contrary intention in the proper construction of the revenue statute.[25]  The Appellate Division stated:

 

“The [conduit] principle rests upon sound and robust common sense; for, by treating the intervening trustee as a mere administrative conduit pipe, it has regard to the substance rather than the form of the distribution and receipt of the dividends.”[26]

 

[42]        A review of the Commonwealth and South African cases shows that the conduit principle was developed to address two separate issues in the context of tax statutes that did not address these issues directly.  The first issue concerned the identification of the taxpayer who was liable to taxation on particular income – was it to be the trustee or the beneficiary?  In that context, the conduit principle was used as a mechanism to ensure that income of a particular nature was taxed in the hands of its true beneficial owner.[27]

 

[43]        The second issue was to protect legislative choices in respect of the favourable or prejudicial income tax treatment of particular categories of income.  In this regard, the conduit principle operated to ensure that income of a particular nature that was earned by a trust and distributed to its beneficiaries did not lose its tax privileged or tax prejudiced nature in the process.  Thus, Armstrong and Rosen involved the income tax exemption then in place in respect of dividends.  In both these cases, dividends distributed to the beneficiaries by the vesting trusts that received the dividends as shareholders did not lose their tax exempt status in the hands of the beneficiaries.  Similar concerns are evident in the Commonwealth decisions.[28]

 

[44]        For present purposes, it is important to emphasise two points.  First, when referring to the conduit principle as being based on “robust common sense”, the Appellate Division in Rosen was dealing with a situation where application of the conduit principle was necessary to protect a legislative choice to treat dividends as non taxable income.  In the present case, there is no issue of any need to protect a legislative choice as to the favourable or prejudicial income tax treatment of particular categories of income or accruals.  On the contrary, absent a clear indication to the contrary in the ITA, “robust common sense” would militate against the application of the conduit principle to the capital gains distributed by a trust.  This is because the legislature has chosen to tax the capital gains of a trust at twice the rate of those of an individual.[29]  Application of the conduit principle to treat capital gains that are distributed on a discretionary basis from a trust to a natural person as capital gains taxable in the hands of the natural person, not the trust, would appear to subvert the legislative intention of taxing capital gains realised by trusts at the higher rate.

 

[45]        Second, the South African and Commonwealth judgments used the conduit principle to answer questions of which taxpayer was to be taxed on particular income and whether that income retained its tax privileged or tax prejudiced status only because the taxation statutes with which they were concerned did not address these issues directly.  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.[30]

 

[46]        In South Africa, the Income Tax Act of 1991[31] (1991 Act) represents a watershed in relation to the conduit principle.  The 1991 Act, for the first time, introduced into the ITA provisions dealing specifically with the taxation of trusts.  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 provisions of the ITA dealing directly with the taxation of trusts

[47]        The 1991 Act was a legislative response to the decision of the Witwatersrand Local Division in Friedman.[32]  There, the court held that a trust was not a taxable entity under the ITA.  Following Friedman, the ITA was amended by the 1991 Act to address the taxation of trusts directly.

 

[48]        The 1991 Act introduced the following amendments into the ITA dealing with the taxation of trusts—

(a)      the definition of “person” in the ITA was amended to include a “trust fund”;[33] and

(b)      section 25B was inserted into the Act to provide expressly for the application of the conduit principle in relation to the taxation of a “trust fund”.[34]

 

[49]        Further amendments relevant to the taxation of trusts were introduced into the ITA by the Income Tax Act 141 of 1992—

(a)      a definition of “trust” was inserted into the ITA;[35]

(b)      the definition of “person” was amended again so that it now expressly included under subparagraph (c) “any trust”; and

(c)      Section 25B was amended into the form that it retained at the time of the transactions relevant to the present case.[36]

 

[50]        The next major development in the amendment of the ITA relevant to the application of the conduit principle took place in 2001 when capital gains tax was introduced by the Taxation Laws Amendment Act 5 of 2001 which—

(a)            inserted Section 26A into the ITA;[37]

(b)            introduced the Eighth Schedule to the ITA to set out the manner in which a taxable capital gain is to be determined;

(c)            provided in paragraph 10 of the Eighth Schedule that natural persons were to be taxed on 25% of their net capital gain and inter vivos trusts (as part of the residual category of “any other case”) at 50% of their net capital gain;[38] and

(d)            in paragraph 80 of the Eighth Schedule to the ITA, specifically addressed the application of the conduit principle in relation to capital gains tax.

 

[51]        The wording of paragraph 80(2) during the 2014 to 2016 tax years has been set out above.[39]  Prior to 2008, the introductory wording of paragraph 80(2) had stated “where a capital gain arises in a trust”.  The 2008 Amendment replaced this wording with “where a capital gain is determined in respect of the disposal of an asset by a trust”.

 

The correct tax treatment of the proceeds of capital gains realised by Zenprop and distributed to Thistle and then on to its beneficiaries

[52]        SARS argues that section 25B cannot be applied to taxable capital gains because it was introduced at a time when those gains were not taxable.  This argument is unpersuasive.  At all times since capital gains became taxable, section 26A has made it clear that taxable capital gains form part of taxable income.  Accordingly, absent contrary indications in the ITA, section 25B would have to be interpreted on the basis that capital gains are taxable income and fall within the phrase “any amount” in section 25B.

 

[53]        However, there are clear indications in the ITA that the application of the conduit principle to the taxation of capital gains in the hands of trusts and beneficiaries is governed not by section 25B, but by paragraph 80.

 

[54]        As pointed out above, Section 26A states that:

 

“There shall be included in the taxable income of a person for a year of assessment the taxable capital gain of that person for that year of assessment, as determined in terms of the Eighth Schedule.”

 

[55]        If the Eighth Schedule said nothing about liability for the taxation of capital gains arising out of the disposal of assets by trusts, it would have been arguable that section 25B (as a specific provision addressing the conduit principle and the taxation of trusts) should govern the application of the conduit principle to the taxation of capital gains realised by the sale of assets by a trust.  However, 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.  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.  Therefore, paragraph 80 governs how the conduit principle is to be applied to establish which taxpayer is liable for taxation on the capital gains realised by the sale of assets by a trust.[40]

 

[56]        Thistle argues that the Eighth Schedule was added to the ITA only to quantify capital gains, not to determine which taxpayer is liable to be taxed on those capital gains.  It is correct that paragraph 80 is not a stand alone provision as SARS argued.  Like all other provisions of the Eighth Schedule, paragraph 80 must be read with section 26A.  As a general rule, the taxing provision is section 26A and the Eighth Schedule concerns itself primarily with questions of the quantification of taxable capital gains.[41]  However, paragraph 80 is a provision of the Eighth Schedule that clearly goes beyond questions of quantification.  It seeks to identify the taxpayer who is liable for capital gains tax on a capital gain realised by the disposal of an asset by a trust and distributed to a beneficiary in the same year of assessment in which the disposal took place.  It must be interpreted accordingly.

 

[57]        As has been pointed out above, in the tax years 2014 to 2018, paragraph 80(2) stated the following in relevant part:

 

“(2)      [W]here a capital gain is determined in respect of the disposal of an asset by a trust in a year of assessment during which a trust beneficiary . . . has a vested interest or acquires a vested interest (including an interest caused by the exercise of a discretion) in that capital gain but not in the asset, the disposal of which gave rise to the capital gain,

the whole or the portion of the capital gain so vested—

(a)        must be disregarded for the purpose of calculating the aggregate capital gain or aggregate capital loss of the trust; and

(b)        must be taken into account for the purpose of calculating the aggregate capital gain or aggregate capital loss of the beneficiary in whom the gain vests.” [42]  (Emphasis added.)

 

[58]        Applying paragraph 80(2) to the present case, we see the following:

(a)            Zenprop (which is a group of trusts) disposed of an asset and determined a capital gain which vested in Thistle (which is also a trust) and which, in turn, distributed the amount of that capital gain to Thistle’s beneficiaries.

(b)            Zenprop disposed of the asset and a capital gain was determined in respect of that disposal.  Thistle is the beneficiary of Zenprop.  Therefore, the capital gain had to be taken into account in determining the aggregate capital gain of Thistle and disregarded for the purposes of determining Zenprop’s aggregate capital gain.

(c)            In terms of paragraph 6 of the Eighth Schedule, the relevant capital gain was therefore included as part of Thistle’s aggregate capital gain.[43]

(d)            Thistle then vested the amount of the capital gain in its beneficiaries.  However, Thistle had not realised the capital gain by disposing of an asset; Zenprop had disposed of the asset.  Therefore, Thistle could not be “the trust” referred to in subparagraph (a) of paragraph 80(2).  Zenprop was the only trust that could be “the trust” contemplated in subparagraph (a).

(e)            Consequently, Thistle could not receive the benefit of having the capital gain disregarded for the purposes of the determination of its aggregate capital gain.

 

[59]        Thistle argues that paragraph 80(2) was capable of an interpretation that allowed Thistle to escape liability for capital gains tax by distributing it to its beneficiaries in the same tax year as it was distributed to Thistle.  In this regard, Thistle emphasises the wide meaning of “determined” in the Eighth Schedule.[44]  It argues that the capital gain distributed to it by Zenprop could be said, through the operation of the conduit principle and paragraph 80(2), to have given rise to a capital gain determined in the accounts of Thistle.  Accordingly, Thistle could be seen as “the trust” referred to in subparagraph (a) when it distributed that capital gain to its beneficiaries.

 

[60]        Thistle is correct that, given the wide meaning of the word “determined” in the Eighth Schedule, the effect of paragraph 80(2) is that a capital gain is determined in the accounts of Thistle.  However, the flaw in the Thistle argument is that paragraph 80(2) is framed so as to identify “the trust” with reference to the fact that it is the trust that disposed of the asset, and not with reference to the fact that it is a trust in whose accounts the capital gain was determined.  Recognising this problem, counsel for Thistle suggested that paragraph 80(2) should be read as though the phrase “in respect of the disposal of an asset” was a parenthetical clause with commas before and after it.  But there are no commas before or after these words in paragraph 80(2) and there are no indications in the ITA that the relevant phrase should be read as a parenthetical clause.  If a statute is not framed in a form that lends itself to the interpretation desired by a litigant, they cannot ask the Court notionally to perform linguistic surgery on the statute by adding or removing commas until the desired interpretation is achieved.[45]

 

[61]        Thistle’s strained interpretation is also to be avoided, because it is inconsistent with the apparent purpose of the 2008 Amendment to paragraph 80(2), namely to prevent the conduit principle from operating in relation to capital gains beyond the first beneficiary trust in a multi tiered trust structure.

 

[62]        The 2008 Amendment changed the introductory wording of the paragraph from “where a capital gain arises in a trust” to “where a capital gain is determined in respect of the disposal of an asset by a trust”.  Prior to the 2008 Amendment, through the operation of paragraph 80(2), the capital gain realised by the sale of assets by Zenprop and distributed to Thistle could be said to be a capital gain which, after distribution by Zenprop, “arose” in Thistle.  By sequential operation of paragraph 80(2), this capital gain would then have had to be disregarded for the purposes of calculating the aggregate capital gain of Thistle and taken into account for the purposes of calculating the aggregate capital gain of its beneficiaries.

 

[63]        In other words, prior to the 2008 Amendment, paragraph 80(2) provided for the conduit principle to apply through multi tiered trusts all the way to the ultimate beneficiaries.  As we have seen above, following the 2008 Amendment, paragraph 80(2) prevented the conduit principle from operating beyond the first beneficiary trust in a multi tiered trust structure.  Therefore, on a linguistic analysis of the 2008 Amendment, its clear purpose was to confine the operation of the conduit principle in this fashion.  If the 2008 Amendment is interpreted in the manner urged by Thistle, it is difficult to identify any change that the amendment made to the meaning of paragraph 80(2) or any other purpose served by the 2008 Amendment.

 

[64]        The purpose attributed above to the 2008 Amendment is confirmed by the 2008 explanatory memorandum.  It stated the following in respect of the amendment that was ultimately made to paragraph 80(2) with the enactment of the 2008 Amendment:

 

“Some commentators have suggested that a capital gain arising under paragraph 80(2) can be attributed through multiple discretionary trusts.  This view has not been accepted and the amendment clarifies this by referring to a capital gain determined in respect of the disposal of an asset by a trust instead of a capital gain arising in a trust.”  (Emphasis added.)

 

[65]        In New Clicks,[46] Chaskalson CJ stated:

 

“In S v Makwanyane and Another I had occasion to consider whether background material is admissible for the purpose of interpreting the Constitution.  I concluded that

‘where the background material is clear, is not in dispute, and is relevant to showing why particular provisions were or were not included in the Constitution, it can be taken into account by a Court in interpreting the Constitution’.

Although it is not entirely clear whether the majority of the Court concurred in this finding, none dissented from it.  I have no reason to depart from that finding and, in my view, it is applicable to ascertaining ‘the mischief’ that a statute is aimed at where that would be relevant to its interpretation.  This would be consistent with the decisions of the Appellate Division in Attorney General, Eastern Cape v Blom and Others and Westinghouse Brake & Equipment (Pty) Ltd v Bilger Engineering (Pty) Ltd and the cases from other jurisdictions referred to in Makwanyane’s case.”[47]

 

[66]        Since New Clicks, this Court has frequently had regard to explanatory memoranda to bills in the process of identifying the purpose of a statute or an amendment to a statute.[48]  So too, has the Supreme Court of Appeal, including in numerous cases involving revenue statutes.[49]

 

[67]        There is a limit to the weight that can be placed on an explanatory memorandum for the purposes of interpreting a statute.  The rule of law dictates that the law should be certain and predictable so that individuals are able to organise their affairs around the law and individuals must have ready access to the law for that purpose.[50]  In order to be predictable, the law must first be accessible.[51]  If the meaning of a law depends entirely on historical research into what was and was not said in an explanatory memorandum issued decades earlier and not easily capable of identification and location, that undermines accessibility of the law and will potentially undermine the rule of law.

 

[68]        Taxation legislation represents a special category of laws in respect of which people proactively organise their affairs to conform to the predictable consequences of the law.  It might therefore be thought that particular caution should be applied before using explanatory memoranda to inform the interpretation of tax laws.  However, both parties before us invoked explanatory memoranda in support of their competing interpretation arguments and the practice of using explanatory memoranda to identify the purpose of revenue statutes is well established.[52]  It is therefore appropriate to have regard to the 2008 explanatory memorandum to identify the purpose of the 2008 Amendment.

 

[69]        To sum up: the wording of paragraph 80(2) shows that the provision applies the conduit principle only to the first beneficiary trust in a multi tiered trust structure.  It is not reasonably possible to interpret paragraph 80(2) to allow the conduit principle to run through a multi tiered trust structure to attribute liability for capital gains tax in respect of the disposal of an asset to a beneficiary beyond the first beneficiary of the trust that realised the capital gain by disposing of that asset.  The legislative history of paragraph 80(2) and the 2008 memorandum both confirm that paragraph 80(2) was amended into its present form for the purpose of preventing the conduit principle operating through multiple discretionary trusts in a tiered trust structure.  Paragraph 80(2) must be interpreted accordingly.

 

[70]        The reasoning above interprets the relevant provisions of the ITA in their form during the 2014 to 2016 tax years without recourse to the 2020 amendment of section 25B of the ITA.  Thistle’s retrospectivity concerns about the Supreme Court of Appeal judgment are accordingly not relevant to this interpretation.

 

The second judgment

[71]        My Colleague, Bilchitz AJ, takes issue with my interpretation of paragraph 80(2).  In his judgment (the second judgment) he raises three different concerns with my interpretation of paragraph 80(2).  First, the second judgment invokes the contra fiscum (presumption that law is not unjust, inequitable or unreasonable) rule of statutory interpretation.  Second, it suggests that the interpretation adopted above is premised on an irrational distinction between the operation of the conduit principle in relation to capital gains distributed through multi tiered trust structures and the operation of the conduit principle in relation to all other forms of income distributed through multi tiered trust structures.  Finally, the second judgment suggests that this interpretation flies in the face of the robust common sense upon which the conduit principle rests.  I respond to each of these concerns in turn.

 

[72]        The second judgment presents the contra fiscum rule as “based upon the idea that no tax can be imposed upon a subject of the [s]tate without words in legislation clearly evincing an intention to lay a burden on him or her”.[53]  Having regard to this foundation of the contra fiscum rule, I have doubts as to whether it is even relevant to the present case.  The rule applies to the interpretation of fiscal statutes to determine whether a particular type of income or activity is subject to tax under the statute.  It is not designed to answer questions as to which taxpayer is going to be held liable for a tax that is unambiguously imposed by the statute.  The present case falls into the latter category, not the former category.  There is no debate whether the capital gain realised by the disposal of assets by Zenprop should be subject to capital gains tax.  The question is whether Thistle or the beneficiaries should be held liable for capital gains tax on the amount in question.

 

[73]        Even assuming that the contra fiscum rule is applicable to the present dispute, it would not, in my view, assist Thistle.  This rule is not a rule of statutory interpretation that applies to override ordinary principles of statutory interpretation.  It is a presumption of statutory interpretation that applies only where ambiguity in fiscal legislation cannot be resolved by the ordinary methods of statutory interpretation.  This has been confirmed most recently by the Supreme Court of Appeal in Telkom,[54] with which the second judgment takes issue.  It was, in fact, also confirmed by the Supreme Court of Appeal in NST Ferrochrome,[55] a case which the second judgment apparently seeks to enlist in support of a stronger application of the contra fiscum rule.  This is clear from the very passage of NST Ferrochrome relied upon in the second judgment,[56] when that passage is read in its full context:

 

“Where there is doubt as to the meaning of a statutory provision which imposes a burden, it is well established that the doubt is to be resolved by construing the provision in a way which is more favourable to the subject, provided of course the provision is reasonably capable of that construction.  (See, for example, Fundstrust (Pty) Ltd (in Liquidation) v Van Deventer 1997 (1) SA 710 (A) at 735G-H; Willis Faber Enthoven (Pty) Ltd v Receiver of Revenue and Another [1991] ZASCA 163; 1992 (4) SA 202 (A) at 216C.)  But, where any uncertainty in a statutory provision can be resolved by an examination of the language used in its context, there is no rule of interpretation which requires that effect be given to a construction which is found not to be the correct one merely because that construction would be less onerous on the subject.”[57]  (Emphasis added.)

 

[74]        For the reasons I have set out above, there is no ambiguity in the meaning of paragraph 80(2) of the sort that would allow recourse to the contra fiscum rule.  The meaning of paragraph 80(2) since the 2008 Amendment is clear when the language of the provision is interpreted in the context of the ITA as a whole and having regard to the clear purpose of the 2008 Amendment.

 

[75]        It is correct that the interpretation that I have adopted above creates a distinction between the operation of the conduit principle under paragraph 80(2) in relation to capital gains distributed through multi tiered trust structures and the operation of the conduit principle under section 25B in relation to all other forms of income distributed through multi tiered trust structures to the ultimate beneficiary that receives the income in the year of assessment.  During the hearing, counsel for SARS was invited to explain the purpose served by such a distinction but he did not take up this invitation.

 

[76]        The second judgment suggests that as SARS failed to offer an explanation for the distinction, “the construction of the provision proposed by [SARS] would render the provision irrational and arbitrary”.[58]  This is unfair to SARS.  Thistle did not allege that if paragraph 80(2) was interpreted to apply the conduit principle to capital gains differently to the manner in which section 25B applied the conduit principle to all other forms of income, this differential treatment would be irrational or otherwise unconstitutional.  As a result, the issue of why section 25B and paragraph 80(2) applied the conduit principle differently was not canvassed on the papers.  SARS was never challenged on the papers to produce evidence to show that there is a rational basis for this differentiation as between income and capital gains, and simple trust structures and multi tiered trust structures.  That being the case, we cannot conclude that the distinction is irrational simply because counsel for SARS failed to offer an explanation for the distinction at the hearing.  At most we can conclude that counsel had understandably failed to prepare for a question on an issue that was not raised on the pleadings, and was therefore unable to answer that question on the spur of the moment in the hearing.

 

[77]        There may well be a rational basis for distinguishing between “ordinary” income and capital gains when it comes to the application of the conduit principle to multi tiered trusts.  For example, the distinction may serve to limit trustees’ capacity to avoid capital gains tax in multi tiered tax structures by making targeted distributions through the structure to “net off” capital losses in the multi tiered trust structure against capital gains in that structure.  Little point is served by speculating further in this regard.  The rationality of the distinction was not canvassed on the papers (or even in the arguments).  It cannot now be invoked by this Court as the basis for an interpretation judgment.[59]

 

[78]        Finally, I take issue with the proposition that robust common sense requires full application of the conduit principle to all situations.  To the extent that the conduit principle rests on robust common sense, it does not assist Thistle at all.  As I have pointed out above, the Commonwealth and South African cases show that the conduit principle was developed to address two separate concerns of “common sense” in the context of tax statutes that did not address these issues directly.[60]  The first was a concern to subject the true beneficial owner of particular income to taxation on that income.  The second was a concern to protect legislative choices in respect of the favourable or prejudicial income tax treatment of particular categories of income.

 

[79]        Absent provisions in the ITA dealing with the application of the conduit principle to the taxation of capital gains realised by the disposal of assets by a trust, neither of these concerns would have assisted Thistle, which is a discretionary trust.  The authorities on the conduit principle consistently refused to apply the principle to the distributions from a discretionary trust to its beneficiaries because the beneficiaries of the discretionary trust were held not to be the true beneficial owners of amounts that vested in the discretionary trust before being distributed to the beneficiaries.[61]

 

[80]        Moreover, as pointed out above, the present case does not involve any need to protect legislative choices in respect of the favourable tax treatment of particular types of income.  The only legislative choice that appears to be relevant in the present case is the legislative choice to tax the capital gains of inter vivos trusts at twice the rate of the capital gains of individuals.  That legislative choice is one which, absent the provisions of paragraph 80(2), would have militated strongly against any application of the conduit principle to capital gains tax.

 

[81]        For these reasons, I am not persuaded by the second judgment to change my interpretation of paragraph 80(2).  The concerns raised in the second judgment appear to me to be misplaced.  The appeal must fail.

 

The cross appeal

[82]        The dismissal of Thistle’s appeal raises SARS’ claim to understatement penalties and the conditional application for leave to cross appeal.

 

Jurisdiction

[83]        The conditional cross appeal engages this Court’s general jurisdiction.  The phrase “bona fide inadvertent error” in section 222 of the TAA is open to different plausible interpretations.[62]  As a result, the dispute over the correct interpretation raises an arguable point of law.  This point of law is of obvious public importance, because it will affect how SARS and the courts approach the imposition of understatement penalties in thousands of future tax cases.  It will also affect the attitude that SARS takes to individual taxpayers who understate their income in even more cases that do not reach the level of disputes before the Tax Court.

 

Leave to appeal in the cross appeal

[84]        Notwithstanding the public importance of determining the proper interpretation of section 222, it is not in the interests of justice to grant leave to appeal.

 

[85]        If this Court is to hand down a judgment on the meaning of “bona fide inadvertent error” in section 222, it will effectively have to do so sitting as the court of first and last instance in relation to this issue.  The Tax Court did not reach the issue of penalties, because it upheld Thistle’s case on the merits.  The Supreme Court of Appeal did not reach the issue of penalties, because SARS did not argue the issue and was understood to have conceded the issue.

 

[86]        It is undesirable for this Court to have to determine a legal point of public importance in a matter where it has no reasoned judgment on the issue from the preceding courts.[63]  If SARS had a strong case in respect of its claim for penalties in this matter, it may nevertheless have been in the interests of justice for this Court to entertain that claim, but SARS has no sustainable case for penalties.

 

[87]        As pointed out above, SARS pins its case for the penalties which it claims to item (iii), alternatively item (ii) of the table in section 223 of the TAA.  These are the categories of “[n]o reasonable grounds for ‘tax position’ taken” and “[r]easonable care not taken in completing return”.  SARS bears the onus of proving the facts that would bring the understatement of Thistle within either of these categories.[64]  It has no reasonable prospects of discharging this onus.

 

[88]        In respect of item (iii), the tax position taken by Thistle in relation to the conduit principle was one taken on legal advice.  It may have been a tax position that this Court has found to be incorrect, but it cannot be said to be a tax position which Thistle had no reasonable grounds to take.  The tax position was not just reasonable, it was a tax position that was upheld by the Tax Court in a reasoned judgment that engaged with the conduit principle and the relevant provisions of the ITA.  To his credit, counsel for SARS declined to submit that there were no reasonable grounds for the Tax Court to have reached the conclusion that it did.

 

[89]        In relation to item (ii), SARS argues that although Thistle was advised on its tax position, the advice Thistle received pointed out that SARS held a contrary view.  On this basis, SARS argues that if Thistle had taken reasonable care in completing its return, it would have ignored the advice given to it and followed the stated SARS position which that advice expressly considered and rejected.  This argument is based on the proposition that no taxpayer can act reasonably on advice that differs from SARS’ statements of its interpretation of tax legislation.  The argument would elevate SARS to the status of an authority that can decree the only reasonable interpretations of tax legislation.  It is an untenable argument.  In Marshall,[65] SARS advanced a similar argument in relation to the relevance of an interpretation note it had issued to explain its view on an issue of VAT law.  This Court rejected that argument in emphatic terms:

 

“Missing from this reformulation is any explicit mention of a further fundamental contextual change, that from legislative supremacy to constitutional democracy.  Why should a unilateral practice of one part of the executive arm of government play a role in the determination of the reasonable meaning to be given to a statutory provision?  It might conceivably be justified where the practice is evidence of an impartial application of a custom recognised by all concerned, but not where the practice is unilaterally established by one of the litigating parties.  In those circumstances it is difficult to see what advantage evidence of the unilateral practice will have for the objective and independent interpretation by the courts of the meaning of legislation, in accordance with constitutionally compliant precepts.  It is best avoided.”[66]

 

[90]        It follows that SARS’ understatement penalties claim will fail on simple factual grounds irrespective of how this Court may determine the meaning of “bona fide inadvertent error”.  In the circumstances it is not in the interests of justice for this Court to sit as court of first and last instance to determine a legal issue that will have no bearing on the outcome of the appeal.  Leave to appeal must therefore be refused in the conditional counter application.

 

Costs

[91]        In Marshall, this Court applied the Biowatch[67] principle in favour of a taxpayer who raised constitutional issues in the context of an application for leave to appeal that did not have good prospects of success.[68]  In the present matter, Thistle has advanced arguments of substance, even if they have not been accepted in this judgment.  One of the issues raised by Thistle was a constitutional issue relating to retrospectivity of statutes and the judgment of the Supreme Court of Appeal.  In view of my conclusions, I have found it unnecessary to address that constitutional issue, but I do not suggest that Thistle acted frivolously in raising it.  In the circumstances, Biowatch applies in favour of Thistle and it should not be ordered to pay the costs of the appeal.

 

[92]        Biowatch does not apply in favour of SARS because it is an organ of state.  SARS must accordingly pay the costs of the cross appeal.  Those costs will include the costs of two counsel.

 

Order

[93]        The following order is made:

1.               The application for leave to appeal is granted.

2.               The appeal is dismissed.

3.               There is no order as to costs in the appeal.

4.               The conditional application for leave to cross appeal is dismissed.

5.               The respondent is ordered to pay the applicant’s costs in the cross appeal, including the costs of two counsel.

 

BILCHITZ AJ (Madlanga J concurring):

 

[94]        I have had the pleasure of reading the judgment authored by my Colleague Chaskalson AJ (first judgment).  The first judgment analyses the language of paragraph 80(2) of the Eighth Schedule of the Income Tax Act[69] (ITA) and finds that it, unambiguously, admits of only one interpretation – that, in relation to capital gains tax, the conduit principle does not apply throughout a multi-tier trust structure and capital gains are taxable once distributed to a second-tier trust.  The first judgment reasons that this interpretation is supported by the text of the provision as well as an explanatory memorandum released by Parliament relating to the relevant amendments to the legislation in 2008.  I, unfortunately, cannot agree with the approach my Colleague adopts to the interpretation of this paragraph.  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.  That interpretation does not arbitrarily block the application of the conduit at the second tier trust or distinguish between capital gains and other taxable amounts without any good reason.

 

[95]        This case raises important questions surrounding the interpretation of fiscal legislation in the constitutional era.  The second interpretation that I argue for is to be preferred in light of the interpretive approach adopted by our courts to statutory interpretation in the constitutional era – for this reason, I proceed as follows.  First, I outline the key principles relating to statutory interpretation and emphasise the important requirement that, where there is ambiguity, statutes should be interpreted to preserve their constitutionality.  Secondly, I indicate how this requirement interacts with the principles that this Court has developed in relation to the rule of law.  In particular, I seek to show how statutory provisions should be interpreted, where reasonably possible to do so, to avoid rendering them arbitrary, or irrational – and, in a manner that discloses a legitimate purpose and that conforms with common sense.  Thirdly, I seek to show how these principles interact with the contra fiscum rule in the constitutional era.  Lastly, I apply these principles to paragraph 80(2) of the Eighth Schedule of the ITA.  I find that there are significant ambiguities in the drafting of the text of this paragraph and that the purpose and context largely support the second interpretation.  Given the existence of two reasonably possible interpretations, the one I prefer is that interpretation which construes the provision in a manner that is rational and non arbitrary – and, in accordance with the contra fiscum rule, in favour of the taxpayer.  I rely on my Colleague’s outline of the background to this dispute, litigation history and the submissions of the parties.

 

Statutory interpretation in the constitutional era

[96]        Given this case concerns the interpretation of key statutory provisions, it is important to commence with the approach our courts have adopted in this regard.  Detailed consideration was given to the question of statutory interpretation in Endumeni,[70] where Wallis JA wrote the following:

 

“Interpretation is the process of attributing meaning to the words used in a document, be it legislation, some other statutory instrument, or contract, having regard to the context provided by reading the particular provision or provisions in the light of the document as a whole and the circumstances attendant upon its coming into existence.  Whatever the nature of the document, consideration must be given to the language used in the light of the ordinary rules of grammar and syntax; the context in which the provision appears; the apparent purpose to which it is directed and the material known to those responsible for its production.  Where more than one meaning is possible each possibility must be weighed in the light of all these factors.”[71]

 

[97]        The approach adopted in Endumeni does not specifically engage with the constitutional context in which statutory interpretation must take place.[72]  Langa DP gave expression to this shift when he wrote the following in Hyundai:[73]

 

“The purport and objects of the Constitution find expression in section 1 which lays out the fundamental values which the Constitution is designed to achieve.  The Constitution requires that judicial officers read legislation, where possible, in ways which give effect to its fundamental values.

. . .

Accordingly, judicial officers must prefer interpretations of legislation that fall within constitutional bounds over those that do not, provided that such an interpretation can be reasonably ascribed to the section.”[74]

 

[98]        More recently, in Cool Ideas,[75] my Colleague Majiedt J brought these various strands of the approach to constitutional interpretation together when he wrote:

 

“A fundamental tenet of statutory interpretation is that the words in a statute must be given their ordinary grammatical meaning, unless to do so would result in an absurdity.  There are three important interrelated riders to this general principle, namely:

(a)        that statutory provisions should always be interpreted purposively;

(b)        the relevant statutory provision must be properly contextualised; and

(c)        all statutes must be construed consistently with the Constitution, that is, where reasonably possible, legislative provisions ought to be interpreted to preserve their constitutional validity.  This proviso to the general principle is closely related to the purposive approach referred to in (a).”[76]

 

Statutory interpretation and the fundamental value of the rule of law

[99]        As is evident from the above quotations, the constitutional context in South Africa has fundamentally shifted the manner in which legislation must be interpreted by judges.  Statutes must be construed in such a way so as to preserve their constitutionality which, for instance, affects the purpose of a provision that may be considered to be legitimate.  As Langa DP wrote in Hyundai (quoted above), a central injunction is for judicial officers to interpret legislation in light of the fundamental values of the Constitution.  Section 39(2) of the Constitution requires a focus on interpretation in light of the spirit, purport and objects of the Bill of Rights.  However, it remains of great importance to interpret legislation in light of other fundamental values too.[77]  In the context of this case, in particular, I draw attention to the fundamental value of the rule of law contained in section 1(c) of the Constitution.[78]

 

[100]     As my Colleague Chaskalson AJ eloquently writes, the rule of law requires that law be certain, predictable and allow individuals to organise their affairs around it.  That, in turn, requires that the law be accessible and as clear as possible in order that people can easily ascertain what the law requires of them.[79]

 

[101]     In addition to these elements, is the importance of rationality.[80]  Sadly, Ackermann J recently passed away – in tribute to his legacy, I quote here his succinct capturing of the nature of the constitutional state he envisaged South Africa as becoming in Makwanyane:[81]

 

“We have moved from a past characterised by much which was arbitrary and unequal in the operation of the law to a present and a future in a constitutional State where State action must be such that it is capable of being analysed and justified rationally.  The idea of the constitutional State presupposes a system whose operation can be rationally tested against or in terms of the law.  Arbitrariness, by its very nature, is dissonant with these core concepts of our new constitutional order.”[82]

 

[102]     Parliament, as the primary legislative organ of a representative democracy, is required to act rationally.  As this Court held in Law Society of South Africa:[83]

 

“The constitutional requirement of rationality is an incident of the rule of law, which in turn is a founding value of our Constitution.  The rule of law requires that all public power must be sourced in law.  This means that [s]tate actors exercise public power within the formal bounds of the law.  Thus, when making laws, the legislature is constrained to act rationally.  It may not act capriciously or arbitrarily.  It must only act to achieve a legitimate government purpose.  Thus, there must be a rational nexus between the legislative scheme and the pursuit of a legitimate government purpose.”[84]

 

[103]     As is evident from the above quotation, the requirement to act rationally involves the following components: (a) the Legislature must not act arbitrarily; (b) a legislative provision must seek to achieve a legitimate government purpose; and (c) there must be a nexus between the legislative provision and the legitimate government purpose.  These requirements in my view, are not only applicable when challenging the validity of legislation, but also to the interpretation thereof.  How then do these requirements interact with the duty on judicial officers to interpret legislation so as to preserve constitutional validity?

 

[104]     The logical consequence of this discussion is that, where it is reasonably possible to do so, provisions in legislation should be interpreted so as to be rational and non arbitrary.  That requires legislation to be construed in a way that is consonant with a legitimate government purpose, and demonstrative of a nexus between the legislative means adopted and its purpose.  Litigants are thus subject to a burden to demonstrate in what way the interpretation they propose construes the provision in such a way that it is rational and non-arbitrary.  That also conforms with their duty to engage with the purpose behind a legislative provision when advancing an interpretation thereof.  Where there are two possible interpretations, preference should be given to an interpretation of legislation that renders provisions non arbitrary and rational rather than one that simply upholds a naked exercise of legislative power.  In short, in the constitutional era, legislation should be interpreted to accord with the requirements that this Court has articulated in relation to the rule of law.  That too harmonises the constitutional imperatives discussed above with the well known common law presumption that statutory law is not unjust, inequitable or unreasonable.[85]

 

Contra fiscum rule, the rule of law and statutory interpretation

[105]     This approach also accords with what has become known as the contra fiscum rule (that legislation must be interpreted against the fiscus).  The rule originated from the idea that legislation giving effect to taxation involves the exercise of significant power over individuals – as a result, just like in criminal matters, the Legislature has a duty to ensure that the law is clear and those subject to the law understand what is required of them.  Where rules are ambiguous, they should be interpreted in favour of the taxpayer.[86]  As held in Glen Anil Development Corporation,[87] the contra fiscum rule is “but a specific application of the general rule that all legislation imposing a burden upon the subject should, in the case of an ambiguity, be construed in favour of the subject”.[88]

 

[106]     The reasoning related to the rule of law provides a strong foundation for this rule in the sphere of taxation.  There is nothing constitutionally suspect about taxation per se: indeed taxation is a feature of all societies and is a duty individuals owe both to the state – to ensure it can perform its functions – and to each other.  In addition to their important social function, tax laws significantly affect how individuals and juristic entities organise their economic affairs.  As such, they must be expressed clearly and in a manner that enables individuals and juristic entities to follow them.[89]  This idea was expressed by Majiedt JA (as he was then) and Davis AJA in the minority judgment in Daikin as follows:

 

“In the case of fiscal legislation, an appropriate standard is the contra fiscum rule which is based upon the idea that no tax can be imposed upon a subject of the [s]tate without words in legislation clearly evincing an intention to lay a burden on him or her.”[90]

 

[107]     This statement follows an earlier recognition of the rule in NST Ferrochrome.[91]  There, the Supreme Court of Appeal stated:

 

“Where there is doubt as to the meaning of a statutory provision which imposes a burden, it is well established that the doubt is to be resolved by construing the provision in a way which is more favourable to the subject, provided of course the provision is reasonably capable of that construction.”[92]

 

[108]     In the more recent case of Telkom,[93] the Supreme Court of Appeal also recognised the existence of the contra fiscum rule in South African law but narrowed its scope significantly.  After having quoted the above dictum from NST Ferrochrome, it then went on to approve of a quotation from a Master’s dissertation.  That quotation, recognised the consistency of the rule with the values of the Constitution.  However, it also stated the following: “to the extent that following analysis, a purposive approach ultimately yields two constructions which are both equally plausible, it is submitted that the contra fiscum rule should apply and the court should ultimately conclude in favour of the taxpayer”.[94]  The requirement here of equal plausibility is in tension with the statements of the rule that simply requires an interpretation to be reasonably possible before it is applied.  Moreover, it is difficult to apply: it will be a rare case where judges will deem two interpretations equally plausible.

 

[109]     There are also different axes upon which plausibility is measured: one interpretation may accord better with the manner in which a provision is phrased; another may give better effect to the context and yet another may better accord with its purpose.  It may be that all are not equally plausible but each may be a reasonably possible interpretation of the statute.  The standard of a “reasonably possible” construction aligns with the dicta in Hyundai and Cool Ideas quoted above.  It is also, in my view, more consistent with the value of the rule of law in requiring Parliament to ensure that fiscal legislation that imposes burdens on subjects is clear, rational and capable of being followed.

 

[110]     There is also no excuse for arbitrary rules in the realm of taxation.  Whilst the Legislature no doubt wishes to raise revenue, specific provisions and distinctions must clearly be capable of justification in realising a legitimate government purpose and being a non-arbitrary and justifiable means to achieve that purpose.  Indeed, in Prinsloo,[95] the Court held as follows regarding the requirement of rationality when differentiation is made between individuals and groups:

 

“In regard to mere differentiation the constitutional state is expected to act in a rational manner.  It should not regulate in an arbitrary manner or manifest ‘naked preferences’ that serve no legitimate governmental purpose, for that would be inconsistent with the rule of law and the fundamental premises of the constitutional state.”[96]

 

Interpreting paragraph 80(2) of the Eighth Schedule

[111]     I agree with my Colleague Chaskalson AJ’s analysis that, in the context of capital gains tax, paragraph 80(2) of the Eighth Schedule (as it read between 2014 and 2016) is the applicable provision to determine in whose hands a capital gain must be taxed.  It is hard to understand why this provision would be necessary if section 25B were directly applicable.  At the same time, as will be discussed further below, where reasonably possible to do so, provisions in tax legislation should be interpreted harmoniously with one another and in a holistic manner, rather than be construed to embody internally inconsistent legal positions.[97]

 

[112]     The question then becomes whether paragraph 80(2) is clear and no interpretation other than the one my Colleague arrives at is reasonably possible.  In my view there is significant ambiguity in paragraph 80(2) when construed in light of the applicable principles and how it applies to multi tier trust structures.  That ambiguity is borne out by the differences between SARS and the legal opinions of senior tax advisors relied on by the applicant as well as academic commentary on the provision which is divided on its interpretation and implications.[98]  I now outline the relevant provisions and then demonstrate why a different interpretation to that adopted in the first judgment should be afforded to the provision.

 

[113]     Paragraph 80(1) – as it read between 2014 and 2016 – was worded as follows:

 

“Subject to paragraphs 68, 69, 71 and 72, where a capital gain is determined in respect of the vesting by a trust of an asset in a trust beneficiary . . . who is a resident, that gain—

(a)        must be disregarded for the purpose of calculating the aggregate capital gain or aggregate capital loss of the trust; and

(b)        must be taken into account for the purpose of calculating the aggregate capital gain or aggregate capital loss of the beneficiary to whom that asset was so disposed of.”

 

[114]     Paragraph 80(2) – as it read between 2014 and 2016 – stated the following:

 

“[W]here a capital gain is determined in respect of the disposal of an asset by a trust in a year of assessment during which a trust beneficiary . . . has a vested interest or acquires a vested interest (including an interest caused by the exercise of a discretion) in that capital gain but not in the asset, the disposal of which gave rise to the capital gain, the whole or the portion of the capital gain so vested—

(a)        must be disregarded for the purpose of calculating the aggregate capital gain or aggregate capital loss of the trust; and

(b)        must be taken into account for the purpose of calculating the aggregate capital gain or aggregate capital loss of the beneficiary in whom the gain vests.”  (Emphasis added.)

 

The conduit principle

[115]     It was common cause that these provisions are clearly designed to apply the conduit principle to capital gains tax.  The first judgment has offered a clear and learned exposition of the background and elements of the conduit principle – it does not, however, engage much with the reasoning and purpose behind the principle.  Armstrong[99] dealt with a company distributing non-taxable dividends to a trust which then distributed them to the main beneficiary.  It was argued that the beneficiary had no direct legal relationship with the company, and so the funds received no longer retained the character of dividends and hence were taxable.  Stratford CJ found that the trust was in fact just a “conduit pipe” to the beneficiary and the dividends retained their tax exempt character.  In making this finding, he essentially identified two rationales.  The first was to prevent double-taxation given that the company had already been taxed before distributing the dividends.[100]  The second was that “in the truest sense the beneficiary derives his income from the company, for that income fluctuates with the fortunes of the company and the Trustee can neither increase nor diminish it, he is a mere ‘conduit pipe’”.[101]  That rationale essentially considers the nature of the intermediary trust as a central consideration in applying the principle.

 

[116]     That rationale was elaborated upon by the Appellate Division in Rosen.[102]  In that case, the Appellate Division had to deal with a distribution of dividends from a discretionary trust to a beneficiary.  Trollip JA articulated the conduit principle as follows—

 

“In effect the Legislature in those provisions has adopted a principle that can be conveniently termed the conduit principle: the registered shareholder is regarded as a mere conduit-pipe for passing the dividends on to the deemed shareholder, the true recipient of them, in whose hands they consequently retain their identity and character as dividends.”[103]

 

[117]     Trollip JA went on to articulate the rationale behind the conduit principle as follows:

 

“The [conduit] principle rests upon sound robust common sense; for, by treating the intervening trustee as a mere administrative conduit pipe, it has regard to the substance rather than the form of the distribution and receipt of the dividends.”[104]

 

[118]     Elaborating upon this reasoning, I would add that the substantive reasoning and common sense involved, emerge from considering the nature of trusts.  The Trust Property Control Act[105] defines a trust as follows:

 

“‘trust’ means the arrangement through which the ownership in property of one person is by virtue of a trust instrument made over or bequeathed—

(a)        to another person, the trustee, in whole or in part, to be administered or disposed of according to the provisions of the trust instrument for the benefit of the person or class of persons designated in the trust instrument or for the achievement of the object stated in the trust instrument; or

(b)        to the beneficiaries designated in the trust instrument, which property is placed under the control of another person, the trustee, to be administered or disposed of according to the provisions of the trust instrument for the benefit of the person or class of persons designated in the trust instrument or for the achievement of the object stated in the trust instrument,

but does not include the case where the property of another is to be administered by any person as executor, tutor or curator in terms of the provisions of the Administration of Estates Act, 1965 (Act 66 of 1965).”

 

[119]     What is immediately evident from this definition is that a trust holds property for the benefit of another person or class of persons.  That fundamental dimension of a trust emerges as well from academic commentary on the common law.  Hahlo, in his seminal article, for instance, writes that “the characteristic feature of the trust is . . . the separation between the control which ownership gives and the benefits of ownership”.[106]  In addition, De Waal writes that “[i]n the most general sense, a trust is an arrangement under which one person is bound to hold or administer property on behalf of another person or for an impersonal object and not for his own benefit”.[107]  The conduit principle essentially recognises this point through embodying the position that, if particular sums of money flow through a trust, as long as they are distributed to the beneficiaries in the same tax year, they are taxed in the hands of the beneficiaries.  This is logical – and prioritises substance over form – in the sense that a trust does not hold the funds it receives for its own purposes but for the purposes of its beneficiaries.  The intervening trusts also add no value to the funds received.  The fact that funds pass through one trust or several trusts is irrelevant to who in fact benefits from those funds.  Once a trust distributes the funds to a beneficiary, it is the beneficiary in whom those funds vest and who should be liable for taxation.

 

[120]     Contrary to the reasoning in the first judgment, this rationale applies equally to a vesting trust and a discretionary trust where a distribution is made in the same tax year.  In a vesting trust, the capital gain will be vested in the beneficiaries once it is realised.  In a discretionary trust, the trustees will have a discretion whether to vest the capital gain in the beneficiaries.  If they fail to do so in a particular tax year, clearly they retain the asset in that trust and it must be taxed in that trust in that year.  If, however, they distribute the capital gain in that tax year to the beneficiary, then they do not hold onto the asset and vest the gain in the actual beneficiary of the trust.  That is what happened in the Rosen case and why Trollip JA referred to the conduit principle as giving effect to the substance rather than the form of the distribution.  In the context of this case, as paragraph 80(2) indicates, what is important is for a beneficiary to have a vested interest in the capital gain – it expressly includes an interest “caused by the exercise of a discretion”.

 

[121]     Whilst the conduit principle was developed by our courts in the past, the Legislature specifically chose to embody the principle in section 25B of the ITA.  When capital gains tax was introduced into South African law, a specific provision – paragraph 80 – applied the conduit principle to capital gains.  There was no compulsion on the legislature to do so – it could have provided that capital gains would be taxed in the hands of the entity which disposes of an asset and realises the capital gain.  The context in which paragraph 80(2) of the Eighth Schedule must be interpreted is thus one in which the Legislature specifically chose to apply the conduit principle to capital gains tax.  It is thus respectful of the legislative intent to apply that principle properly.

 

[122]     The difficulty that has arisen in this case concerns the application of the conduit principle in the context of multi-tier trust structures.  The applicant contends that intervening trusts remain conduits so long as distributions to beneficiaries happen in the same tax year as the capital gain arrives in the account of the intervening trust.  The respondent, however, contends that the conduit is effectively blocked at the first beneficiary to whom the capital gain is distributed – in the case of a multi-tier trust structure, that would render the second-tier trust liable for taxation on capital gains received.  Their argument is rooted in a construction of the language of paragraph 80(2).

 

Text

[123]     I do not consider paragraph 80(2) to be a model of clear legal drafting: difficulties in interpretation arise from the use of the passive voice, indefinite articles, lack of punctuation and complexity of the drafting.  There are two reasonable constructions of the provision: the first, which is the holding of the first judgment, requires the capital gain to be determined in the same trust that disposes of the asset.  The trust referred to in the first line of the provision thus is the first-tier trust and it is the trust referred to in sub-paragraph (a).  The beneficiary would on this reading be the second-tier trust.  This interpretation focuses on reading the words “disposal of an asset” together with “by a trust”, thus linking the capital gain with the disposal of the asset.  The conduit pipe would be blocked once a distribution is made to a second-tier trust and taxation on capital gains in multi-tier structures would take place in relation to second tier trusts and not the ultimate beneficiaries.

 

[124]     In my view, a second plausible reading is to see the provision as applying to any trust – including a second-tier trust – which receives a capital gain from the disposal of an asset.  If that trust distributes the capital gain to a beneficiary, it is only the ultimate beneficiary that is taxed.  That reading requires linking the word “determined” in the first line with “by a trust” which can be any trust (first, second or third-tier) which, in its financials, reports on such a capital gain.  Put differently, “determined” and “by a trust” would link up thus: “where a capital gain is determined . . . by a trust”.  Whether the determination is by a trust in Zenprop’s position (the first-tier), Thistle’s position (the second-tier) or by one further down in the tiered trust structure, it will still be “in respect of the disposal of an asset” as required by paragraph 80(2).  On this reading, the disposal does not have to be done by the same trust as the trust in which the gain is “determined”.  This reading appears to me to be plausible even without the insertion of parenthetical commas after the word “determined” and the word “asset”.

 

[125]     The words “disposal of an asset”, in this context, are critical both to explain how the capital gain arose but, also importantly, in their statutory context, to distinguish paragraph 80(2) from paragraph 80(1).  The latter provision regulates circumstances where a trust vests an asset in a beneficiary and acquires a capital gain in that process; whereas paragraph 80(2) addresses circumstances where a capital gain is realised from the disposal of an asset and distributed to a beneficiary.  The latter provision is simply not clear as to whether the disposal of the asset has to be by the same trust that made the capital gain or whether the reference to disposal of an asset was added by the 2008 Amendment simply to explain the circumstances in which the provision applies and distinguish the provision from paragraph 80(1).  The indefinite article before the first use of the word “trust” confirms the ambiguity relating to which trust in a multi tier structure is being referenced.

 

Purpose

[126]     This reading becomes even more plausible when we consider the purpose of the provision.  It is admitted by the respondent that the goal of the provision is to apply the conduit principle to capital gains from one trust to the beneficiary of that trust.  The respondent, however, contended that the conduit stops at a second-tier trust in a multi tier structure.  It however, made no effort on the papers to justify its reading or suggest any purpose for why the conduit principle should be restricted to the second tier in a multi tier structure.  During the oral hearing, when asked, the respondent’s counsel could not provide any rational basis for this restriction – indeed, as indicated above, the very point of the conduit principle is for tax to be levied on the ultimate beneficiary.  Counsel for the respondent could also not explain why there is a differentiation between capital gains tax – where the conduit principle stops at the second tier trust – and other forms of accruals, such as dividends and interest, for instance, where it does not.  Without any rationale or purpose suggested, the construction of the provision proposed by the respondent would render the provision irrational and arbitrary.

 

[127]     This is not merely, as the first judgment finds, an understandable failure by the respondent’s counsel to respond to a surprise question in an oral hearing.  Instead, it goes to the heart of the approach adopted by SARS throughout when approaching the interpretation of section 25B and paragraph 80(2).  As was indicated above, a central feature of the approach to statutory interpretation in the constitutional era is the need to understand the purpose of a provision and construct the wording in that light.  Where the respondent makes no effort to demonstrate how its construction would realise a legitimate purpose, then it fails to make out a central dimension of its own case.  Without such a purpose or rationale, the reading advocated for by a party becomes arbitrary and irrational.

 

[128]     Indeed, SARS’ responses to the applicant have been replete with a statement of its approach without justifying its stance in terms of any purpose sought to be achieved by its proposed interpretation.  In its letter disallowing an objection to its assessment that the second-tier trust was liable for tax, it stated the following:

 

“If a trust makes a capital gain during the year, and vests it in another trust, paragraph 80 deems the gain to be made by the other trust (beneficiary).  However, paragraph 80 does not apply if this other trust (beneficiary) distributes the gain to its beneficiaries.  This is due to the fact that the second trust did not dispose of the asset and did not make the original capital gain.  The second trust cannot distribute the gain to its beneficiaries for tax purposes.  Even though the beneficiaries may become entitled to the gain in law, the second trust is still taxed on the gain.”

 

[129]     This reasoning makes little sense when one considers that paragraph 80 is a legislative encapsulation of the conduit principle.  The whole point of the principle, as indicated above, is that an intermediary entity which distributes a gain to a beneficiary is a mere conduit and does not hold onto the amount it receives.  If we attempt to apply SARS’ statement to dividends such as in the cases of Armstrong and Rosen, a company obviously generated the dividends and distributed them to a second trust.  If the logic of SARS is to be applied, then they should be taxed at the level of the second trust – but, the conduit principle, that the legislature has enshrined in statute, has recognised that they are taxed in the hands of the ultimate beneficiaries.  There is no attempt to explain why the conduit should be blocked in relation to capital gains but not in relation to dividends or interest.[108]

 

[130]     The interpretation I adopt utilises the rationales behind the conduit principle to understand the meaning of paragraph 80(2).[109]  As was common cause, the Legislature sought to give effect to the conduit principle through this provision.  Given the rationale behind the principle is not to reify intervening trusts but to tax accruals in the hands of the ultimate beneficiaries, there is no good reason why the Legislature should be understood arbitrarily to restrict the operation of the principle to the second-tier trust in a multi-tier trust structure.  If, as the first judgment suggests, the Legislature wished to tax capital gains at the higher rate applicable to trusts, it is unclear why it should have legislatively incorporated the conduit principle at all.  Indeed, had there been no intermediary trust or the gain vested immediately in the beneficiaries were Thistle to have been constituted as a vesting trust, then the capital gain would have been taxed in the hands of the beneficiaries.  If the Legislature had wished to tax capital gains at the higher rate applicable to trusts, then, it failed to adopt an efficient means to achieve that end.  An interpretation of the provision rooted in such a purpose would thus fail to construe the provision in a manner that meets the constitutional standard of rationality.

 

[131]     The first judgment also speculates that the rationale for distinguishing capital gains may be to address tax-avoidance strategies that could be utilised in complex multi tier trust structures in this regard.  As the first judgment indicates, this rationale is entirely speculative and goes beyond the papers – the respondent, which is well placed to understand the rationale for the particular legislative provision, failed to make out even a rudimentary case for what the purpose was behind the interpretation it proposed.  Moreover, such a speculative rationale also again fails to explain why the full application of the conduit principle only gives rise to tax avoidance concerns in relation to capital gains: multi-tier trust structures could presumably be used to avoid tax in relation to other categories of monetary accruals.[110]  It is unclear why the Legislature allows for the application of the conduit principle at all, if its goal was to counteract tax avoidance with this provision.

 

Context

[132]     Apart from purpose, the interpretive principles adopted by the courts require an examination of various contextual factors.  Paragraph 80(2) appears in the context of the Eighth Schedule that deals with capital gains tax.  It also co-exists with section 25B in the ITA.  The latter provision, it is common cause, applies the conduit principle to all other forms of income throughout a multi-tier trust structure.  If we are to construe the provisions of the Income Tax Act harmoniously, it would seem that section 25B and paragraph 80(2) should be interpreted to reinforce one another, rather than as enshrining different approaches to the conduit principle in the same statutory scheme.  That is particularly the case given that there seems to be no good reason for interpreting paragraph 80(2) differently.

 

[133]     Apart from the statutory context, we now also have subsequent evidence that the relationship between section 25B and paragraph 80(2) was regarded by the Legislature as being unclear in its application to multi-tier structures.  Indeed, a further amendment to section 25B and paragraph 80(2) was given effect to in 2020.  The subsequently amended section 25B reads as follows—

 

“any amount (other than an amount of a capital nature which is not included in gross income or an amount contemplated in paragraph 3B of the Second Schedule) received by or accrued to or in favour of any person during any year of assessment in his or her capacity as a trustee of the trust, shall, subject to the provisions of section 7, to the extent to which that amount has been derived for the immediate or future benefit of any ascertained beneficiary who has a vested right to that amount during that year, be deemed to be an amount which has accrued to that beneficiary, and to the extent to which that amount is not so derived, be deemed to be an amount which has accrued to that trust.”

 

[134]     The 2020 amendment to paragraph 80(2) reads as follows—

 

“[s]ubject to paragraphs 64E, 68, 69 and 71, where a trust determines a capital gain in respect of the disposal of an asset in a year of assessment during which a beneficiary of that trust (other than any person contemplated in paragraph 62 (a) to (e)) who is a resident has a vested right or acquires a vested right (including a right created by the exercise of a discretion) to an amount derived from that capital gain but not to the asset disposed of, an amount that is equal to so much of the amount to which that beneficiary of that trust is entitled in terms of that right—

(a)        must be disregarded for the purpose of calculating the aggregate capital gain or aggregate capital loss of the trust; and

(b)        must be taken into account as a capital gain for the purpose of calculating the aggregate capital gain or aggregate capital loss of that beneficiary.”

 

[135]     What is evident from section 25B is that it now expressly excludes capital gains from the application of the conduit principle therein.  The language in paragraph 80(2) is also modified to make it clearer that the conduit is stopped at the immediate beneficiary of the trust that disposes of an asset and realises a capital gain.  The amended text of paragraph 80(2) utilises express language that identifies the trust disposing the asset as being the same trust that determines the capital gain.  It also directly links the beneficiary to the trust disposing of the asset.

 

[136]     The 2020 explanatory memorandum indicates the intention expressly to exclude section 25B from applying to capital gains, and for paragraph 80 to govern capital gains.  Whilst it does not explain the modification of the language in paragraph 80(2), that amendment happened at the same time as section 25B was altered and these two sections should be read in harmony with one another.  It is thus clear that the Legislature considered it necessary to amend the ITA so as to make its intention clear that the conduit principle be restricted to the immediate beneficiary of the trust that disposes of an asset and realises a capital gain – namely, the second-tier trust in a multi-tier trust structure.  The unavoidable inference is that the prior position was not clear – and, indeed, reading section 25B and paragraph 80(2) harmoniously would have required the full application of the conduit principle.  It is, in my view, impermissible for this Court to re write the legislation retrospectively to cure an ambiguity in favour of the fiscus rather than the taxpayer – as was held by the Tax Court.

 

[137]     Much is made by the first judgment of the 2008 explanatory memorandum which, it is claimed, evinces a clear intention for the conduit to be stopped at the second tier trust.  It seems to me that limited weight should be placed on such a memorandum: the Legislature is duty-bound due by the requirements of the rule of law to ensure that the legislation it passes is as clear as possible and enables individuals to know how to order their affairs.  The Legislature must, in the legislative instrument itself, say what it means and cannot cure an ambiguity by relying on an explanatory memorandum.  This is particularly so where there is very limited treatment of this issue in the explanatory memorandum.  In particular, no explanation is given in the 2008 explanatory memorandum for the purpose of limiting the conduit principle or reasons for the differentiation in this regard between the taxation of capital gains and other monetary gains.  The memorandum simply asserts the legal position it seeks to arrive at without explaining the rationale for doing so which, ultimately, should be the purpose of an “explanatory” memorandum.[111]

 

[138]     Indeed, this Court has, for instance, utilised an explanatory memorandum in Assign Services[112] to ascertain the purpose of legislative provisions rather than the meaning of the provisions themselves.[113]  Where an explanatory memorandum fails to articulate the rationale for a provision but simply asserts an interpretation of the statutory provision, the weight to be attached to such a document is very limited.  Reference to such an explanatory memorandum alone cannot cure an ambiguity in the language of the provision itself and dislodge the need to interpret legislation in light of the applicable interpretive principles and in a manner so as to preserve its constitutionality.

 

[139]     As I have indicated, we are required to interpret legislation in such a way that ensures conformity with the Constitution and its foundational values.  This Court should be hesitant to adopt an interpretation of legislation that renders sections thereof arbitrary and involving distinctions that have no rational purpose.  As I have discussed above, the contra fiscum rule requires that fiscal legislation must be clear and, in the event of an ambiguity, interpreted to favour the tax subject.  We are thus duty bound in light of the interpretive principles I have discussed to prefer the interpretation that renders this legislation rational, non-arbitrary and in favour of the taxpayer.  That interpretation is the second one I have explicated that does not arbitrarily restrict the operation of the conduit principle in the context of capital gains tax.  I have sought to show why this interpretation is preferable when the text of paragraph 80(2) is construed in light of its statutory context and in relation to its manifest purpose.

 

[140]     Apart from the need to construe legislation in a non-arbitrary and rational manner, I believe this reasoning also conforms to the equities involved: given the lack of clarity of the legislation relating to multi-tier trust structures, it is unjust and inequitable retrospectively to impose a large tax bill on a second-tier trust.  Indeed, expert tax advisors were unable to ascertain its true meaning (as was evident from the differing opinions in this case), and academics have noted the lack of clarity in this regard.[114]  The Tax Court and the Supreme Court of Appeal reached completely different conclusions about the applicable tax regime.  In these circumstances, once again, it is equitable to adopt an interpretation in favour of the taxpayer.

 

[141]     For these reasons, had I commanded the majority, I would have found in favour of the applicant and upheld the appeal.  In these circumstances, there would be no need to decide the cross-appeal though I concur with the reasoning of my Colleague Chaskalson AJ in that regard.

 

For the Applicant:	W Trengove SC, T Emslie SC, C Steinberg SC and M Sibanda instructed by Werksmans Attorneys
For the Respondent:	M A Chohan SC and L Kutumela instructed by Madiba Motsai Masitenyane and Githiri Attorneys

---