AFRICA CASH AND CARRY (PTY) LIMITED Appellant

and

THE COMMISSIONER FOR THE SOUTH AFRICAN

REVENUE SERVICE Respondent

Neutral citation: Africa Cash & Carry (Pty) Ltd v The Commissioner for the South African Revenue Service (783/18) [2019] ZASCA 148 (21
November 2019).

Coram: Navsa, Swain and Zondi JJA and Koen and Hughes AJJA

Heard: 6 November 2019

Delivered: 21 November 2019

Summary: Income and value added tax – Tax Administration Act 28 of 2011 – powers of tax court to alter assessment under s 129(2)(b) – whether SARS proved

that the methods of assessment used were reasonable – whether the tax court ought to have remitted the assessment - s 89 quat interest.

ORDER

On appeal from: The Tax Court (Satchwell, Makume et Mali JJ) and N. Singh (assessor).
The appeal is dismissed with costs, such costs to include the costs of two counsel where employed.

JUDGMENT

Koen AJA (Navsa, Swain, Zondi JJA and Hughes AJA concurring):

The appellant (the taxpayer) operates a cash and carry business and as the facts set out hereunder demonstrate, it generates substantial amounts of cash. During 2011 the respondent (SARS) raised estimated assessments (the assessments) in relation to the taxpayer in respect of additional income tax and value added tax (VAT), penalties and interest for the financial years 2003 to 2009, totalling some R600 million. It is not in dispute that the taxpayer fraudulently suppressed its sales figures for those years which resulted in its income tax and VAT liability being under-declared. On the version of its own experts it is liable to the fiscus for unpaid taxes of at least R68 million. The taxpayer objected to the assessments, and when the objections were disallowed, appealed to the Tax Court (tax court). The taxpayer raised the defence of prescription, which was not persisted with as the assessments arose from a fraudulent understatement of income. The tax court dismissed the appeal and ordered that the additional tax per the assessments be altered in terms of s 129(2)(b) of the Tax

Administration Act 28 of 2011 (the Act) to lesser amounts, as fixed in the tax court’s order. This appeal is against those parts of the judgement of the tax court:

dismissing the taxpayer’s first point that SARS was bound by the assessments;

finding that the assessments were reasonable;

concluding that the tax court had jurisdiction in terms of s 129(2)(b) of the Act to alter the assessments and grant the order it did; and

concluding that the interest in terms of s 89quat of the Income Tax Act 58 of 1962 (Income Tax Act) should not be remitted altogether.

Background

During 2007 SARS became privy to certain information1 concerning tax evasion in the Jumbo Group of companies (Jumbo) which operated in the same cash and carry market as the taxpayer. It shared certain shareholders in common with the taxpayer, notably a Mr Hathurani, who, in discussions with SARS officials, which included a Ms Pretisha Rowesh Khoosal, a specialist forensic auditor employed by SARS, admitted that Jumbo had under declared cash sales of approximately R200 million. It had under- reported its sales by using a functionality in its point of sale (POS) system, which allowed manual manipulation of its sales figures. The point of sale system used by the taxpayer is similar to the one used by Jumbo.

SARS thereafter commenced an investigation into the tax affairs of the taxpayer. According to its annual financial statements it traded at a negative gross profit margin and only returned a positive profit margin through the earning of rebates and discounts. The investigation then turned to a scrutiny of the taxpayer’s records of its sales and stock.

SARS visited the taxpayer’s premises on 19 March 2009 to obtain and copy the relevant financial data to perform its investigation. A hopelessly inadequate disclosure of the taxpayer’s financial records caused SARS to apply2 for search and seizure

1 These related mainly to the Jumbo group of companies and specifically Jumbo Cash and Carry. An enquiry had been launched into its affairs in terms of s 74C of the Income Tax Act.

2 These were applied for in terms of s 74D of the Income Tax Act and s 57D of the Value Added Tax Act 89 of 1999 (the VAT Act) as they then read.

warrants. These were executed at the taxpayer’s premises on 20 March 2009. Computer equipment was seized, sealed in evidence bags and removed. Available computer hard drives were also imaged at the taxpayer’s premises. Documents were placed in boxes, each having a template indicating the location from which documents were removed and a brief description of the documents contained therein, and removed. What was seized was acknowledged in writing by the taxpayer’s legal representative. About 74,000 documents in pallets were seized comprising sales invoices, supplier invoices, supplier trade agreements, and financial records.3

Ms Khoosal was present during the visit to the taxpayer’s premises on 19 March 2009, and thereafter during the execution of the warrants. She has personal knowledge of all material dealings SARS had with the taxpayer regarding its tax liability, throughout, until her resignation from SARS in 2015. Her factual evidence as to what records were made available by the taxpayer and the inadequacy thereof, stands unchallenged. It is not in dispute that the records that could be found and were seized were incomplete and inadequate. The taxpayer’s representatives were invited to be present when the seals to the records that were seized would be broken. None of the taxpayer’s representatives however attended.

The software related to the POS system4 used by the taxpayer to record its sale transactions is distributed by React Solutions (Pty) Ltd (the React System). The taxpayer used a DOS version (the DOS version) of the React system from early trading, customised to its requirements. During July 2008 the DOS version was replaced with a standard Windows version of the React System (the Windows version). Both the DOS and Windows versions have a sale suppression functionality which allows for the manual manipulation of sales details.5

3 These records were subsequently discovered by SARS for the purpose of the trial before the tax court.

4 The point of sales programme used by the taxpayer recorded sales of stock and monitored stock levels in the warehouse/shop. Transactions are recorded at the terminals by the cashiers and the information then feeds into the server and the server records all the information from the Point of Sale terminals.

5 These are referred to as a P-Type adjustment.

The only source data which SARS was able to image, was from the Windows version, which only covered the seven month period from August 2008 to the end of February 2009 (the seven month period). As regards the period prior to that when the taxpayer used its DOS version, SARS requested the particular DOS version software to access the data, but this was not forthcoming. SARS did not find any backup data for this period during its search and seizure operation. Requests for copies of the backup data for the period before August 2009 were met with the response that the DVDs containing the backup data for that period were stolen during a burglary at the taxpayer’s premises. No evidence to confirm a theft of the backup data was adduced by the taxpayer before the tax court. SARS requested the relevant customised DOS software application from the taxpayer. The taxpayer alleged that this DOS version was on drives seized by SARS. Ms Khoosal denied that these drives had been seized by SARS. Her evidence stands unchallenged.

The result was that SARS was restricted to the so-called raw data from the Windows version, which only covered the seven month period. Ms Khoosal, analysed this data and discovered that the quantities sold per the stock table exceeded the sales count in the sales table.6 These quantity variances were caused by the taxpayer suppressing its disclosed sales to below the level of its true sales. SARS adduced the evidence of Mr Moosa Loonat Ebrahim and Mr Imitaz Ahmat Ishmael Kara, former employees of the taxpayer, who testified about the ‘ooplang’7 practices of the taxpayer and the suppression of sales, which occurred in an erratic fashion. Their evidence was not disputed.

In addition to the raw data for the seven month period, SARS had some records from the Pastel accounting program used by the taxpayer on which data was captured across from the React system, hard copies of some management reports, VAT returns, and annual financial statements for each of the financial years ending

6 A total of 1076 P-type adjustments were found to have been made throughout the period from August 2008 to February 2009 according to the report from independent accountants PriceWaterhouseCoopers (PWC). The adjustments were apparently actioned by two users, user 42 and user 202, identified as Mr Moulana Sajjad and Mr Ziyaad Bhikhoo, both of whom were allocated the “ADMIN-POS’ role within the React system used by the taxpayer.

7 An ‘ooplang’ practice essentially involves the keeping of two sets of financial accounts for an enterprise with the object of manipulating financial records, usually to evade tax.

February 2003 to 2009. The manipulated sales figures were carried forward in all these records, thus rendering them unreliable.

Having determined the quantity variance of under-disclosed sales for the seven month period, SARS, by using the pricing field from the sales table in the database it was able to access, calculated the sales value variance for the seven months to be R38 994 851 (the R38 million variance). This amount was added to the fictitious sales figure declared by the taxpayer for that seven month period, resulting in a figure of R1 518 847 563. The cost of sales for that seven-month period was then calculated by extracting the value of the opening stock as at 1 August 2008 and the value of the closing stock as at 28 February 2009 from the stock table in the database. An amount of R1 526 104 367 was extracted in respect of purchases from the ‘goodsdoc’ table and reduced by R 11 694 337 for returned goods. The gross profit for the seven month period using these figures was calculated as follows:

Sales disclosed by the taxpayer R1 479 852 712

Sales variance  R38 994 850 R1 518 847 563

Cost of sales:

Opening Stock R219 461 109

Purchases R1 151 410 031

Closing Stock         (R268 954 558) R1 464 916 582

Gross Profit R53 930 981

The gross profit expressed as a percentage of the adjusted sales figure was 3,6%.

Section 95 of the Act provides for the estimation of assessments by SARS in the following terms:

‘(1) SARS may make an original, additional, reduced or jeopardy assessment based in whole or in part on an estimate if the taxpayer –

fails to submit a return as required; or

submits a return or information that is incorrect or inadequate.

SARS must make the estimate based on information readily available to it.

If the taxpayer is unable to submit an accurate return, a senior SARS official may agree in writing with the taxpayer as the amount of tax chargeable and issue an assessment accordingly, which assessment is not subject to objection or appeal.’

An estimated assessment, unlike an assessment in terms of s91 of the Act, is not a precise determination of the undeclared taxable income, because all the necessary data for a proper estimate is not available. SARS perforce had to base the estimated assessments on such records as were made available to it by the taxpayer.

SARS applied a gross margin percentage of 3,6% for the seven month period to the taxpayer’s 2003 to 2008 and the full 2009 years of assessment. It did so by using the cost of sales figures in the taxpayer’s annual financial statements declared for those years, to calculate the sales which would produce a gross profit percentage of 3,6% for each year.

Following that methodology, SARS claimed estimated additional income tax from the taxpayer as follows:

Year

Under declared sales

Income tax

Additional tax at 200%

2003

R35 555 740

R10 666 722

R21 333 444

2004

R71 884 116

R21 565 235

R43 130 470

2005

R86 937 318

R26 081 196

R52 162 392

2006

R112 416 989

R32 600 927

R65 201 854

2007

R89 057 501

R25 826 676

R51 653 352

2008

R116 517 421

R33 790 052

R67 580 104

2009

R145 602 847

R42 224 826

R84 449 652

SARS applied the same methodology in respect of under-declared sales to claim additional VAT from the taxpayer as follows:

Tax period

Output tax on under declared sales

Additional tax at 200%

2003

R4 977 804

R9 955 608

2004

R10 063 776

R20 127 552

2005

R12 171 225

R24 342 450

2006

R15 738 379

R31 476 758

2007

R12 468 050

R24 936 100

2008

R16 312 439

R32 624 878

2009

R20 387 399

R40 768 798

On 19 November 2010 SARS issued a letter of its preliminary audit findings based on the above methodology claiming the above amounts, followed by a Letter of Findings, and finally it issued the estimated assessments which formed the subject matter of the dispute before the tax court. It did so acting in terms of s 78 of the Income Tax Act and s 31 of the Value Added Tax Act (prior to their repeal and substitution respectively by the provisions of the Act).8 In terms of s 270 (3) of the Act, 9 which came into operation on 1 October 2012, these assessments are to be regarded as having been issued under the comparable provisions of the Act, namely ss 95 and 102(2).

The taxpayer objected to these estimated income tax and VAT assessments, principally on the grounds that it had not under-declared its income tax or VAT liability, that SARS’ approach to establish sales was incorrect, that the Windows version contained certain defects, and that the period within which additional assessments

8 At the time, s 78 of the Income Tax Act provided that the Commissioner, if not satisfied with a return 'may estimate the taxable income of a taxpayer'. Section 31 of the VAT Act empowered the Commissioner to issue a further assessment if 'the Commissioner is not satisfied with any return or declaration'. Neither of these provisions specified that SARS was to consider information 'readily available' to it. It is not provided under s 95 of the Act.

9 Section 270(3) of the Act provides:

‘(3) A form, notice, demand or other document issued, given or received by a person or SARS under the provisions of a tax Act repealed by this Act, must be regarded as issued, given or received in terms of any comparable provision of this Act, as from the date that the form, notice, demand or other document was issued, given or received under the repealed provisions’.

could be issued had lapsed.10 The objections were disallowed by SARS by letter dated 17 March 2012. The taxpayer then filed a notice of appeal and thereafter delivered its Letter of Appeal, noting the appeal against such disallowance. That is the appeal which served before the tax court.

SARS’s ‘Statement of Grounds of Assessment’ delivered on 23 May 2013 essentially mirrors the contents of its letter of assessment, the import whereof has been set out briefly above.

On 30 April 2015, preparatory to the hearing before the tax court, SARS filed a notice in terms of rule 37 (a) and (b) of the rules11 promulgated in terms of the Act, in respect of the expert testimony of Ms Khoosal. This notice recorded that in preparing for the tax appeal, Ms Khoosal identified three errors in the previous estimation by SARS, namely that:

the value of the variance of R38 million previously calculated did not take into account so called positive P adjustments;12

the value of the purchases previously used in the calculation of cost of sales had been overstated due to an error in the computer program script,13 which merged the data in respect of goods returned and goods received. A correction of this error resulted in the variances identified being reduced to nil and therefore no adjustment, as had previously been made, was required to purchases; and

In its calculations SARS made a typographical error with regard to the cost of sales figure for the 2008 year of assessment, resulting in an over-statement of the assessment raised for that year by R1 706 543.

10 These were set out in the taxpayer’s letter of objection dated 18 October 2011.

11 SARS's notice in terms of r 37 (a) and (b) was dated 30 April 2015.

12 Suppressed sales were negative P adjustments. The converse of that is the positive P adjustment which would have the effect of bringing the sale back into sales. They arise where something that was previously a negative P adjustment would be reversed and brought back into sales as a positive P adjustment.

13 A script is an instruction fed into the software program of a computer instructing it to do certain calculations. Ms Khoosal explained that there was a script error in the merging of the goods doc and goods doc item tables. The program did not take into account goods returned on a different date of its receipt. Her uncontradicted evidence was that this only changed the amount, not the calculation or the methodology followed by SARS.

She explained the methodology used by SARS in detail and confirmed that it had remained the same.

Taking the aforesaid into account, Ms Khoosal recalculated the stock variance for the seven months and determined that the quantity of sales units under-declared reduced from 3 412 578 to 2 390 259. She confirmed that any variance from the reconciliation between the stock and sales tables was not attributable to a purchase from or a return to a supplier,14 but arose due to the manipulation of sales in the Windows version, either being unrecorded sales using the stock adjustment account, or the reversal of recorded sales by recognition of a negative sale.

The value of the under–declared sales for the seven-month period was recalculated as R28 020 064 (the R 28 million variance). Ms Khoosal accepted that the initial R38 million variance was wrong.

The taxpayer’s experts, Mr Charles Arthur Stride and Mr Daniel Sabbagh, do not dispute the variance in sales due to the sales figures being understated. Indeed, it became common cause before the tax court that the R 28 million variance reflects the value of sales understated by the taxpayer during the seven month period. No explanation for such variance has however been forthcoming from the taxpayer.

As was done previously, SARS added the R28 million variance to the contrived sales per the React POS system disclosed by the taxpayer in its records for the seven month period. The purchases were adjusted by R 11 694 337, which had previously been taken into account erroneously, and recalculated as R1 526 550 542. The gross profit for the seven month period from 1 August 2008 to 28 February 2009 was re- calculated, as follows:

14 A reconciliation of the outflow of stock items per the stock table to the outflow of stock items per the sales table, generated a variance report of a movement in stockstatsdaily but not in saleitem and in some instances a movement in saleitem but not in stockstatdaily.

Sales

R1 479 852 712

Sales variance

R    28 028 064

R1 507 880 776

Cost of sales:

Opening Stock

R219 461 109

Purchases

R1 526 550 542

Closing Stock

    (R268 954 558)

R1 477 057 093

Gross Profit

R30 823 683

The recalculated gross profit expressed as a percentage of sales produced a now reduced gross profit percentage of 2,04%.

Again, as was done previously, the 2,04% gross profit percentage was extrapolated to the 2003 to 2008 and full 2009 years of assessment, to revise the contrived annual sales figures reflected in the annual financial statements submitted for those years, to result in a gross profit expressed as a percentage of sales of 2,04% for each year. The total income tax and VAT liability payable for those years were recalculated as R 208 288 350,47 (R140 632 842,84 income tax and R 67 655 507,63 VAT).

On 3 September 2015 Ms Khoosal produced a factual report. Annexure FR 48 to her report sets out how these amounts were calculated for each of the tax years from 2003 to 2009. That was the first time that the corrections and revised amounts SARS would persist with, were set out for each year of assessment. The schedule of her calculations based on the 2,04% gross profit percentage is reproduced in the schedule below. The income tax and VAT payable were calculated as R192 755 633,20 and R92 116 071,04 respectively, excluding additional tax interest and penalties. Her previous calculations based on a gross profit percentage of 3,6% and containing the errors alluded to by Ms Khoosal, which was annexed to the assessment letter, annexure FR 47 to Ms Khoosal’s factual report, are set out in the second schedule below.

AFRICA CASH AND CARRY (PTY) LTD

EXTRAPOLATION OF FINDING

AFRICA CASH AND CARRY (PTY) LTD

EXTRAPOLATION OF FINDINGS

The methodology adopted in calculating the original R38 million variance, which resulted in a gross profit percentage of 3,6% and gave rise to the estimated assessments, and the methodology adopted in calculating the R28 million variance and resulting in a gross profit percentage of 2,04%, were identical. The difference was caused by two incorrect input values, the understated sales (due to a lower calculated sales variance) and the purchases figure (due to an incorrect script and the typographical error in the cost of sales figure for 2008).

The original assessments were not withdrawn15 nor amended, but remained together with the objection to the assessments by the taxpayer, the disallowance of the objections, and the appeal as framed by the pleadings. Ms Khoosal’s expert summary simply concluded that:

’47. In Ms Khoosal’s opinion the analysis and extrapolation conducted provides a sufficient and reasonable foundation and methodology for SARS to assess the taxable income and VAT on the basis explained above.

48. The calculations, inferences and methodologies followed by SARS, having regard to the Taxpayer’s failure to maintain and produce proper accounting records and source documents, are reasonable and appropriate in order to arrive at the estimates, determinations and quantifications identified above and in the assessments. The assessments must be evaluated taking into account what is stated in this expert summary.’

On 19 October 2015 the taxpayer filed its amended Grounds of appeal with the caveat that the taxpayer disputed that there had been ‘assessments’ as defined in the Act. In response to an enquiry by the taxpayer’s attorneys, SARS’s attorneys on 23 October 2015 in a letter advised the taxpayer’s attorneys that:

’20. Although our client intends proving the reasonability of the methodology used at the tax appeal, the Tax Court will be asked to confirm the reduced calculated margin of 2,04%, as extrapolated. Therefore, the Tax Court will be asked to make an order in terms of s 129(2)(b) of the Tax Administration Act, Act 28 of 2011.

15 SARS maintains that it was not obliged to withdraw the assessments.

21. In light of the above, we place on record that our client does not, as currently advised, intend issuing reduced assessments prior to determination by the Tax Court of the tax appeal.’

At the commencement of the proceedings before the tax court, SARS’s counsel, according to the judgement of the tax court, placed on record that SARS, ‘did not intend to and was not proceeding with an insistence upon full liability in accordance with the figures of 2011 and 2015 but that it was proceeding only with the figures set out in the amended Khoosal report based upon undeclared sales for the seven-month period of the lesser amount of R28 million resulting in the 2,04% gross profit margin – as set out in the expert witness summary of April 2015 and confirmed in the correspondence of 23 October 2015’.

SARS adopted the stance during the trial before the tax court that the ‘amount of the liability’ had changed ‘at appeal stage’, and that it was not required of SARS to amend the assessments accordingly.16

The tax court dismissed the appeals and ordered that ‘the assessments are altered in accordance with s 129(2)(b) of the Tax Administration Act as indicated below, and it is directed that SARS alter the assessments in respect of the 2003 to 2009 income tax assessments and VAT assessments, forming the subject of this tax appeal accordingly’.

The issues in the appeal

The taxpayer raises what it terms are four key issues, namely that:

it was not open to SARS to defend the assessments by contending for a materially different tax liability to that reflected in the assessments, in the absence of revised assessments or a valid concession of the appeal, whether in whole or in part;17

16 It filed an expert summary in respect of an independent expert, Mr Louis Strydom, employed by PWC. He confirmed that the calculations of SARS were reliable.

17 This issue is perhaps more accurately described in the judgement of the tax court as the argument that 'Change in calculation of an amount is not an “assessment”’.

the tax court, by substantially ‘altering’ the assessments, exercised the functions of SARS, which it may not do, hence it acted beyond its powers;

if the aforesaid arguments both fail, that SARS has not discharged the burden of proving the reasonableness of the assessments effectively revised by the tax court; and

the s 89 quat interest ought to have been remitted.18

In raising these issues, it was emphasized that the audi alteram partem principle was not adhered to.

Was SARS bound by the assessments?

Section 1 of the Act defines ‘assessment’ as meaning:

‘the determination of the amount of tax liability or refund, by way of self-assessment by the taxpayer or assessment by SARS’.

Based on the wording of this definition, the taxpayer emphasises that not only must the amount of liability be determined, but that the amount must be determined by SARS, not by the court.

It is clear that SARS must fix a liability in a certain amount and communicate it to a taxpayer, for such notification to qualify as an ‘assessment’. An assessment cannot mean the unexpressed thoughts of an assessing revenue officer. It requires the written representation of those thoughts.19 The taxpayer consequently argues that an assessment requires a ‘formal act’ where an assessing officer in SARS records the amount of the liability, such as the initial assessments. In its submission, the amount of the liability cannot change without the assessment being changed by SARS.20 It follows, it was submitted, that if SARS is asked what assessment it relies upon that it must be able to point to the letter of assessment (and pleadings). It was also submitted, that an assessment is ‘administrative action’ for the purposes of the Promotion of Administrative Justice Act 3 of 2000 (PAJA) and that once made, unless challenged

18 The tax court failed to address this aspect in the judgement, but made an order that s-89quat interest is payable.

19 Irvin & Johnson (SA) Limited v Commissioner for Inland Revenue 1946 AD 483 at 487.

20 SARS cannot change an assessment at will or in any manner it devises, which is not in accordance with the law, and it is not at liberty to waive tax, unless a tax law specifically allows it to do so. Carlson Investments Share Block (Pty) Ltd v Commissioner for the South African Revenue Services 2001 (3) SA 210 (W) at 231 F-G.

successfully, would harden into finality.21 Finally, in terms of the principle of functus officio, the taxpayer submitted that once an assessment has been made, it may be revisited only within the parameters of prescribed powers in the Act.22

The argument advanced is that these powers in the Act allow SARS only to either issue additional assessments, reduce assessments in terms of s 93(1)(e)(ii)23 or withdraw assessments in terms of s 98 (1)(d)(i)(bb)24 if there was ‘a processing error by SARS’ provided the jurisdictional requirements are met, or to concede the appeal whether in whole or in part, provided such concession was made in compliance with the law. The taxpayer contends that none of this happened and hence that the approach of SARS is fatally flawed.

SARS does not dispute that the assessments were not reduced or withdrawn. But is does maintain that the assessments were conceded in part. The issues arising accordingly are whether there was a valid concession and/or whether the assessments could be lawfully altered as the tax court ultimately did.

21 Oudekraal Estates (Pty) Ltd v Municipality of Cape Town [2004] ZASCA 48; 2004 (6) SA 222 (SCA)

para 26.

22 Retail Motor Industry Organisation v Minister of Water and Environmental Affairs and another [2013] ZASCA 70; 2014(3) SA 251(SCA) paras 23-25. Carlson Investments Share Block (Pty) Ltd v Commissioner for the South African Revenue Service 2001 (3) SA 210 (W).

23 Section 93 provides:

‘SARS may make a reduced assessment if –
(a) …

(b) …

(c) …

(d) …

(e) a senior SARS official is satisfied that an assessment was based on-

…

A processing error by SARS;

…’

24 Section 98 provides:

‘SARS may, despite the fact that no objection has been lodged or appeal noted, withdraw an assessment which-
(a) …

(b) …

(c ) …

in respect of which the Commissioner is satisfied that –

it was based on- (aa) …

(bb) a processing error by SARS; or (cc) ...’

Was there a valid concession by SARS?

The taxpayer complained that SARS for the first time during argument before the tax court contended that it conceded the appeal in part. SARS however maintained that it had given notice that it conceded part of the assessments to reduce the amount of the taxpayer’s liability, in accordance with what was set out in the expert summary of Ms Khoosal much earlier, referring firstly to the summary of her expert evidence delivered on 30 April 2015, secondly SARS’s attorney’s letter of 23 October 2015, and thirdly SARS’s counsel’s address at the commencement of the proceedings before the tax court. The tax court held that the expert summary of Ms Khoosal and the attorney’s letter of 23 October 2015 both constituted notice of a concession.

The issue raised is whether there was a concession of part of the appeal before the tax court. Concessions are governed by s 107 (7) of the Act and rule 46 of the tax court rules.

Section 107(7) of the Act provides:

‘(7) SARS may concede an appeal in whole or in part before –

the matter is heard by the tax board or the tax court; or

an appeal against the judgement of the tax court or higher court is heard.’

Rule 46 of the tax court rules promulgated under s 103 of the Act,25 provides:

‘(1) If at any time before it has been set down under rule 39 an appeal or application under Part F is withdrawn by the taxpayer or conceded by SARS under s 107 of the Act, notice of the withdrawal or concession, whichever is applicable, must be given to the other party.

If an appeal or application has been set down for hearing under rule 39, or is part heard, and the taxpayer withdraws or SARS concedes the appeal or application, the relevant party must–

deliver a notice of withdrawal or concession, whichever is applicable, to the other party and to the registrar; and

in such notice, indicate whether or not the party consents to pay the costs of the other party.’

25 These rules provide for ‘the procedure to be followed in lodging an objection and appeal against an assessment or decision and the conduct and hearing of an appeal before a tax court’.

Neither Ms Khoosal’s letter, nor the attorney’s letter, nor counsel’s address referred to the actual amounts in respect of the actual years of assessment with which SARS would persist as representing the taxpayer’s additional income tax and VAT liability. The expert summary of Ms Khoosal made no express reference to a concession. It simply recorded that ‘the assessments must be evaluated taking into account what is stated in this expert summary’. The summary of her expert evidence did however pre-warn the taxpayer as to what SARS’s approach would be so it would not be taken by surprise.

The same observations, so the taxpayer contended, applied in respect SARS’ attorney’s letter of 23 October 2015, which simply advised that, ‘…the Tax Court will be asked to confirm the reduced calculated margin of 2,04%, as extrapolated. Therefore, the Tax Court will be asked to make an order in terms of s 129(2)(b) of the Tax Administration Act, Act 28 of 2011’. The amounts ‘conceded’ were not stated. SARS’ counsel’s address at the beginning of the appeal before the tax court also did not record the actual amounts conceded. In all three instances, so it was submitted, the amounts conceded would have to be calculated, being the difference between what was originally sought as per the assessments and the amounts of the tax liability SARS would persist with based on the 2,04% gross profit percentage.

It is clear that SARS, when faced with the enquiry by the taxpayer as to whether the original assessments would be withdrawn, belatedly resorted to the mechanism of a concession. The belated reliance on s 107(7) was opportunistic and not conscientious. Whether the tax court correctly concluded that the expert summary of Ms Khoosal and the attorney’s letter of 23 October 2015, constituted a valid notice of a concession, does not have to be decided because the tax court ultimately resorted to an alteration of the assessments, in terms of s 129(2)(b). Whether it had the authority to do so and did so correctly, is for the reasons that follow, dispositive of this appeal.

The tax court’s power to alter an assessment in terms of s129(2)(b) of the Act

At the conclusion of the hearing the tax court was faced with the original assessments and evidence which did not support the amounts in the assessments, but a tax liability for lesser amounts, based primarily26 on calculations, applying a gross profit percentage of 2,04%, certain other evidence (including that of Mr Louis Strydom) and the errors in the previous calculations identified by Ms Khoosal. The taxpayer maintained that the assessments should in those circumstances have been set aside in toto, and if the tax court was so minded, the matter remitted to SARS, particularly as the 16 page draft order handed up during oral argument at the end of the hearing, was not the relief the taxpayer had pleaded to. The draft order it submitted, provided for a tax liability substantively and substantially different to the amounts determined in the assessments.

The tax court however decided to order the assessments ‘to be altered’ to result in the order which was granted. It did so relying on s 129 of the Act27 which provides:

‘Decision by tax court

The tax court, after hearing the ‘appellant’s’ appeal lodged under s 107 against an assessment or ‘decision’, must decide the matter on the basis that the burden of proof as described in s 102 is upon the taxpayer.

In the case of an assessment or ‘decision’ under appeal or an application in a procedural matter referred to in s 117(3), the tax court may–

confirm the assessment or ‘decision’;

order the assessment or ‘decision’ to be altered; or

refer the assessment back to SARS for further examination and assessment.

In the case of an appeal against an understatement penalty imposed by SARS under a tax Act, the tax court must decide the matter on the basis that the burden of proof is upon SARS and may reduce, confirm or increase the understatement penalty.

If SARS alters an assessment as a result of a referral under subsection (2) (c), the assessment is subject to objection and appeal.

….’

26 The order also took account of the evidence and the errors identified by Mrs Khoosal (even prior to the hearing commencing).

27 The predecessor to s 129 was s 83 of the Income Tax Act.

The taxpayer contended that the tax court ‘altered’ the assessments in a manner not contemplated by s 129(2)(b) of the Act. It construed the power to ‘alter’ to mean something not comprehended by ss 129(2)(a) or (c), as the power to ‘alter’ stands in the alternative to those subsections. In particular it argued that to ‘alter’ means either to set aside in the macro sense as in Commissioner, South African Revenue Service v Stepney Investments (Pty) Ltd,28 or to change in a manner and to a degree which leaves the original assessment and its function as a notice to the taxpayer (permitting no more than what was termed ‘keyhole surgery’) as in ITC 1869 (75 SATC 329) intact.

The taxpayer specifically maintained that there was a fundamental and substantial change and not just an obvious error which needed to be corrected, which resulted in a change in the basis of the assessment and called for a reassessment. Further, to interpret s 129 of the Act as giving the tax court the power to reduce an assessment in line with a new liability relied upon by it:

violates the separation of powers doctrine as the court would be usurping a function ‘close to the “heartland”29 of executive power’, as the altering of the assessments in the manner done by the tax court is a non-judicial function not expressly provided for in the Constitution and not connected with the core function of the judiciary.30

violates the taxpayer’s right to a fair trial entrenched in s 34 of the Constitution, as SARS, having rejected the objection and determined the tax liability in the assessments, cannot fairly after the event change the basis for its ‘estimation’ relied on in its Statement of Grounds of Assessment. In terms of rule 34 of the tax court rules, the Statement of the Grounds of Assessment read with the taxpayer’s Statement of the Grounds of Appeal and, if any, the Reply, constitute the pleadings and define the issues in the appeal. In essence the argument is that s 129 (2)(b) does not allow the tax court itself, in effect, to ‘re-issue’

28 Commissioner, South African Revenue Service v Stepney Investments (Pty) Ltd [2015] ZASCA 138; 2016 (2) SA 608 (SCA) para 29.

29 South African Association of Personal Injury Lawyers v Heath & others 2001(1) SA 883 (CC) para 24 and National Treasury & others v Opposition to Urban Tolling Alliance & others [2012] ZACC 18; 2012

SA 223 (CC) paras 67 and 69.

30 That is applying the test in National Society for the Prevention of Cruelty to Animals v Minister of Agriculture, Forestry and Fisheries & others [2013] ZACC 26; 2013 (5) SA 571(CC) para 38.

assessments under the guise of an ‘alteration’ thereby depriving the taxpayer of procedural entitlements under the rules.

The Oxford English Dictionary defines the word ‘alter’ as:

‘to make (a thing) otherwise or different in some respect; to make some change in character, shape , condition, position, quantity, value etc without changing the thing itself for another; to modify, to change the appearance of’

‘to become otherwise, to undergo some change in character or appearance.’

It defines the word ‘examination’, which appears in s 129(2)(c) of the Act dealing with the power of a tax court to ‘refer the assessment back to SARS for further examination and assessment’, as:

‘a testing, trial, proof’ or ‘the action of testing or judging by a standard rule’ or ‘the action of investigating the nature, qualities or condition or any object by inspection or experiment, minute inspection, scrutiny’ or ‘the action or process or searching or enquiry into (facts, opinions, statements etc), investigation, scrutiny.’

The tax court held that:

‘Subsection (b) envisages that when an assessment is ordered to “be altered”, the assessment is changed or modified in identified respects but the assessment is not completely transmuted or transmogrified into an entirely new entity comprising new DNA. Subsection (c) envisages that the assessment is referred back to the creator thereof, SARS, for a further process of investigation so as to test the subject matter and arrive at a further result.’

That is a correct interpretation of s 129(2)(b). It also emphasizes the distinction between instances where a tax court may ‘alter’ an assessment, and those instances where it needs to refer an assessment back to SARS ‘for further examination and assessment’.

That a tax court in principle has this power to alter an assessment, cannot be doubted. The clear wording of s 129(2)(b) provides for such eventuality. It is exactly what happened in ITC 186931 where a management fee of some R12 million had been allowed twice as a deduction. The court granted an order amending the original

31 Income Tax Case 1869 (75 SATC 329).

assessment to exclude the double deduction of the management fee. Such an alteration would not offend against the separation of powers doctrine. The taxpayer however argues that in ITC 1869 the court was faced with an obvious error that did not affect the rest of the assessment, and that this is not the position in the present matter. That submission is incorrect. The reduced tax liability calculated on the 2,04% arose because of an obvious error on the part of SARS in the three limited respects, identified and easily corrected by Ms Khoosal.

In Commissioner, South African Revenue Service v Pretoria East Motors,32 SARS disallowed certain income tax and VAT input tax deductions in respect of payments allegedly made in cash as rentals to a landlord in respect of an additional parking space leased by the taxpayer. Following an interruption of a witness’ testimony by the court, no further evidence was adduced on behalf of the taxpayer of these cash payments. The tax court concluded that SARS had failed to appreciate that the issue between the parties was not the entire rental but just the alleged cash component. The evidence in that regard was insufficient to discharge the onus, but the insufficiency of the evidence was considered to be as a result of the intervention by the court. Accordingly, it was found that the appeal had to succeed in relation to this item but that the matter had to be remitted in terms of s 83(13)(a)(ii) of the Income Tax Act33 to the Commissioner for further investigation and assessment of the parking rentals allegedly paid. Unlike Pretoria East Motors, in the present matter there is no further evidence requiring further investigation and assessment. All the facts, investigations and assessments underlying the calculation of the tax liability based on the gross profit percentage of 2,04 had been disclosed timeously to the taxpayer, was placed before the tax court, and the tax court was in a position where it could grant the order it did.

In Avenant v Commissioner for the South African Revenue Service,34 referred to in the judgment of the tax court,35 SARS had issued an additional assessment in

32 Commissioner, South African Revenue Service v Pretoria East Motors and another [2014] ZASCA 91; 2014 (5) SA 231 (SCA). The comments above arise from the third ground of appeal in that matter.

33 The applicable section prior to its repeal by the provisions of the Act which introduced s 129(2)(c), provided that the tax court may ‘refer the assessment back to SARS for further examination and assessment’.

34 Avenant v Commissioner for the South African Revenue Service [2016] ZASCA 90; 2016 JDR 1025 (SCA).

35 Income Tax Case No 1873 77 SATC.

respect of ‘closing stock from farm operations’. In that case the tax court found there was an error as to the amounts included as taxable income in respect of the closing stock from farming operations, which was ‘manifestly erroneous, unfair and unreasonable’ and which ‘arose out of employing an illogical and incongruous methodology’. The tax court noted that ‘no evidence was adduced on behalf of SARS in support of the error’. Following an amendment by SARS of its grounds of assessment ‘no alternative value was fixed in respect of closing stock and the taxpayer was not informed what SARS’ case was. SARS also did not respond to the taxpayer’s request for reasons for the assessment as required by the rules of the tax court. The tax court concluded that:

‘With the paucity of evidence adduced on what would constitute a fair and reasonable method of quantification, this court is not in a position to substantiate respondent’s calculation with that of its own.’36

The issue of determining the value to be placed on the income was submitted to SARS for further consideration and reassessment in terms of s 129(2)(c). On appeal37 to this court, the appellant initially submitted that the tax court erred in referring the matter back to SARS, for further consideration and re-assessment. However, at the hearing of the appeal, the appellant conceded that if the appeal failed on the merits, the tax court correctly referred the re-assessment to SARS. The tax court simply did not have that information before it. The present appeal is plainly distinguishable from Avenant on that basis. No further information was required by the tax court to grant the order it did.

In Stepney Investments,38 SARS contended for a net asset valuation methodology in respect of the value of shares in a company which held a casino licence, whereas the taxpayer contended for a discounted cash–flow method of valuation. This court accepted certain concessions made by SARS at the hearing of the appeal but also found that the taxpayer’s valuation was flawed in certain respects. The methods of valuation clearly required further examination and investigation. There were insufficient facts before the court. The court accordingly ordered that the matter

36 Ibid para 64.

37 Note 34 above para 37.

38 Note 28 above.

be remitted to the Commissioner for further investigation and assessment. There is no need for any such further examination and investigation in the present appeal.

The question whether an alteration of an assessment is competent must, like the issue of the reasonableness of an assessment or the methodology used to determine the amount of an estimated assessment, be answered in the light of the facts and circumstances of each case. It is not necessarily determined by the magnitude, in monetary terms, of the alteration but is dictated by considerations of fairness, with due observance of the audi alteram partem principle.

The point of departure should always be that a tax court is a court of revision and, ‘not a court of appeal in the ordinary sense’.39 The legislature ‘intended that there could be a re-hearing of the whole matter by the Special Court and that the Court could substitute its own decision for that of the Commissioner’, if justified on the evidence before it.40 A tax court accordingly rehears the issues before it and decides afresh whether an estimated assessment is reasonable.41 It is not bound by what the Commissioner found. In rehearing the case it can either uphold the opinion of SARS or overrule it and substitute it with its own opinion. The powers of the tax court and its functions are unique. It places itself in the shoes of the functionary and re-evaluates the facts and circumstances of the subject matter on which the assessments were based. By its very nature an estimated assessment is subject to change based on an evaluation of the evidence and any information that becomes available. What is important is that the methodology used and the assumptions on the strength of which the estimated estimates were made should remain the same, otherwise the conclusions reached by the tax court might not be procedurally fair. The tax court must place itself in the shoes of the functionary to determine whether the methodology followed and the assumptions on which the estimated assessment are based, are reasonable and produce a reasonable result.

39 Bailey v Commissioner for Inland Revenue 1933 AD 204 at 220.

40 Rand Ropes (Pty) Ltd v Commissioner for Inland Revenue 1944 AD 142 at 150.

41 Johannesburg Consolidated Investment Co v Johannesburg Town Council 1903 TS 111.

Being a court of revision does not mean that a tax court is free of restrictions. It too must observe an administratively fair process.42 That will entail inter alia that the dispute must be resolved on the issues raised by the parties and the enquiry confined to the facts placed before court.43 In this regard the pleadings are important and the parties will be kept to their pleadings, where any departure from the pleadings would cause prejudice or prevent a full enquiry. But within those limits a tax court has a wide discretion, for pleadings are made for the court and not the court for pleadings. Where a party has had every facility to place all the facts before the tax court and the investigation into all the circumstances has been thorough, then there is no justification to interfere simply because the pleadings had not been as explicit as they might have been.44

The taxpayer is entitled to enjoy all the benefits of the audi alteram partem rule. It must not be taken by surprise, and the process must have been conducted fairly.45 Whether the taxpayer was adequately forewarned to present its case fully, is again a question of fact dependent on the circumstances of each case.

The evidence in this matter established, inter alia, that the taxpayer:

was appraised of the errors in the calculation of the assessments and SARS’ intention to ask for a reduced tax liability to be fixed based on only a 2,04% gross profit margin. That was communicated inter alia by the expert summary of Mrs Khoosal on 30 April 2015 and in the attorney’s letter of 23 October 2015;

had the aforesaid basis for recalculation reiterated by SARS’ counsel at the commencement of the hearing;

by then stood possessed of annexure ‘A’ to SARS’ assessment letter, which is reproduced as the second schedule in paragraph 24 above;

42 Note 32 above para 11.

43Director of Hospital Services v Mistry 1979 (1) SA 626 (AD) at 635F-H.

44 Shill v Milner 1937 AD 101 at 105.

45 Bel Porto School Governing Body & others v Premier, Western Cape & another [2002] ZACC 2; 2002

(3) SA 265 (CC) para 87.

by then also had received the revised quantification, annexure FR 48 to the factual report by Mrs Khoosal dated 3 September 2015, reproduced as the first schedule in paragraph 24 above;

had prepared fully and was able to cross-examine SARS’ witnesses comprehensively on the calculations based on the same methodology employed in respect of the assessments, but now based on the R28 million variance and a 2,04% gross profit margin, as opposed to the assessments which were based on the R36 million variance and a 3,6% gross profit margin. A perusal of the extensive evidence adduced by the taxpayer and the detailed cross examination of SARS’s witnesses on the quantification of these input values and matters related thereto, unequivocally demonstrate that the taxpayer was well aware of the details of the case it had to meet and that it confronted that case squarely and with meticulous detail.

in regard to the so-called ‘third’ draft order presented by SARS at the end of the trial, was allowed an adjournment to allow supplementary heads to be filed, which process would address any potential prejudice it might have suffered. Certainly no prejudice has been identified.

Most significantly, did not have to deal with a change in the methodology of the assessment, but was confronted merely with an arithmetical recalculation.

Resorting to terminology, such as that a tax court can only alter an assessment if it is not a substitute for the executive, is unhelpful. Every alteration by a tax court resulting in the determination of a tax liability different to that in an estimated assessment, could bring such determination by the court within the definition of ‘assessment’, and could be criticised for being new and something which SARS could and perhaps should have done. To conclude that in those circumstances a court would not be empowered after hearing all the evidence and pursuant to a considered judgment on all the facts before it, to ‘alter’ the assessment, would be to render the power to alter an assessment in s 129(2)(b) worthless.

A tax court does not have inherent jurisdiction. However, if the evidence before it does not sustain the amount determined in an estimated assessment of a taxpayer’s liability, or it determines that the amount in the estimated assessment is unreasonable

then, subject to constitutional principles and compliance with the audi alteram partem principle, and fairness, provided that the basis for taxation is not now entirely different, and provided the court has all the information it requires to decide the matter before it, a tax court can alter an assessment, rather than ’refer the assessment back to SARS’.

Such an alteration, if in compliance with the aforesaid principles and justified on the facts as reasonable, will fall within the powers conferred in s 129(2)(b) of the Act, accordingly be competent, not offend against the separation of powers doctrine, amount to a proper discharge of the obligation imposed upon a tax court, and be entirely consistent with the tax court’s function in the greater constitutional framework as a court of revision. The tax court will simply be discharging one of its core functions.

What was said by the full court in Arepee Industries Ltd v Commissioner for Inland Revenue,46 albeit in a different context, is instructive:

‘It is at the hearing itself that the Commissioner’s case must be presented, and it is only what is raised by him at the hearing that is of any relevance to the matters in issue. In this connection the Commissioner’s reasons are totally irrelevant, for it is not his reasons that are in any way at stake at the hearing before the Special Court. At most it is his grounds that may be an issue, but even this statement must be qualified, because whatever grounds the Commissioner might earlier have had when he made the assessment or when he disallowed the objection concerned may also become irrelevant. The reason is that the Commissioner is not obliged to adhere to any such earlier reasons or grounds as he may have had and, provided the Special Court itself applies the audi alteram partem rule at the hearing of the appeal, it is not precluded from confirming an assessment merely because its conclusion is based upon a ground which differs from that which the Commissioner relied on when making it or relies on before the Special Court.’47

The taxpayer’s complaint that the alteration of the assessments by the tax court would offend against the separation of powers doctrine is unfounded, provided the tax court properly restricts itself to the confines of s 129(2)(b) of the Act. If the taxpayer’s rights to a fair trial are not impaired in any way, then a tax court would be perfectly entitled, and indeed legally obliged, to invoke and exercise its powers in terms of

46 Arepee Industries Ltd v Commissioner for Inland Revenue 1993 (2) SA 216 (N).

47 Ibid at 222H-223B.

s 129(2)(b) of the Act. The absence of any explanation from SARS as to why it did not withdraw the assessment and issue a fresh assessment, is irrelevant provided s 129(2)(b) found proper application.

Both Ms Khoosal and Mr Strydom respectively stated in their evidence that the 3,6% gross profit margin was incorrect. But that does not mean, as an inevitable consequence that where a lesser amount would be justified, regardless of the quantum of how much less, that it follows simpliciter that the appeal should be upheld, the assessments be set aside, and the taxpayer’s tax liability be referred back to SARS to assess afresh.

The methodology used to justify the reasonableness of the amounts determined in the assessments is the same methodology used, but now with different lower input values in respect of the sales variance (the R 28 million variance instead of the R38 million variance), and the purchases figure for that seven month period, resulting in a different gross profit percentage (2,04% instead of 3,6%), which was extrapolated and applied to the same cost of sales figures as previously, for the 2003 to 2009 tax years.

Whether the gross profit percentage methodology adopted by SARS was reasonable is a different issue. There is no reason why the assessments cannot be altered by the tax court to give effect to a lesser correct tax liability, calculated by applying input variables which are justified on accepted evidence. In altering the amounts, the tax liability of the taxpayer is not determined by an official of SARS (Ms Khoosal) or an independent expert from PWC, Mr Strydom, none of whom has the power in terms of the Act to make assessments, as the taxpayer complains. The evidence of these experts was adduced to assist the court. The court was persuaded by the evidence of these witnesses, and adopted their reasoning. The alterations to the tax liability for each of the years of assessment are simply consequential upon the determinations which the court is empowered and enjoined to make as a court of revision.

The tax court correctly concluded that s 129(2)(b) of the Act is the appropriate tool:

‘where portion of the original assessment can be set aside with clarity, where the taxpayer had not been taken unawares and proper notice has been given of the proposed alteration, where the provenance of the alteration is known to all and has been carefully examined by SARS and the taxpayer and the court, where there can be no prejudice to either the taxpayer or SARS, where a matter is brought to finality and it is appropriate so to do.’

The reasonableness of SARS’ gross profit percentage methodology

The taxpayer contended, in the alternative, that if the tax court had the power to alter the assessments, that it erred in using SARS’ gross profit percentage methodology in doing so, because it was not reasonable to rely on that methodology:

where there are several alternative and more reasonable extrapolation methods48 available to it, the contention being that the gross profit method results in distorted results and fails to take account of information readily available to SARS;

where there is an issue of the effect of rebates and discounts;

where it is reasonable to allow adjustments for shrinkage provided for in s 22(1) of the Income Tax Act, typically to closing stock at year end, which were not made.

In terms of s 102(2)49 of the Act:

‘(2) The burden of proving whether an estimate under s 95 is reasonable or the facts on which SARS based the imposition of an understatement penalty under Chapter 16, is upon SARS.’

The Act does not provide any guidance or criteria to determine whether an estimate made by SARS is reasonable. Following what was said in Head of the Western Cape, Education Department and others v Governing Body of the Point High

48 The taxpayer submitted that this court is not asked to select the most appropriate method but to uphold the appeal and to refer the matter back to SARS for further investigation and reassessment.

49 It makes no material difference with s 78 of the Income Tax Act or s 31 of the VAT Act or s 95 of the Act is applicable because all these provisions require that the estimations must be reasonable.

School and others,50 in a different context with reference to what is meant by ‘unreasonableness’ in s 6(2)(h) of PAJA, reasonableness would require that SARS strike a balance fairly and reasonably open to it on the facts before it or available to it. Reasonableness requires that a balance must be struck between a range of competing considerations in the context of a particular case.51 The principal enquiry is whether SARS struck a balance fairly52 and reasonably open to it on the facts before it, or readily available to it.53 If the choice of the gross profit percentage method is one that reasonably could be applied, then a court will not interfere with that decision.54 What is required for a decision to be justifiable, is that it should be ‘a rational decision taken lawfully and directed to a proper purpose’.55

Clearly, if the results of a decision are patently distorted, it cannot be reasonable.56 An estimated assessment by SARS may also not be an ‘arbitrary guesstimate’.57 If a decision ‘is so unreasonable that no reasonable person could have so exercised the power’, it will be reviewable.58 In all instances a discretion must ‘be exercised with care by properly experienced and suitably qualified personnel, since it may otherwise be reduced to an arbitrary guesstimate, with grave consequences for the taxpayer’.59

SARS had to consider all reliable information readily available to it in arriving at the assessments and must have acted rationally, in accordance with principles established in Bato Star60 and Bel Porto School Governing Body.61 Factors relevant to

50 Head of the Western Cape, Education Department and others v Governing Body of the Point High School and others [2008] ZASCA 48; 2008 (5) SA 18 (SCA) para 16.

51 This is the kind of balancing exercise required to be performed referred to in Bato Star Fishing (Pty) Limited v Minister of Environmental Affairs and Tourism & others [2004] ZACC 15; 2004 (4) SA 490 (CC).

52 However as was said in Bel Porto School Governing Body para 86, ‘The unfairness of the decision in itself has never been a ground for review. Something more is required. The unfairness has to be of such a degree that an inference can be drawn from it that the person who made the decision had erred in respects that would provide grounds for review. That inference is not easily drawn’.

53 Note 51 para 44.

54 Note 45 above para 87.

55 Note 45 above para 89.

56 Foodcorp (Pty) Ltd v Deputy Director- General, Department of Environmental Affairs and Tourism: Branch Marine and Coastal Management 2006 (2) SA 191 (SCA) para 19.

57 Mokoena v Commissioner for the South African Revenue Service 2011 (2) SA 556 (GSJ) para 10.

58 Associated Provincila Picture Houses Ltd v Wednesbury Corporation [1948] 1 KB 223 at 233-234; referred to in Bato Star para 44.

59 Note 57.

60 Note 51 above paras 42-50.

61 Note 51 above paras 87-90.

determining whether a decision is reasonable or not would include amongst others the nature of the decision, the identity and expertise of the decision–maker, the range of factors relevant to the decision, the reasons given for the decision, the nature of the competing interests involved and the impact of the decision on the lives and well-being of those affected.62 This list is not exhaustive.

The issue is not whether the decision to adopt the gross profit methodology is necessarily the best decision in the circumstances. What this court has to decide is whether the decision to apply the gross profit methodology struck a reasonable equilibrium between the applicable principles and objectives sought to be achieved, in the context of the established facts of this case.63

In support of its argument that the gross profit methodology was not reasonable the taxpayer relied on the opinions of the two experts, namely Mr Stride and Mr Sabbagh. SARS relied on the evidence and opinions of Ms Khoosal and Mr Strydom. It is trite law that an evaluation of expert testimony, involves a determination of whether and to what extent, their opinions are founded on logical and cogent reasoning.64 That would also apply to the experts’ opinions on what is reasonable in this appeal.

Experts generally express opinions in respect of established facts. No factual evidence was however adduced by the taxpayer, its management or directors. The only factual evidence available to SARS were the limited records of the taxpayer found at the taxpayer’s premises, inaccurate and contrived as some may be, of which Ms Khoosal had knowledge and which she could identify as she was personally and directly involved in finding, securing and evaluating those records. That was the only information on which SARS could estimate the tax liability of the taxpayer in respect of the suppressed sales. The absence of any factual evidence from the taxpayer, detracted from the cogency of some of the opinions expressed by Mr Stride and Mr Sabbagh, as genuine or helpful as they might have wanted to be, and their criticisms of SARS’ methodology, as shall be demonstrated below.

62 Note 51 above para 45.

63 Ibid para 54.

64 Michael & another v Linksfield Park Clinic (Pty) Ltd & another [2001] ZASCA 12; 2001 (3) SA 1188 (SCA) para 36; Twine & another v Naidoo & another [2018] 1 All SA 297 (GSJ) para 18.

It was not disputed that:

there was a manual sales suppression functionality available in the Windows version and in the DOS version used by the taxpayer;

sales were suppressed resulting in the R28 million variance;

the taxpayer had an ooplang system in place which was planned, calculated, and part and parcel of the business operations of the taxpayer, which it implemented continuously, albeit erratically, according to the evidence of Mr Ebrahim (an employee of the taxpayer) and Mr Kara (a chartered accountant and former financial manager at the taxpayer). Their evidence was not disputed by the taxpayer;

SARS even resorted to a search and seizure warrant to obtain the prime records of the taxpayer, with limited success. It was unable to gain access to the React POS backup data under the DOS version and only managed to obtain the seven months’ worth of React POS data in the Windows version;

Mr Strydom recalculated every aspect of the gross profit margin calculations of SARS and found them in line with the conclusions reached by SARS.

The gross profit percentage methodology resulted in only the sales figures in the taxpayer’s annual financial statements for 2003 to 2009 being adjusted. The other variables in the extrapolation calculations influencing the cost of sales were, and had to be accepted in the case of the 2003 to 2008 years and the first five months of the 2009 year, as per the taxpayer’s annual financial statements. There was no other information available to SARS. The exact same methodology was employed when the sales variance was reduced to R28 million with a resultant gross profit margin of 2,04%. The revised calculations were attached to Ms Khoosal’s factual report, and are set out in the first schedule in paragraph 24 above. The present is not an instance, such as in Pretoria East Motors, where the tax official responsible for the additional assessments did not familiarise herself with the workings of the accounting system utilised by the taxpayer and simply proceeded on the basis that where ‘she did not understand something she was free to raise an additional assessment and leave it to the taxpayer to prove in due course at the hearing before the tax court that she was wrong’.65 Mrs Khoosal had extensive experience of the cash and carry business, also

65 Note 32 above paras 10-11.

from her involvement with the tax affairs of Jumbo. She was as meticulous and thorough in her approach as the limited data available to her permitted.

In its heads of argument the taxpayer maintained that the gross profit percentage methodology used by SARS gave rise to a flawed result, because:

it’s application resulted in distorted figures;

the calculation does not take account of rebates;

SARS used inaccurate figures for stock; and

SARS mixed the data it used.

The alleged distorted figures

In applying the gross profit percentage methodology SARS, in its initial calculation, concluded that there was an under declaration of sales of R145 million for the 2009 year, R38 million in the seven months analysed and R107 million in the other five months. This, in the view of Mr Sabbagh, was an ‘unbelievable variation’ and a distortion which should have alerted SARS to a likelihood that there was something seriously wrong with the application of the gross profit methodology, because a variance of ‘X’ for seven months, but 3’X’ for 5 months did ‘not make sense’. In his view SARS ought to have had regard to the VAT returns for the whole 2009 tax year and not only for seven months. He compiled a schedule setting out the sales as per the VAT returns for the period March 2008 to February 2009 which reflected the sales as R 1 479 847 468, and the sales for March 2008 to July 2008 (five months) as

R950 067 489, giving a total of R 2 429 914 957 (which figure differed from the sales figure in the annual financial statements by R2). This he concludes demonstrates that the taxpayer’s VAT records were reliable and that the gross profit methodology is not reasonable.

During her cross examination Ms Khoosal however convincingly explained that it was not SARS’s methodology that caused any distorted result. What at first might appear as a distorted result arose from the taxpayer for the full twelve months of the 2009 tax year having reported a negative gross profit margin of some 2,2%. Having regard to the React POS data for the seven months which she analysed, the taxpayer during the preceding five months for which SARS had no reliable data available,

operated at a profit margin in excess of negative 2% and in the seven-month period (excluding the suppressed sales), operated at a positive margin of 0.19%. The suppressed sales were considerably more in the first five months. When the sales were adjusted for the full year in accordance with the gross profit percentage of 2,04 calculated on the data available for the seven month period, almost three times the volume of sales to that which has been calculated in respect of the seven month period now appeared in respect of the preceding five months from 1 February to 31 July 2008 because the suppression of sales was greater in the first five months of the tax 2009 year. This explains the alleged ‘distortion’.

All of this points to the reasonableness of the gross profit percentage methodology adopted by SARS, while detracting from the suitability and reasonableness of a range of possible alternative methodologies. Ms Khoosal’s explanation of the distortion makes obvious sense and is logical.

Mr Strydom similarly confirmed that the apparent distortion was the result of the taxpayer’s own declared (contrived) negative gross profit margins during the first five months of 2009. He explained how, using the taxpayer’s figures extracted by Mr Sabbagh, excluding the add back in respect of suppressed sales, the taxpayer made a smaller loss in the seven months analysed than it made in the preceding five months.

SARS could not use the VAT returns. The taxpayer has not denied that the sales figures recorded in its records were suppressed. That would include the sales figures used in the VAT returns. The suppressed sales figures carried forward in the VAT returns would be the same as those in the annual financial statements. To what extent they might have been less suppressed in the last seven months of 2009, only the taxpayer would know. It chose not to take the tax court into its confidence. The raw data for the first five months of the 2009 tax year could not be accessed because the taxpayer used the customised DOS React version which SARS could not access. The taxpayer failed to provide any back up records in respect of those five months. SARS cannot be criticised for not attaching weight to the VAT returns as they are unreliable, if not fraudulent.

On the data available, the alleged distortion, is not as a result of the methodology adopted by SARS but arises only as a consequence of the negative gross profit margin declared by the taxpayer for the full 12 month 2009 tax year compared to the positive 2,04% margin as per the Windows React POS system for the seven month period for which data was available and was analysed.

Rebates and discounts

The taxpayer complained that SARS disregarded rebates and discounts, which ordinarily affect the cost of sales calculation and hence the gross profit percentage. Both Mr Sabbagh and Mr Stride criticized SARS for not taking proper cognizance of rebates. That criticism was persisted with before this court.

Mr Sabbagh explained, and this was common cause, that the rebates and discounts were reflected in only the taxpayer’s Pastel accounting package, and in its annual financial statements, the same annual financial statements from which SARS derived the cost of sales figures which it used to determine the extent of suppressed sales. He also referred to rebate agreements with suppliers providing for the rebates being contained in five lever arch files which were allegedly seized during the raid by SARS. Accordingly, he maintained that the absence of any reference to rebates in the React raw data which SARS used to calculate the 2,04 gross profit percentage meant that the raw data did not constitute a reliable source to benchmark acceptable trading performances.

Rebates and discounts ordinarily affect the calculation of gross profit (this was rightly accepted by Ms Khoosal) because it affects the purchases figure and could affect the net difference between opening and closing stock. The taxpayer maintained that if SARS had interrogated the information in the annual financial statements for the periods of assessment, it would have realised that the taxpayer’s reported gross profit was an average of negative 2,2% before rebates and 5,4% after rebates, a 7,6% difference, which would mean that the taxpayer achieved gross profit margins of some 10% (3.6%, or 2.04% +7.6%) taking rebates into account.

There is however nothing improbable about a 10% gross profit margin, unless a valid comparison with past trading history, or a comparison with similar type businesses showed otherwise. Comparisons with the results of Massmart, for example, are in the nature of similar fact evidence. Absent factual evidence from the taxpayer that such a comparison would be valid, they are not helpful. The taxpayer elected not to adduce evidence of its trading results.

More significantly however, rebates and negative profit margins are not relevant to the objective quantification of ‘suppressed sales’ and would have no effect on the variances in sales during the seven month period. It would only impact on purchases and opening and closing stock. In respect of purchases, it would have the effect of reducing the purchases figure and hence the cost of sales, resulting in an increased gross profit and hence a higher gross profit percentage. An increased gross profit percentage would, when applied to the opening stock, closing stock and purchases figures which the taxpayer apparently accepted as it reflected those figures in its annual financial statements for the years of assessment under review, would have resulted in an even larger understated sales figure, to the detriment of the taxpayer.

What is important is that SARS, in applying its methodology, used figures excluding rebates and discounts in respect of the seven month period analysed,66 and applied the resultant 2,04% gross profit percentage similarly to cost of sales figures for the entire 2009 and the preceding 2003 to 2008 tax years, which likewise did not include rebates and discounts.

The taxpayer’s criticism that SARS failed to take into account rebates and discounts in the calculation of cost of sales is further unfair as it lacks a sound evidential base, is based on an impossibility, and therefore is at best a theoretical challenge. Ms Khoosal repeatedly stated that it is not the function of SARS to reconstruct the books of the taxpayer. On what records are available, Mr Stride and Mr Sabbagh confirmed that it would be a matter of impossibility to link rebates and discounts, which are often in respect of a calendar year, to financial data such as

66 Mr Stride confirmed that in SARS's gross profit methodology, when using the seven months’ data that was available, rebates and discounts were not deducted from cost of sales and also not added to revenue, as was done in the financial statements.

purchases for a particular tax year (which does not coincide with the calendar year). Mr Sabbagh conceded that one ‘absolutely’ could not take a rebate into account against sales over a different period and that the period and the rebate must be matched properly. He stated that any exercise to match the rebates to purchases with any degree of reliability, would be a ‘mammoth’ task. That is assuming it would be possible.

Mr Stride confirmed that he did not do the calculation of cost of sales reduced by deducting rebates and discounts. Mr Sabbagh also understandably said that he could not state what the correct amounts of the rebates and discounts for any year would be. He also did not conduct an investigation as to the correct opening or closing stock in any financial year. SARS not only did the best with what it had at its disposal, but most importantly, consistently used figures in the calculation of cost of sales which excluded rebates and discounts.

It is trite law that the manner of a taxpayer’s accounting is irrelevant for the purposes of determining a tax liability.67 Although generally accepted accounting practice and other accounting directives might recommend that rebates and discounts should generally be taken into account against purchases and stock figures, the reality is that the taxpayer deliberately did not do so. It elected, for reasons best known to itself but never explained, to report rebates and discounts in its financial statements as ‘other income’. This ‘other income’ it reflected in its annual financial statements until the 2007 financial year below the gross profit calculation, so-called ‘below the line’, as part of the calculation of its net profit and taxable income. Treating it thus, rather than deducting it from purchases and stock ‘above the line’ as part of the gross profit calculation, would leave the ‘bottom line’ calculation of net profit and net taxable income unaffected. In the tax years post the 2007 tax year the rebates and discounts were still not taken into account against purchases and stock, but were simply added ‘above the line’ as a revenue item to ‘sales’. Treating it in that manner would distort the sales figures and would not be an accurate representation of actual sales. What it meant was that rebates and discounts were again taken into account as a ‘revenue’

67 Secretary for Inland Revenue v Eaton Hall (Pty) Ltd 1975 (4) SA 953 (A) at 958B-D.

item, as it had previously been when ‘under the line’, and not in any manner affecting cost of sales.

There was no evidence or explanation forthcoming from the taxpayer as to why rebates and discounts were treated in this manner. An inference might be that the taxpayer, or its accountants, were simply unable accurately to allocate rebates and discounts to particular purchases or stock of a particular tax year, in calculating the cost of sales for such year. In all likelihood it is related to the manipulated financial records.

Ms Khoosal furthermore confirmed that if one is to allocate rebates and discounts to purchases and stock, then it must be done consistently and that one cannot selectively adjust purchases but not stock. She confirmed that there was no evidence to suggest that any adjustments had been done by the taxpayer in respect of its stock. She also agreed that if a taxpayer adjusted closing stock for rebates, that it would not benefit the taxpayer much at all because one year’s closing stock simply becomes the next year’s opening stock. Ms Khoosal confirmed that the figures used by her in respect of the seven-month period analysed to calculate the cost of sales (stock and purchases) did not take account of rebates and discounts.

Ms Khoosal and Mr Sabbagh both accepted that the figures in the financial statements in respect of the calculation of cost of sales for the years going back to the commencement of the 2003 tax year, were unlikely to be understated and hence reliable, as one can safely assume that a taxpayer generally would not understate cost of sales.68 If a taxpayer understated its cost of sales then it would increase its gross profit and eventually its net taxable income, resulting in it paying more tax. The cost of sales figures in the taxpayer’s annual financial statements were the only figures available in respect of those years. Unreliable as the sales figures in the annual financial statements may be, based on the assumption afore stated, the cost of sales figures were probably reliable, or sufficiently reliable, for SARS to use in the calculation

68 Mr Sabbagh also agreed that as a basic assumption taxpayer very seldom understate an expense, specifically cost of sales.

of the estimated assessments. If they were not, then it was for the taxpayer to adduce evidence casting doubt on the reliability of those figures.

The cost of sales figures in these previous financial periods certainly did not include rebates and discounts as these were specified separately, either ‘under’ and later ‘above the line’. If the cost of sales calculations in the annual financial statements for the tax years under review included rebates and discounts, then rebates and discounts would have been included twice, which would have overstated the net profit position of the taxpayer, an unlikely result. SARS therefore was comfortable in its knowledge that in the extrapolation exercise it performed it applied a gross profit percentage unaffected by rebates and discounts, to cost of sales figures in the prior years which likewise did not include rebates and discounts.

Mr Strydom confirmed that SARS’s calculations indeed compared ‘apples with apples’. The issue of rebates and discounts was, as the tax court rightly observed, a ‘red herring’.

Inaccurate opening and closing stock figures.

Opening and closing stock figures are an integral part of the calculation of cost of sales. The closing stock of one financial or tax year is carried forward as the opening stock of the next financial period. The tax court found that SARS accurately reflected the opening stock as at 1 August 2008 and the closing stock as at 28 February 2009, as per the React POS system data. The taxpayer contended that this finding was incorrect but does not dispute that the figures were extracted from the React POS data. The taxpayer’s criticism is that:

SARS used the values for opening stock and closing stock as reflected in the REACT POS where rebates are not reflected as they were only accounted for in Pastel and following from there in the annual financial statements;

Stock reflected in the React POS is reflected at a value prior to any s 22(1)(a) (of the Income Tax Act) adjustments to closing stock for damage, deterioration, change of fashion, decrease in the market value, or for any other reason satisfactory to SARS;

The tax court erred in finding that the taxpayer’s experts did not dispute and challenge the accuracy of the opening stock at 1 August 2008 or the closing stock at 28 February 2009;

Accordingly, and even if SARS started with figures as reflected in the REACT raw data, a reasonable approach would require that adjustment should be made even if Pastel and the annual financial statements are disregarded. In short, that it was not reasonable for SARS to disregard acceptable adjustments for shrinkage and rebates, as allowed by s 22(1)(a).

It is not in dispute, and accepted by Ms Khoosal and Mr Strydom that adjustments are normally made to stock at year end, that rebates would reduce the figure for closing stock, and that inventory should generally be at the lower of cost or selling price as a principle of accounting. As much as such an adjustment might not have been made to the closing stock figure used in respect of the seven-month period, the opening stock figure would similarly have been obtained from raw data, and would not have been adjusted for rebates and discounts or for any other possible allowances. SARS is not privy to the extent of any stock damage, deterioration, change of fashion, decrease in the market value, shrinkage, wastage, redundant stock et cetera which might justify an adjustment to stock values. It is unclear whether any such adjustments had been made in prior years. If they were, then there should have been some adjusting journal entry. Mr Sabbagh did not identify any such journal entry. The taxpayer did not adduce evidence of any adjustments made by it in the general ledgers or any other books of account in previous years for shrinkage, or whether there was shrinkage, and if so, the value thereof. In the absence of any such evidence SARS can hardly be criticised for not making any end of year adjustments. The suggestions of shrinkage were not substantiated by reference to any documents or evidence. This information falls peculiarly within the knowledge of the taxpayer who could have advanced evidence in that regard, if so advised. The taxpayer simply did not adduce any evidence as to any specific types of adjustment, or the amounts of any such adjustments.

The mixing of data

This aspect has already been touched on earlier. The taxpayer complains that SARS calculated the gross profit percentage by inter-alia using the REACT raw data which does not account for rebates and adjustments, yet had determined the gross profit percentage taking into account adjusted sales, and cost of sales figures (opening and closing stock and purchases) in the annual financial statements which did include rebates. That accusation is unfounded and proceeds on a factual premise which was not proved, and which on the probabilities is unlikely.

Ms Khoosal explained that she used costs of sale figures from the annual financial statements which did not take account of rebates, or at least were not proved to take account of rebates and discounts, and compared these to REACT raw data figures which similarly excluded rebates, thus comparing like for like.69 No factual basis was advanced by the taxpayer to show that she had erred in that regard.

Opening and closing stock figures would not generally be significantly dissimilar and accordingly would not affect the calculation of the gross profit percentage significantly, even if the opening and closing stock figures in the annual financial statements included rebates and discounts. There was no evidence from the taxpayer on this issue. In the absence of more detailed information from the taxpayer, it has not been shown that the tax court erred in concluding that the approach adopted by SARS was reasonable in the circumstances.

SARS failed to consider information readily available to it

The taxpayer contended that SARS should not only have had regard to the REACT POS data but that it should also have considered other information available, or considered alternative sources of information alleged to be available, such as the taxpayer’s VAT returns, the Pastel accounting system, the annual financial statements, the REACT management reports, and source documents seized by the

69 The notes in the annual financial statements record that the inventories were shown ‘at the lower of cost or net realisable value, or the lower of cost and selling price’, but these were not proved nor properly audited.

taxpayer during the raid. In the view of Mr Sabbagh this was because the REACT POS system did not reflect rebates, and that Pastel and the annual financial statements in contrast provided useful information on rebates and the effect on trading margins, that the accounting records and VAT returns provided information for the full 2009 year and the prior years from 2003 to 2008 as opposed to the seven months in the REACT raw data, and finally that the management reporting records would contain information on the composition of the taxpayer’s turnover and product mix affected by the adjustments to sales.

This additional information, as Ms Khoosal confirmed when cross examined, contained ‘the manipulated’ results. Her evidence has not been rebutted. The raw REACT source data, prior to manipulation, was the only reliable data. That explanation did not however satisfy the taxpayer which contended that SARS selectively used information from the annual financial statements for purposes of its extrapolation exercise. That is so, but the data used by SARS from the financial statements related only to the calculation of the cost of sales, which cost it has been assumed safely, any taxpayer would not readily understate because it would result in a greater reported gross profit. No one from the taxpayer testified that these figures were inaccurate.

Mr Sabbagh nevertheless persisted that it would have been prudent to consider not only what SARS viewed as reliable, but also the other available (probably unreliable) information. The contention is that SARS would then have realised that the gross profit methodology gave rise to distorted and unreasonable results, and the additional information would have assisted SARS to benchmark its findings to constitute reasonable assessments. The additional information which he contends should have been so considered, include the VAT returns, Pastel records, the annual financial statements and the source documents. All of these however contained fictitious and unreliable data, at least insofar as the sales figures were suppressed. That much has become common cause. His arguments are nevertheless considered briefly in turn below.

The criticism that SARS should have considered the VAT returns, must fail. It did, but the VAT returns contained the under-declared sales figures. In the words of Mr Strydom the VAT returns as a source:

‘... is unsubstantiated in terms of the data per React. There is no data to compare it with. I do not know what type of variance exists in those periods.’

The Pastel system would contain general ledgers, creditors’ ledgers, cashbook and trial balances and some information on rebates, for the entire 2009 and prior years. The value of referring to the Pastel records apparently lies therein, as stated by Mr Sabbagh, that the Pastel records correlated with both the annual financial statements and the VAT returns. That is not surprising as the suppressed sales figures were transferred to the Pastel records and that incorrect data was used in the VAT returns and other records. The fact that they correlate does not alter the unreliability of that information. Ms Khoosal however explained that she was unable to gain access to the Pastel records and that she was advised that the information was corrupted. She was criticised for not thereafter again asking the taxpayer to make the Pastel data available, which would seem senseless, if the data had indeed been corrupted. The pertinent issue is whether it was represented to her that the data had been corrupted and was not available to SARS. On that issue the evidence of Ms Khoosal stands unchallenged. The fact that Mr Sabbagh might subsequently have been given access to the Pastel records does not detract from the fact that as far as SARS was concerned, it was not given access thereto. This is not a valid criticism.

The financial statements of the taxpayer were not used by SARS, save as indicated above in respect of the cost of sales for the periods under review. Mr Sabbagh contended that the annual financial statements ‘are very useful additional sources of information regarding a business’ financial performance’. He was critical of what he referred to as SARS’ ‘selective use of the annual financial statements’ and maintained that the financial statements would have provided SARS with information against which it could have benchmarked its findings. That argument is misplaced, particularly where it has become common cause that the sales figures in the financial statements were suppressed. No factual evidence was presented on behalf of the taxpayer as to what conceivable benefit could have been achieved by consulting the financial statements over and above the extent that they were. The financial

statements only had the limited benefit that it disclosed the best and only evidence available regarding the opening and closing stock and purchases figures, at least on the basis of what the taxpayer was seemingly prepared to accept, in the absence of it taking the tax court into its confidence and adducing evidence that the cost of sales figures were also misrepresented, if in fact they were

According to Mr Sabbagh SARS seized about 74,000 documents on 19/20 March 2009, which it is maintained were readily available to SARS. These allegedly comprised some sales invoices, supplier invoices and supplier trade agreements. These were subsequently discovered and made available for use by the taxpayer at the hearing before the tax court. It was contended by the taxpayer’s experts that these may well have contained extracts from the creditors’ journal evidencing rebates and the like. No factual evidence to that effect was however given on behalf of the taxpayer. Furthermore, in the absence of all the financial records of the taxpayer, not just some 74,000 documents being available, an accurate and complete reconstruction of the taxpayer’s financial records was impossible. That is leaving aside that there is no obligation on SARS to reconstruct the taxpayer’s financial records, assuming in the first place that such a reconstruction would have been possible.

It was reasonable for SARS to use, as the best evidence available, the figures relating to cost of sales contained in the annual financial statements, which it can reasonably be assumed, in the absence of evidence to the contrary, to be unlikely to have been understated.

In regard to not consulting REACT Management reports, Mr Sabbagh contended that the management reports produce far more extensive information than the raw data contained in REACT. Ms Khoosal had only limited access to these reports until the licence affording SARS access to React expired. She was unequivocal that these other sources would not have given any additional information regarding the non-disclosure or under disclosure of sales and the extent thereof. No factual evidence of what is actually recorded or not recorded in the management reports was adduced to suggest that she was wrong in that view.

Alternative methodologies

Apart from criticising the suitability of SARS’ methodology, the taxpayer contended that there were alternative methods of extrapolation which would have been more reasonable to adopt, namely the Percentage of sale method, the Identified variance approach, or the Adapted SARS methodology. These were suggested for the first time only during or about 2015.

Mr Sabbagh confirmed, in answer to a question from the court, that no accounting rule requires one particular extrapolation method over another. However, in his view, SARS’ failure to have due regard to the nature of the taxpayer’s operations as a cash-and-carry business, resulted in Ms Khoosal selecting an inappropriate extrapolation method. This was also the view of Mr Stride. That criticism is unfounded. Ms Khoosal was well aware of the nature of the operations and as stated above, was well familiar with Cash and Carry operations.

Mr Sabbagh and Mr Stride considered other methods of extrapolation. These alternative methods according to comparative calculations done by Mr Sabbagh would result in a lower estimated tax liability for the taxpayer. They will be considered and evaluated in the context of the taxpayer’s operations, the task SARS had to perform, the reliability and extent of information/data and records available to SARS, and similar relevant considerations.

The tax court did not consider the alternative methodologies suggested as reasonable, commenting that ‘(t)he only disputes and challenges made to evidence were either technicalities or presentation of a different view on what would be or was a reasonable assessment. But facts were not challenged’.

The percentage of sales method

Both Mr Sabbagh and Mr Stride recommended the percentage of sale method as a more reasonable alternative. This method expresses the variance in sales (R28 million) as a percentage of the reported sales (for the seven-month period). That percentage would then be applied to the suppressed sales figures reported by the

taxpayer for each of the tax years from 2003 to 2009 to arrive at the understated sales. The appeal of this methodology is said to lie in the fact that it uses only the one variable, namely sales, and ignores the values for opening stock, closing stock and purchases, thereby eliminating any difficulties pertaining to rebates, mixing of information, et cetera.

Accepting the R 28 million variance, Mr Stride calculated that the percentage of sales variance averaged 1,977%, consequently that the understated sales for the period would be R198,954,177 and that the taxpayer’s tax liability would be some R85

539 598. The taxpayer accordingly submitted that the percentage of sales methodology was on the evidence a more reasonable alternative to apply as it is simple, uncomplicated and involves sales only, which is the disputed issue.

Ms Khoosal simply acknowledged, without any further comment, that the percentage of sales method was an ‘alternative method’. Mr Strydom was however clear, with respect correctly so, that one cannot simply ignore the influence of the cost of sales in the determination of suppressed sales, and therefore taxable income.

The percentage of sales method assumes that the extent of under-disclosed sales was consistent. Ms Khoosal confirmed that in the data she analysed, the monthly under-disclosure of sales was erratic. It, for example, resulted in the distortion in the 2009 year dealt with earlier. There is no reason why that would not also be the case in prior years. According to the former employees of the taxpayer, Mr Ebrahim and Mr Kara who were called as witnesses by SARS, the under-disclosure of sales and the syphoning off of the money to senior managers was erratic also during the earlier years. Simply applying a percentage to the under-disclosed contrived sales figures disclosed by the taxpayer in its financial statements and records could result in either a gross over or under statement of sales. That would give rise to extremely unreliable and unsatisfactory results.

The extent of true sales during any given period, and smoothed out over a tax year, would correlate more closely to the cost of sales, that is what the actual sales generated during a particular tax year would have cost. Sales generally vary in extent

depending on the cost of sales figure. The cost of sales figures for every tax year appearing in the annual financial statements can, for reasons indicated earlier in this judgement, be accepted as reliable. A more accurate and reasonable method to estimate actual sales would therefore be to apply the gross profit percentage to the cost of sales figure (to which it bears some relationship and will take account of fluctuations from period to another), rather than using inconsistent fictitious suppressed sales figures adjusted by a fixed percentage of sale method. Not adopting the percentage of sale method was reasonable and certainly not irrational.

The identified variance approach

Mr Sabbagh contended, in the alternative, for the identified variance approach as a preferred reasonable alternative, because in his view the bulk of the variance was made up of certain lines of products. He extracted information on purchases of the alleged lines of products for which sales were understated for the years of assessment 2003 to 2009 from Pastel and applying the trading margins for those products calculated the understated sales to be R162 million, resulting in a tax liability of R 68 million.

Ms Khoosal likewise acknowledged that this was an alternative way to extrapolate, without any material comment. Mr Strydom rightly indicated that although he agreed with the principle encapsulated by the method, he had no information regarding the particular products in respect of the periods for which there was no React POS data. Accordingly, in his opinion this methodology would not be ‘an accurate or a reasonable variance...to follow’.

Mr Strydom’s criticism is well founded. In the absence of factual evidence from the taxpayer as to what the actual product lines were in respect of which sales were suppressed, assuming that it is only the sales in respect of those particular product lines that were consistently suppressed (which assumption will also require some factual foundation being laid), and what the trading margins would be, a proper evaluation of this methodology and its comparison to the gross profit percentage method used by SARS could simply not be made.

The identified variance approach has not been demonstrated to be a more reasonable method to have adopted.

The adapted SARS methodology

Mr Sabbagh, maintained that if the gross profit percentage methodology adopted by SARS was to be applied then he would have used information for the full 2009 year, he would have accounted for rebates, and he would have made certain adjustments for the lengthy period of assessment. In his view the information in Pastel and the end of year financial statements correlated to an acceptable level to justify the use of that information.

The information in the annual financial statements obviously correlated to that in Pastel, as the annual financial statements were produced on the basis of what is contained in Pastel. This information was unreliable and in respect of sales was accepted to have been understated. The information available to SARS also did not extend to the full 2009 taxation year and little information, other than annual financial statements which contained suppressed sales figures, were available to SARS in respect of the prior years under review.

Further, as indicated previously in this judgment, it was impossible to correlate rebates with particular purchases. This was conceded by Mr Sabbagh who said it is almost impossible to reconcile the details of rebate agreements exactly with financial statement figures because one has different financial years and calendar years for the rebates.

Making ‘certain adjustments’ for the lengthy period of assessment would also remain at best speculative in the absence of factual evidence from the management or officials of the taxpayer as to what these should be. Adjustments would presumably have to be informed by what factual differences there might have been from one tax year to the next. There is simply no evidence on these aspects. It is accordingly safer to assume that over time, barring any catastrophic or similar events (of which there

was no evidence), the financial market in which the taxpayer operates would have been relatively consistent.

Mr Strydom also disputed the reasonableness of the adapted SARS methodology for ignoring the difference between the closing stock in the React POS system and the closing stock reported in the 2009 annual financial statement. No documentary or other evidence was presented to explain the difference.

The onus is on SARS to show that the methodology used was reasonable. That required no more than satisfying the tax court that an acceptable methodology, recognised as an acceptable methodology in the accounting discipline, was used and that there were cogent reasons for doing so. The taxpayer’s approach of simply picking away at SARS’s methodology in the absence of factual evidence in rebuttal to sustain the criticism levelled does not, even at the level of an evidentiary onus, demonstrate that the use of the gross profit percentage method was irrational, or not rationally justified, and that the decision to use the gross profit percentage method,70 was unreasonable.

The tax court did not err in concluding that SARS’s assessment was reasonable. The criticisms by the taxpayer’s expert witnesses, amongst others, failed to take into account that there was a systematic plan in which the taxpayer suppressed its sales and manipulated its records. The taxpayer has elected not to make a full disclosure of its activities in this regard. There is simply no basis to set aside the methodology employed by SARS as being not reasonable.

The section 89quat interest

The taxpayer argued that its contention that it was not liable for the amount set out in the assessment (calculated on the 3,6% gross profit percentage), on SARS’s own version (SARS having accepted that it should be 2,04%), was reasonable and

70 Rustenburg Platinum Mines Ltd (Rustenburg Section) v Commission for Conciliation, Mediation and Arbitration [2006] ZASCA 175; 2007 (1) SA 576 (SCA) para 20.

that the tax court therefore ought to have remitted the interest in terms of s 89quat (3) of the Income Tax Act as it read at the relevant time.

The relevant provisions of s 89quat read:

‘(2) If the taxable income of any provisional taxpayer as finally determined for any year of assessment exceeds –

R 20,000 in the case of a company; or

R 50,000 in the case of any person other than a company,

and the normal tax payable by him in respect of such taxable income exceeds the credit amount in relation to such year, interest shall, subject to the provisions of subs (3), be payable by the taxpayer at the prescribed rate on the amount by which such normal tax exceeds the credit amount, such interest being calculated from the effective date in relation to the said year until the date of assessment of such normal tax.

(3) Where the Commissioner having regard to the circumstances of the case is satisfied that any amount has been included in the taxpayer’s taxable income or that any deduction, allowance, disregarding or exclusion claimed by the taxpayer has not been allowed, and the taxpayer has on reasonable grounds contended that such amount should not have been so included or that such deduction, allowance, disregarding or exclusion should have been allowed, the Commissioner may, subject to the provisions of s 103(6), direct that interest shall not be paid by the taxpayer on so much of the said normal tax as is attributable to the inclusion of such amount or the disallowance of such deduction, allowance, disregarding or exclusion.‘

The tax court reduced the s 89quat interest in accordance with the reduced quantum of the under declared taxable income. In respect of income tax, the standard order, with details inserted specific to each assessment period, provided that:

‘The Section 89quat (2) interest on the under payment of provisional tax, be altered to be calculated on the amount with which the normal tax payable (on the total taxable income taking into account the altered under declared taxable income referred to above) exceeds the credit amount in relation to the … year of assessment, at the prescribed interest rate, from the effective date (….) until the assessment of such normal tax, being …’

In each instance of VAT assessment, the standard order, with details inserted specific to each assessment period, provided that:

‘The interest in terms of s 39(1) be altered to be calculated at the prescribed rate on the outstanding balance (taking into account the altered under declared output VAT), from…, to date of payment.’

In granting that order, the tax court has not erred and no basis has been advanced to interfere with its order.

Conclusion

There is no basis to interfere with the order granted by the tax court. This appeal accordingly falls to be dismissed.

Costs

SARS has been successful. The taxpayer asked in the event of the appeal being upheld with costs, that such costs should include the cost of three counsel as the multiplicity and complexity of the legal and factual issues made it reasonable for the taxpayer to engage the services of three counsel. A similar request emanated from SARS, which likewise employed three counsel, in the event of the appeal being dismissed.

The appeal did involve some complex issues. Two counsel could however deal with these issues adequately. In the exercise of this court’s discretion on the issue of costs the cost of two counsel only is allowed.

Order

The appeal is dismissed with costs, such costs to include the costs of two counsel where employed.

P A Koen

Acting Judge of Appeal

APPEARANCES

APPELLANTS: J J Gauntlett SC QC (with C Louw SC and E Muller) Instructed by: Faber Goertz Ellis Austen Inc, Bryanston

McIntyre Van der Post, Bloemfontein

RESPONDENT: E E Coetzee SC (with C Naudé) Instructed by: Gildenhuys Malatji Inc, Pretoria

Honey Attorneys, Bloemfontein