---

IN THE HIGH COURT OF SOUTH AFRICA

GAUTENG LOCAL DIVISION, JOHANNESBURG

Case number: 12026/2012

Date: 8 August 2016

In the matter between:

NATIONWIDE AIRLINES (PTY) LTD (IN LIQUIDATION)                                             Plaintiff

and

SOUTH AFRICAN AIRWAYS (PTY)                                                                         Defendant

JUDGMENT

NICHOLLS J:

Introduction

[1] This is a delictual claim, the first of its kind, arising out of the anti­ competitive practices of our national carrier South African Airways ("SAA"). The plaintiff, Nationwide Airlines (Pty) Ltd (in liquidation) ("Nationwide"), was a competing airline during the relevant period, 1 June 2001 to 31 March 2005.

[2] Nationwide claims R170 million in loss of profit from SAA arising out of a breach of the Competition Act 89 of 1998 ('the Act"). SAA denies that the anti­ competitive conduct caused Nationwide's loss, but contends that if damages are to be awarded, the maximum amount payable is R 20 million.

[3] The genesis of the claim is a complaint made by Nationwide to the Competition Tribunal ("the Tribunal") relating to SAA's anti-competitive conduct. There were two complaints made to the Tribunal but for the purposes of this claim, it is the second complaint that is relevant. The first complaint dealt with the period October 1999 to the end of May 2001. The Tribunal found in favour of Nationwide and handed down its decision[1] on 28 July 2005. Civil proceedings brought by Nationwide in respect of the first complaint were settled between the parties.

[4] The complaints which are the subject matter of this case culminated in a decision[2] handed down by the Tribunal on 17 February 2010 in which, as with the first complaint, it was held that SAA's conduct constituted a prohibited practice in terms of section 8(d)(i) of the Act[3]. The Tribunal, having found that SAA was a dominant firm, held that the override incentive agreements and TRUST agreements that SAA entered into with various travel agents contravened the Act by inducing them to deal exclusively with SAA at the expense of its rivals. TRUST was an acronym introduced by SAA that stood for "True partnership; Respect rules; United focus; Support; Training."

Findings of the Tribunal

[5] The Tribunal held that SAA's conduct was an exclusionary act as defined in section 1 of the Act, in that it "impedes or prevents a firm, entering into, or expanding within, a market"[4]• Further, that travel agents have the ability to divert potential customers, and did indeed divert passengers, to SAA. As regards the degree of foreclosure, the Tribunal found that the override agreements covered 56% to 76% of all travel agents' outlets by number, but rose to between 70% and 90% when weighted by revenue or passenger numbers. It was noted that the travel agents themselves confirmed the effects of the foreclosure were significant. There is no dispute that this court is bound by the findings made by the Tribunal.

[6] Of relevance to this case is the finding that the growth of internet and other direct sales channels, did not erode the ability of the travel agents to influence customers' airline preference to any significant extent. The overwhelming majority of ticket sales of all the airlines at the time were through travel agents. Other sales channels constituted approximately 30% of all sales, which in effect meant that travel agents' sales made up approximately 70% of sales of domestic airline tickets. Notwithstanding this, so the Tribunal found, the anti-competitive conduct was experienced in the entire domestic airline travel market and not solely in the travel agent sector.

[7] The Tribunal accepted the evidence of Nationwide's Chief Executive Officer, Vernon Bricknell ("Bricknell') that over the relevant period Nationwide was unable to grow on most of its routes, particularly its core routes. It determined that the effect of SAA's conduct was to divert its discretionary business away from its rivals, including Nationwide, thereby reducing Nationwide's passenger numbers.

[8] The Tribunal held that the first relevant market was that for the purchase of travel agent services for the sale of domestic airline tickets. The second relevant market was the market for scheduled domestic airline transportation in South Africa. Nationwide's contention was not that they were completely foreclosed but that their growth was impeded by SAA's conduct. The Tribunal found that the fact that the market share of Comair and Nationwide remained the same and did not decline does not support  a finding that they  were  not foreclosed from expanding into the segment of the market covered by the SAA override agreements and distributed by travel agents.[5]

[9] The Tribunal's conclusions are summarised as follows:

"After considering the evidence and arguments in these proceedings we have concluded that SAA's incentive scheme was in contravention of section 8(d)(i) in that it induced travel agents to deal with SAA at the expense of its rivals and led to foreclosure of the rivals in the market for scheduled domestic airline travel. We have found that the two relevant markets for this period are the market for travel agent services to airlines and the market for scheduled domestic air travel. We have found SAA to be dominant in both markets. Despite the recent developments in the domestic airline market, such as the launch of low cost carriers and the growth of alternative distribution channels, we have found that travel agents still constituted the most significant and optimal route to market for domestic airlines. While low cost carriers accounted for most of the growth in the domestic airline travel market we have found that during the relevant period, the market for these developments did not warrant market segmentation into a low cost/time sensitive/price sensitive market and time sensitive/price sensitive market. While the effects of SAA's conduct may have had a greater impact on that segment of the domestic airline market distributed through travel agents, we have concluded that the conduct had the effect of reducing competition in the total domestic airline market."[6]

[10] The findings of the Tribunal were upheld by the Competition Appeal Court[7] ("CAC"). It confirmed the centrality of travel agents both to consumers and as the main distribution channel of airline tickets. The CAC found that the override commissions had a material effect on the airlines' share of the defined market. This was particularly marked in directional selling by travel agents of time-sensitive (primarily business) passengers in favour of SAA as opposed to

its rivals. Therefore while Nationwide was able to grow the non-time sensitive market (primarily leisure passengers), it suffered significant foreclosure in the travel agent's sector which dealt with the more lucrative time-sensitive market.

[11] Pursuant to the Tribunal's decision, a certificate of decision in terms of section 65(6)(b)[8] of the Act was issued in terms of which the Tribunal certified that the incentive agreements and TRUST payments were prohibited practices.

It is common cause that section 65(6) of the Act contemplates a claim for damages. This section enables those found to have engaged in prohibited anti­ competitive conduct to be liable for damages to any person harmed by that conduct. The certificate is conclusive proof of its contents and is binding on a civil court.[9]

Issues before this court

[12] Various issues have been agreed. The general features of the incentive scheme are undisputed. There is consensus that the period of the infringement for the purposes of this claim is 1 June 2001 to 31 March 2005. Delictual liability has, for all intents and purposes, been admitted. Insofar as SAA initially disputed causation, it is apparent that this does not go to the merits of the claim but rather to the quantum, namely that no actual damages were caused by the anti­ competitive conduct. SAA's stance is that if there were any losses suffered by Nationwide, which it denies, this was caused by factors unrelated to SAA's anti­ competitive conduct.

[13] SAA denies in its plea that any loss of profit was as a result of abuse of its dominant position in the market, but instead any losses incurred were as a result of perceptions as to how Nationwide was run and managed. It is specifically pleaded by SAA that:

"... such loss was as a result of the market's and passenger's perception of the plaintiff's business, the way the plaintiff's business was run and managed, competition of rivals, including carriers other than the defendant, the growth of online bookings, the decision by the State to prefer the defendant as the airline of choice and the instruction to government and semi-government departments to fly with the defendant, the defendant's superior frequent travel rewards, the defendant's increased competitiveness which included a new fleet of aircraft, good on-time performance, better service delivery and additional capacity and perceived passenger concern about the age and safety of the plaintiff's aircraft."

[14] The central focus of this case, therefore, is the quantification of the damages, if any, to be awarded to Nationwide. To this end each of the parties submitted reports prepared by aviation economists from the United Kingdom, who were called as expert witnesses to assist the court. Robin Noble ("Noble") from Oxera Consulting LPP ("Oxera") testified for Nationwide and Luisa Affuso ("Affuso") from PricewaterhouseCoopers ("PwC") for SAA.

History of SAA's I ncentive Agreements

[15] In the 1990s, SAA introduced an incentive scheme whereby it paid travel agents a standard commission of 7% in respect of each ticket sold. These agreements were referred to by the Competition Tribunal as the first generation agreements. The standard commission is a flat rate commission deducted at source by the travel agent who sells the ticket to the passenger. The remainder of the value of the ticket revenue is remitted to the airline through the Billing and Settlement Plan ("BSP") of International Air Transport Association ("IATA"), a clearing house for travel industry suppliers and customers.

[16] BSP should be differentiated from "flown revenue" which is the revenue received once the passenger has flown as opposed to having merely purchased a ticket. BSP represents the revenue from tickets sold by IATA through travel agents while flown revenue is the rand value of revenue generated from all tickets once the passengers actually board the flight Flown revenue is therefore an artificial construct and not a revenue stream in the true sense of the word. Generally the BSP figures will be 10 - 20 % higher than the flown revenue. The reason given for this discrepancy by the First Tribunal is that flown revenue excludes revenues owing to travel agents for commission plus cancellations and interline  revenue  (where  a flight  booking  makes  use of another  airline for  a particular leg of the journey). [10]

[17] Historically the standard commission constituted the major portion of a travel agent's remuneration. In October 1999 SAA introduced new incentive agreements (the so-called second generation agreements) which applied across all of its domestic flights in the country during the relevant period, being October 1999 to May 2001. The second generation agreements consisted of override incentive agreements coupled with the Explorer scheme which rewarded individual staff of travel agents with free international tickets and the allocation of bonus points to the agency for tickets sold.

[18] The override commission was 'back to rand one' (the first sale made) once a certain level of sales had been reached. An incremental commission 'back to base' applied only to those sales above the target and increased the more the sales exceeded the target level. The base values used to determine the growth rewarded by commission payments due to the travel agency was typically defined as the domestic SAA flown revenue generated for the preceding financial year.

[19] Following the first complaint from Nationwide in July 2005, the Competition Tribunal[11] found that SAA's second generation agreements contravened section 8(d)(i) of the Competition Act in that they provided strong incentives to travel agents to divert passengers from rival airlines' flights to SAA. The Tribunal found that the scheme constituted an abuse of dominance and its effect was to incentivise travel agents to deal with SAA thereby foreclosing its competitors from the market and simultaneously reinforcing its own position of dominance.[12]

[20] The third generation agreements, which are the subject matter of this case, were introduced by SAA on 1 June 2001 and remained in effect until 31 March 2005. To a great extent they were similar to the second generation agreements and consisted of override agreements and TRUST payments.

[21] As with the second generation agreements, travel agents were compensated at the flat rate commission until the target revenues were reached. These were usually set on the previous year's sales of the particular travel agent and were individually negotiated with each travel agent at consortium level. Once the target was reached the travel agent received a commission calculated on a 'back to rand one' basis in addition to the basic commission. Unlike the second generation agreements the incremental commission was flattened so that the incentive rate remained the same for all sales exceeding the base level.

[22] Travel agents were induced to reach base and then strive for maximum growth above that target, at which level they were handsomely rewarded for their loyalty. Essentially the purpose was to increase the number of passengers for SAA According to Conrad Mortimer ("Mortimer"), the commercial director of Tourvest, one of the largest travel groups in the country, the targets became more aggressive and more difficult to reach over the relevant period. Mortimer testified as follows: "essentially what an override is doing, it is saying you will earn a commission if you achieve a volume. If you don't achieve that volume you will earn nothing, but if you achieve that volume then you will achieve an incentive which is often described as a percentage of what your volume is."

[23] The express terms of the agreement were couched in the following terms: "SAA wishes to incentivise and reward the agent by paying a cash override incentive for achieving and exceeding the base total ... Accordingly, this override agreement is established to stimulate and encourage the increase in total SAA flown revenue and flown passenger numbers in return for the cash override incentive payments. "[13]

[24] Targets would be set and the computation of achievement of targets would be on the basis of flown revenue rather than BSP figures. In 1997 SAA formed a strategic alliance with South African Airlink ("SAL") and South African Express ("SAX"). The sales of SAA's SAX's and SAL's tickets, on all routes, were the subject of SAA's incentive and TRUST agreements from 2001 until the first quarter of 2004. Until approximately June 2004, flown revenue sold on SAX and SALS were included in the override agreements with major travel agents for the purpose of computing performance and payments to travel agents.

[25] Commissions paid by SAA in accordance with the second and third generation agreements were paid over and above BSP commission paid for each ticket sold. In respect of override agreements, classes of tickets were differentiated, but the differentiations differed from period to period. The profitability of a travel agent in any given month may have been affected by the commission which an agent earned in the respective month.

[26] TRUST payments were made in addition to the domestic override incentives and were intended to compensate travel agents for the flat commission post target introduced by the third generation agreements. TRUST payments were lump sum payments made to travel agents for receiving specific objectives. The precise formula for TRUST payments differed across agents and through time, but the essential features were that they were based on sales outcomes of growth and support. TRUST agreements were not adopted by all travel agents and were not made "back to base".

[27] In November 2004, SAA announced that as from 1 April 2005 travel agents would receive no commission, no overrides and no corporate deals. SAA later paid a commission to agents ranging from 1-4% based on passenger volume. (This was not generally known in the industry.) The November announcement precipitated an immediate diversion of business away from SAA to other airlines. This resulted in travel agents charging a booking fee on air tickets sold to compensate for the absence of a commission.

[28] The response of the travel agents to SAA's announcement is best summed up in an email dated 31 January 2005 written by William Puk. He was the CEO of Sure Travel, another of the large travel agent groups, and sent the email to all Sure Travel Managers. It stated:

"It has become very clear that we cannot rely on saa for a decent override agreement in future and our basic commission is about to disappear altogether...I am formerly advising you that our group strategy is to move our discretionary business away from saa onto more agent friendly carriers, hence the new deals with Virgin & Nationwide. our international priorities must now lie firmly with British Airways, Virgin, Lufthansa, Nationwide, Cathay Pacific etc and domestically with BA/Comair & Nationwide as a first priority, an effort and directive to this effect must therefore be communicated by you to all your staff. it makes sense from a business point of view, 0% commission from saa and generally expensive GOS fares to sell to consumers, versus standard guaranteed commission for other airlines (provided we move the business to them) and generally better value fares for the consumer, we need to show saa in the months of feb/mar/& apri that travel agents are still vital to their business and that we can and will, direct the business away from them." (sic)

[29] This is indeed what occurred and immediately after this announcement travel agents moved a large portion of their discretionary business away from SAA to other airlines, albeit for a limited period. Around this time there was an exponential growth in the use of low cost carriers ("LCC's"). This was coupled with a sharp increase in internet bookings which was the preferred sales method of the LLC's. The above trends marked a fundamental shift in airline booking patterns.

[30] The LCC's made their entry into the South African airline industry in the early 2000's. Kulula and 1Time commenced operations in August 2001 and February 2004 respectively. This was a significant development and the aim was to tap into a market of potential passengers who had never flown before by luring them with low fares and no frills service (no food or beverages were served). It was only at the start of 2005 that the LCC's really gained popularity. The concomitant growth of direct sales via the internet and call centres impacted on the market during this period, particularly for price-sensitive passengers. Nonetheless the Tribunal found that travel agency services still accounted for 70% of the total domestic airline distribution during the relevant period.

The Rise and Fall of Nationwide Airlines

[31] It was in this aviation milieu that Nationwide flew its maiden flight on 5 December 1995 from Johannesburg to Cape Town with 30 passengers. Nationwide was the brainchild of Vernon Bricknell. He was the CEO of Nationwide for the entire period of its existence and its only shareholder through the Vernon Bricknell Trust.

[32] In the mid 70's Bricknell established a charter business which had various contracts with United Nations agencies in Africa. This later came to be known as Nationwide Air Charter ("Charter") and played a pivotal role in getting Nationwide off the ground by funding the airline. Later it provided guarantees to Nationwide in order that it could obtain unqualified audits.

[33] Nationwide's first aircraft were all BAC-111 (British Aerospace Corporation) purchased from Ryan Air. Later Boeing 737's were added to the fleet and eventually replaced all the BAC-111's. All Nationwide's aircraft: were purchased by Charter, which then leased them to Nationwide. Nationwide started its own ground-handling service and its own Aviation Maintenance Organisation ("AMO") to do its maintenance. Bricknell Aviation Properties provided accommodation for crew and staff. Like Nationwide Charter and Nationwide Airlines, Bricknell was the sole shareholder of all these entities through his trust.

[34] More routes were added as the airline grew. The Durban /Cape Town route commenced in December 2001; the Johannesburg/Port Elizabeth route in August 2002; the Durban/Cape Town and Port Elizabeth/ Durban route in April 2003. The Johannesburg/George, Johannesburg/Cape Town and Johannesburg/Durban routes were referred to as the 'golden routes' because of their profitability. At some point a Johannesburg/Mpumalanga route was introduced, as well as regional routes to Livingstone and Tanzania and international flights to London Gatwick. The regional and international routes are not relevant for present purposes, as this claim is only in respect of domestic airline travel.

[35] Nationwide flew both leisure and business class passengers - time sensitive and non-time sensitive passengers. Bricknell said the intention was to produce a unique brand and superior product with a business class interior. The interiors were reupholstered in leather and re-carpeted in the Nationwide livery to create an elegant upmarket feel. Nationwide strove to differentiate itself from its rivals by creating a unique on-board customer experience. Flight attendants spoke to every passenger personally. Nationwide was a legacy carrier - free food and alcoholic beverages were served. The focus was for the passenger to have the best possible on-board experience, so immense effort was put into training of service staff and cabin crew.

[36] To oversee this process, in 1997 Nationwide employed Roger Whittle ("Whittle"), a Canadian national who worked extensively in cabin safety with a number of Canadian airlines, including Canada's regulatory body. The tenor of his evidence was that he was pivotal in drawing up the Canadian safety regulations. He testified that there was very little regulation in the South African aviation industry at the time and the Civil Aviation Authority ("CAA") decided to adopt the Canadian aviation regulations. These later became known as the CAA regulations.

[37] Whittle's specific mandate with Nationwide was in respect of regulatory compliance in the area of cabin safety. This was undoubtedly his area of expertise and he authored manuals and conducted training programmes with SAA and British Airways Comair on various aspects of cabin safety. After a few years with Nationwide, Whittle became responsible for corporate quality which entailed taking responsibility for the IATA operational safety audit (IOSA). Towards the end of his time he became involved with all operational areas of business. Whittle said that for the last few years Nationwide was run by himself, Bricknell and Peter Griffiths ("Griffiths'), the head of finance.

[38] After Puk's email in response to the announcement of the termination of SAA's override in November 2004, Sure Travel moved its discretionary business away from SAA. As a result Nationwide's BSP share of Sure Travel increased from 10% to 14.5%. Other travel agents followed suit and there was a surge in passenger numbers away from SAA to Nationwide. This was described by Whittle as a 'step change' and resulted in a significant improvement in Nationwide's fortunes. The increase in passenger numbers was sustained throughout 2005 with a drop in 2006.

[39] In this period of growth there were no new routes introduced by Nationwide. No new code share agreements were signed and no new airline entered the market. 1Time had commenced operations in 2004, just prior to the announcement. Nationwide submits that the only reason for the growth in passenger numbers over this period was the shift in discretionary business as a result of the termination of the override agreements. This, it is contended, is what would have occurred from the outset, absent the abuse. It should be noted that although the growth in passenger numbers was significant for Nationwide, it represented only 2% of SAA's passenger numbers.

[40] From early 2005 Nationwide's passenger numbers grew steadily. It entered into its own override agreements with some travel agents. These were halted in late 2005 when it was discovered that agents were 'double dipping' by charging passengers a booking fee when they were being paid a 7% override commission by Nationwide. As a result Nationwide announced that its commission would drop from 7% to 1% as from January 2006. According to Whittle the travel agent sector behaved in much the same way as it had when SAA made its announcement and Nationwide's BSP dropped. Unbeknownst to Nationwide, SAA had implemented an amended incentive scheme which further contributed to a decrease in passenger numbers.

[41] Nationwide won awards in 2005 and 2006 for the best domestic airline from ASATA, the Association of South African Travel Agents. It also successfully completed an IOSA safety audit in July 2006. This is a pre-requisite for IATA membership and is globally accepted as the benchmark for airline safety. Notwithstanding this, the death knell of Nationwide was the 'separation' of an engine on one of Nationwide's Boeing 737's on take-off from Cape Town International Airport on 7 November 2007.

Nationwide's Maintenance Issues

[42] There is a dispute as to the exact cause of the engine separation in November 2007. Roger van Putten, a former maintenance manager from Nationwide, went to inspect the scene of the incident immediately thereafter and testified that the cause of the separation was the result of the acoustic liner being ingested into the engine. He stated that this caused a massive surge or stall resulting in the engine twisting off the wing. This was in contradiction to the CAA report on the incident, which found that the cause of the incident was the failure of an aft cone bolt through fatigue. The latter was suggestive of poor maintenance rather than an unforeseeable once-off incident.

[43] After the CM investigation into the engine separation, an emergency airworthiness directive (AD) was issued for the grounding of all Boeing 737's in order that their cone bolts could be inspected. Although this was applicable to all the airlines, it was particularly detrimental to Nationwide as they utilized more 737's than the others. The CM documents reflect that after the engine separation, some serious maintenance issues were identified with Nationwide aircraft in mid to late 2007. On 30 November 2007 the CM grounded the entire Nationwide fleet and the airline effectively ceased operations with the exception of a month or two. Nationwide went into liquidation on 29 April 2008.

[44] To some extent maintenance issues dogged Nationwide throughout its existence. Their first aircraft, the BAC-111's, had to be replaced with Boeing 737's because they were not fitted with an automatic drop down oxygen system, only a manual system which required the passenger to plug the oxygen mask into the overhead panel if necessary. Whittle himself made reference to perceptions in the aviation industry that Nationwide had an aging fleet. He believed that any negative safety perceptions as a result thereof were unfounded and that Nationwide was uncompromising in ensuring that regulations and standards were maintained at the highest level. He pointed out that in 2006 Nationwide passed the stringent and rigorous IOSA audit after a scrutiny of Nationwide's entire operation and finding only one minor non-compliance out of thousands. This, he said, was unheard of in the airline industry.

[45] Van Putten, who was employed as a line station maintenance manager for Nationwide at all their stations and based at Johannesburg International Airport from February 1996 to late 2007, also gave a favourable account of Nationwide's maintenance standards and compliance with audit requirements of the CAA. He explained that scrutiny and oversight was not only conducted by the CAA but also provided by the manufacturers of the different components of the aircraft, as well as internal and external audits.

[46] The evidence of poor public perceptions of Nationwide's maintenance came from an employee of SAA in corporate sales, Ms Donna Kritzinger ("Kritzinger''), who was employed by Comair during this period. She said that amongst corporate customers, Nationwide's second hand aircraft were seen as a safety concern. Kritzinger was constrained to concede that there were articles about SAA's aging fleet over the same period which would have had an equally detrimental effect on public perception. Mortimer of the Tourvest Group said although it was not as large as SAA or even Comair, Nationwide was an "airline you could do business with.....it was a competent airline". An SAA witness, Nicolaas Vlok, formerly employed as deputy chief executive in charge of operations, described Nationwide as "formidable, not formidable that's maybe the wrong word but they were a good competitor ..." In summary that there were some concerns about Nationwide's older aircraft seems to be undisputed but it is unlikely that this was a significant factor in reducing the volume of passengers.

[47] The engine separation took place 18 months after the end of the abuse period. Although a lengthy time was spent by Mr Pretorius, counsel for SAA, canvassing the cause of the engine separation with various witnesses, it was stated in argument that the purpose was merely to show that the perceptions regarding Nationwide's poor maintenance were justified. The inordinate amount of time spent on this issue was unwarranted in view of the findings already made by the Tribunal and the CAC. The Tribunal alluded to the bad publicity which Nationwide had received and observed that it appeared to have a poorer safety record than that  of  its  competitors.  Nonetheless, said  the  Tribunal,  that Nationwide continued to have good growth even though it had an inferior safety record, did not mean that SAA's conduct failed to have a foreclosing effect[14]•

[48] Likewise the CAC held that while Nationwide's safety record and financial difficulties may have been to blame for the drop of its market share during the relevant period, it was SAA's incentive agreements that had the anti-competitive effect. [15] In short both the Tribunal and the CAC acknowledged the shortcomings in the safety record of Nationwide but nevertheless found that SAA's abusive conduct was the major cause of the decrease in volume of Nationwide's passengers. Those are findings which cannot be faulted but, in any event, to which this court is bound.

[49] Accordingly, insofar as it is alleged that the reason for Nationwide's loss of passenger volumes was the perception that it operated an aging and unsafe fleet, these are irrelevant in light of the above findings of the Tribunal and the CAC. These make it clear that the override agreements had an identifiable and separate anti-competitive effect of inducing travel agents to divert passengers away from Nationwide to SAA. The override agreements were found to have had an anti-competitive, exclusionary and foreclosing effect  on Nationwide, based on the actual market's outcome and the assessed strength of the competition in that market.

Damages

[50] Any damages suffered by Nationwide would amount to its lost profit over the relevant period. It is common cause that this involves a comparison of the actual situation in the relevant markets with the hypothetical position or the so­ called  counterfactual  scenario  in  the  same  markets,  absent  the  abuse  of dominance. In simple terms the lost profit is the difference between what Nationwide would have earned but for SAA's abusive conduct and what Nationwide in fact earned. "The damage is then the difference in the wealth of economic actors in both scenarios."[16]

[51] The numerous variables to be taken into consideration make it an impossible exercise to quantify the damages with any precision. However, this is not an unprecedented situation in delictual claims. Our courts have previously referred to future loss of earnings in personal injury matters, as being the preserve of fortune tellers and soothsayers because predictions as to the future are required which are, by their very nature, speculative. [17] A court is enjoined to do the best it can on the material available, even if it is merely an estimation. However, a plaintiff is obliged to produce all evidence at its disposal to assist the court in making as accurate a decision as possible.[18]

[52] There are various possible methods to quantify damages arising out of anti-competitive conduct. The Practical Guide on Quantifying Harm and Actions for Damages in the European Union[19]  at para 123 stated:

"It should be stressed that it is only possible to estimate, not to measure with any certainty and precision, what the hypothetical non-infringement scenario would have looked like. There is no method that could be singled out as the one that would in all cases be more appropriate than others. Each of the methods described above has particular features, strengths and weaknesses that may make it more-or-less suitable to estimate the harm suffered in a given set of circumstances. In particular, the methods differ in the degree to which they are simple to apply, in the degree to which they rely on data that is the outcomes of actual market interactions or on assumptions based on economic theory and the extent to which they take into account factors other than the infringement that may have affected the situation of the parties."

[53] Essentially what a court has to do is compare the performance of Nationwide before and after the abuse period to try and reach some estimation of how it would have performed absent the override agreements. SAA contends that Nationwide in fact performed worse once the abuse ended and therefore no damages were suffered as a result of the anti-competitive conduct. Affuso is of the view that, if any damages have been suffered, the interpolation approach is a reasonable methodology to estimate an upper limit of the damages.

[54] Various economic models have been placed before court and depending on the model used and the variables taken into consideration, the figures vary from zero damages to as high as R355 million. Noble produced four economic models - the linear trend[20] and the linear interpolation[21] using both passenger numbers and market shares. He then pooled the results to obtain a median of R170 million. Although it is apparent that each model has its own particular attributes, the linear interpolation is the one model agreed upon by the parties. Therefore for the purposes of this judgment the method relied upon to determine damages, if any, is the linear interpolation.

Oxera's Approach

[55] Noble uses a three-step methodology to determine the final damages amount. A diagrammatic representation of his framework is set out hereunder:

NB: Please consult the PDF version for the images

[56] The first step is to estimate the passengers Nationwide would have carried but for the abuse. The actual passengers are known. The lost passengers amount to the counterfactual passengers minus the actual passengers. This will be discussed in greater detail when dealing with the linear interpolation which is the preferred methodology to determine the number of lost passengers.

[57] Noble's next step is to ascertain the lost revenues as a result of the lost passengers. Essentially this amounts to Nationwide's counterfactual revenue minus its actual revenue. The actual revenue is not in dispute. The counterfactual revenue is estimated by multiplying the number of lost passengers by the ticket price Nationwide would have obtained for the sale of tickets to the lost passengers. Noble has used the average revenue per passenger that in fact flew on Nationwide flights over the relevant period. It should be noted that Affuso rejects the two step approach to establishing lost revenues and contends that it is unnecessary to calculate lost passenger numbers as the only reliable indicator of lost revenue is the BSP data. This issue will be dealt with in greater detail later when determining the appropriate dataset to be used.

[58] The third and final step in Noble's methodology is to ascertain the lost profit which is the lost revenue multiplied by the profit margin Nationwide would have obtained on each ticket. Essentially this amounts to lost revenues multiplied by the relevant profit margin less avoided cost. It is agreed by the experts that the estimated profit Nationwide would have earned from flying additional passengers is 50-51%.

[59] For the estimate of lost profits Affuso relies on the same methodology employed by Noble to calculate lost profit from lost revenues. This method involves applying an incremental profit margin to lost revenues.

[60] In estimating the profit margin it is important to determine how the extra passengers would have been carried. There are three possibilities - that they would have been carried on existing flights; that additional flights would have been put on utilising the existing planes; or new planes would have been leased or purchased. All have a different cost implication, with the profit margin being highest where the passengers are carried on existing flights. The cost of flying additional passengers on an existing flight is minimal but once additional flights are required, the costs of flying these passengers is much higher and the profit will be commensurately lower.

[61] Four separate cost categories were identified by Noble. The first is the cost incurred with every additional passenger transported and has a direct link with the individual passenger (such as travel agent commission, complementary meals and drinks).The second category is the costs incurred with every additional sector flown and is associated with individual flights (such as fuel and air traffic control costs). If an additional plane is required then costs will be incurred in the procurement thereof (such as monthly leasing costs). The final category of costs is those incurred independently of the number of passengers (such as overheads for the head office and marketing).

[62] Noble contends that the additional flights for Nationwide in the infringement period would have been accommodated on existing aircraft and there would have been no necessity to lease or purchase additional aircraft. Affuso disagrees with this contention on the basis that Oxera's analysis relies on an average load factor whereas in reality load factors vary significantly depending on the time of day and the time of the week and month. By averaging the load factor this obscures the fact that the demand at peak periods such as early morning and late afternoon or on a Friday or Monday will be far greater than at other times thus requiring additional aircraft to carry these passengers.

[63] This is known as the 'avoided costs framework'. The advantage of this method is that one does not have to evaluate the entire financial situation of the company, merely an estimate of those costs that have not been incurred because of the infringement[22]. In essence what is being done is to estimate the costs that would have been incurred by Nationwide had the extra passengers been carried (but were not as a result of the abuse). Noble used Nationwide's actual costs over the relevant period to calculate the avoided costs. To the extent that they were not actually incurred, these costs are hypothetical: they were not incurred but avoided. Mr Pretorius took issue with Noble on his comment that avoided costs were hypothetical. This line of reasoning was also raised in argument. What Noble meant was that the avoided costs were hypothetical in the sense that they were never incurred (because of the loss in passenger volumes) as opposed to the actual costs which were incurred.

[64] These hypothetical costs are then deducted from the revenue that would have been earned absent the abuse and multiplied by the agreed profit margin. This figure will amount to the damages suffered.

[65] In their final joint minute Noble and Affuso agreed on the avoided costs framework where the lost profit amounts to lost revenues less avoided costs. The lost revenues equal Nationwide's counterfactual revenues minus its actual revenues; the avoided costs are the difference between Nationwide's actual costs and the costs that would have been incurred but for SAA's conduct.[23]  It was further agreed that Nationwide's profit margin would have been 50-51% subject to one caveat as to whether additional aircraft would have been required to carry the additional passengers.[24]

[66] Affuso was at pains to point out that the agreement on the profit margin was premised on the underlying data being correct. This, she says, she was unable to validate. Before dealing with Affuso's other criticisms of the Oxera approach it is appropriate to deal with the dispute over avoided costs. A finding that Nationwide have not proved their avoided costs would be largely dispositive of the case.

Avoided costs dispute

[67] The question of avoided costs became an issue during the course of the trial. There is agreement that avoided costs amounted to the difference between the costs that Nationwide actually incurred and the costs that Nationwide would have incurred but for SAA's abusive conduct. It is not this formula that is in dispute but whether Nationwide adduced any admissible evidence of its costs. Despite agreement on the profit margin, SAA's case is that the avoided costs have not been proven. Any agreement Affuso might have reached with Noble on the profit margin was dependent on the underlying data being proven. This, it is argued, Nationwide has been unable to do. As the plaintiff, it has not established its avoided costs and absolution from the instance should accordingly be granted.

[68] The data Nationwide used to ascertain the avoided costs came from 3 sources of cost information: the route profitability spreadsheet, management accounts and audited financial statements. The financial statements are for the entire period of the abuse and aggregate all types of costs, including those for international and regional routes. The management accounts are available on a monthly basis, although February 2004 is missing. They provide partially disaggregated data. The route profitability is the document that Noble primarily relied upon when determining the avoided costs. It is disaggregated with all input and output data recorded individually.

[69] The main complaint is that Nationwide presented no admissible evidence of its costs other than the audited financial statements as confirmed by its auditor, Isak Buys "(Buys)", and Bricknell. It is contended that the management accounts and more particularly the route profitability spreadsheet, which were utilised by Noble in preparing his avoided costs data, amount to nothing more than inadmissible hearsay evidence.

[70] The route profitability is a voluminous spreadsheet providing very detailed cost and revenue data per route, per month and the number of passengers. It was a contemporaneous document prepared over a period of many years and utilised as a management tool in order that Nationwide could set its fares on the different routes. It listed costs such as fuel, air traffic control fees, ground handling fees aircraft, maintenance and crew accommodation. Noble used the line items of costs actually incurred from this document to identify the costs Nationwide would have incurred had it flown additional passengers.

[71] Griffiths was a director of Nationwide and the financial manager who drew up the route profitability on the instructions of Bricknell. The purpose of this document was to enable the performance and profitability of the airline to be monitored and assessed on a regular basis. Bricknell testified that it was a 'live' document that was updated every 10 days, or even 5 days if necessary. The management team would get together to discuss the overall management of the airline using the route profitability spreadsheet. For example it was used to make decisions as to whether cheaper tickets would be added to a particular route; whether to operate certain routes and to alter frequencies of flights.

[72] Mr Gotz, counsel for Nationwide, stated from the bar that Griffiths had refused to testify without substantial remuneration. Mr Pretorius pointed out that as a director Griffiths owed a fiduciary duty to Nationwide and contended that an adverse inference be drawn from the fact that he did not testify. Nationwide was criticised for not calling other available witnesses who could have shed light on the actual costs but whom Nationwide failed to produce.

[73] The following examples are cited as persons or evidence which could have been produced to prove the actual costs. Ms Barbara Buchanan ("Buchanan") who occupied the next most senior position in the finance department could have been called to testify on the route profitability spreadsheet. Van Putten testified that each aircraft had its own credit card for fuel which was carried in the cockpit of the plane - this real evidence could have been produced. It was confirmed by the liquidator that SARS downloaded Nationwide's server and they could get back-ups of all the financial information - this data could have been provided.

[74] SAA submits that instead of calling these witnesses, reliance is placed on the route profitability spreadsheet. It is alleged that the costs and standards contained therein reflect serious discrepancies and anomalies which point to the actual costs having been significantly underestimated. The further criticism is that there are significant differences between the audited financial statements and the route profitability report which Griffiths should have been called to explain.

[75] According to Mr Gotz, Nationwide was first alerted that costs were to be disputed a few days before the commencement of the trial. Prior to that, because both experts had agreed to the profit margin at 50 - 51%, it was assumed that avoided costs were not an issue. Neither was Nationwide alerted that this would be an area of dispute during the judicial case management process, which the parties were obliged to undergo before the trial could be allocated for hearing. In short Nationwide contends that the avoided costs had been agreed or at least not pertinently disputed until the eleventh hour.

[76] Insofar as the agreement on profit margin was dependent on the caveat of whether additional aircraft would be required, Noble dealt with this in his first expert report[25] and  concluded that the additional passengers could be accommodated on the existing aircraft. Using the actual load factor during 1999-2004 of 60-70%, he said that all counterfactual passengers could be carried on the remaining 30-40% available capacity for most months. However, additional flights would be required for certain periods. He was not cross-examined on this aspect.

[77] Bricknell testified extensively on how the route profitability spreadsheet was generated. The ticket sales on a particular route were obtained from the revenue accounting department and the expenses would be obtained from Buchanan. From that information the average ticket price could be extrapolated as well as the load factor. A target figure would be generated for every flight and ticket prices adjusted accordingly depending on the rate they were being sold. This was monitored on an hourly basis.

[78] Bricknell explained that the direct operating costs such as the costs in respect of approach fees, landing fees, ground  handling, ACSA  parking fees, crew accommodation, distribution costs and revenue costs were extracted from the actual invoices. He stated that the costs of the individual routes were calculated as follows:

"Well, the aircraft and leasing would be broken down by the monthly - by the hours flown on that particular route. The fuel, as I said, was from an average of each route. The aircraft fees would be the actual fees, the approach fees. The in­ route charges would be taken from the invoices, the landing fees. Ground handling - that was a cost recovery from Ground Support. Catering was for actual catering uplifted for the flights pertaining to that route. Crew accommodation would be for crew accommodation for those flights. And the liquor and drinks - that would be part of the catering. Parking fees would be the ACSA account, and distribution costs again the sale of ticket prices."

[79] SAA's biggest complaint is with regard to fuel costs where there are said to be various discrepancies and inaccuracies. In this regard, Bricknell said the fluctuation in the fuel price was monitored because there was a levy on the fuel price which was adjusted with every increase or decrease in price. He went on to say:

"We knew what the fuel burn on each sector was so that we would have averaged out because on each sector every day you get different winds, assuming Cape Town I Joburg you always have headwinds going down to Cape Town and as a result you always have tailwinds. So the headwind would vary and if it varies going down you would make it up coming back, so the overall time on both sectors would almost average out to be the same. So we had to average that out because some days we did about 54 flights, so to take each invoice for each uplift was very, very difficult and when an aeroplane arrived it arrived with so much fuel and then we needed to uplift sufficient fuel to take us to the destination airport, the alternate airport, and then have a diversion fuel - not a diversion fuel, the alternatives for the diversion. Then we had to have a contingency fuel, which l think was 5% to allow for varying weather and that would be the fuel required. So the fuel required less the fuel on board would be the uplift and that would vary all the time because depending where the aeroplane had come from it would depend on how much fuel it had remaining in the tanks so it was always that calculation that was involved. We had uplifted the amount of fuel and we'd go to the destination report. If we had no diversion and everything went according to plan, we land with the fuel that we estimated we would land with, and that fuel was recorded from the fuel gauge so there again it was as accurate as it could be. We had to just take the reading off the fuel gauge and so the repeat for the next flight would be exactly the same so that fuel was based on an average fuel per sector."

[80] Buys from PwC (South Africa), who audited the books of Nationwide in its latter years, described the route profitability report as a management tool used by all airlines.  He said it was particularly useful to Bricknell in deciding which routes to close and which to continue with. It should also be noted that Nationwide made full discovery of the hard drives of Buchanan and Griffiths which contained all the relevant information.

[81] Noble observes that if one compares the management accounts with the route profitability, many items are similar. However, the costs in the route profitability are about 4% higher than the equivalent in the management accounts, which means that the costs are arguably more conservative in the route profitability. Because of this, and as an alternative calculation, Noble provided recalculated costs based on the management accounts. This lowers the final damages amount by 6%. Flown domestic and regional revenues from the route profitability data are very similar to the management account data. The output data in the route profitability spreadsheet, such as monthly passenger statistics, is within 3% of ACSA data over the relevant period.

[82] Noble analysed the audited financials with the management accounts and found a close correlation. The sale/revenue figures were identical. The costs in the audited financials were slightly lower. Noble indicated that there was a 3% difference, which he did not consider significant.   Noble testified that the various line items had  been checked  with  Nationwide  staff over  a  period  of several years. The fuel price had been checked against the monthly price advisory from BP, Nationwide's fuel supplier, all of which were produced as exhibits.

[83] It has long been established that in some cases damages are difficult to ascertain with any precision, but this does not relieve the wrongdoer of the necessity of paying damages. In Esso Standard SA (Pty) Ltd v Katz 1981 (1) SA 964 (A) the plaintiff was the owner of a garage and filling station who claimed damages for loss of petrol due to a leaking petrol tank. The defendant sought absolution from the instance on the basis that the plaintiff was in a position to lead evidence to show the quantum of damages but elected not to do so and instead relied on a schedule of damages. The question was whether the plaintiff had produced all evidence reasonably available to him. The Appellate Division upheld the trial court's decision that it had no cogent reason for doubting the accuracy of the records on which the schedule was based. Similarly, in this case I have no reason to doubt that the route profitability spreadsheet is, to a substantial degree, an accurate reflection of Nationwide's costs.

[84] In any event the evidence relating to the route profitability is covered by section 3 of The Law of Evidence Amendment Act 45 of 1988. The amendment was introduced as a response to the tendency of South African courts to exclude evidence which was reliable, but which could not be admitted because of the rule against hearsay.[26]  The amendment confers a judicial discretion on courts to admit hearsay evidence if it is in the interests of justice. Six factors which must be considered are set out. [27]

[85] Section 3 of the Law of Evidence Amendment Act provides:

(1) Subject to the provisions of any other law, hearsay evidence shall not be admitted as evidence at criminal or civil proceedings, unless:

(a)...;

(b...;

(c)     The court is of the opinion that such evidence should be admitted in the interests of justice, having regard to:

(i)   the nature of the proceedings;

(ii)   the nature of the evidence;

(iii)  the purpose for which the evidence is tendered;

(iv) the probative value of the evidence;

(v) the reason why the evidence is not given by the person upon whose credibility the probative value of the evidence depends;

(vi)  any prejudice to a party which the admission of such evidence might entail;

(vii)   and any  other factor  which  should  be taken  into account."

[86] Nationwide submits that on each requirement of section 3(1)(c), it has led sufficient evidence warranting the admission of the route profitability spreadsheet. As regards the nature of the proceedings, it is contended that it is not open to SAA to deny a victim of established anti-competitive harm, damages to which it is entitled on the basis that there ought to be better possible evidence relevant to the claim. The nature of the evidence and the purpose for which the route profitability was tendered is to show a contemporaneous business record created by Nationwide's senior management for their use, based on actual cost data, actual revenue data and actual passenger volume data. Because one half of the route profitability spreadsheet (that relating to revenues) is uncontested by SAA, it is not open to them to dispute the probative value of the other half. Nationwide submits that it has provided a sufficient explanation why it was unable to secure the evidence of Griffiths, upon whose credibility the probative value of the evidence depends. It has also explained why the underlying flight coupons, invoices, receipts and other primary source documents are unavailable. (The liquidator testified that after the liquidation of Nationwide in 2008, these documents were removed by SARS and a potential purchaser of the business). Finally, it is argued that SAA has yet to show what prejudice it suffers by the admission of the route profitability spreadsheet.

[87] I am satisfied that the evidence has shown that the figures contained in the route profitability spreadsheet are largely reliable. Further, that it is an accurate reflection of the incremental costs of operating the airline in the relevant time period. The level of de-aggregation of the costs in the spreadsheet makes it a particularly useful record from which one can extrapolate the avoided costs of Nationwide.

[88] It is clear that the route profitability is a reliable record of the revenue, passenger volumes and cost inputs of Nationwide. I am not persuaded that in the absence of the testimony of Griffiths the route profitability spreadsheet has not been proven. Nor indeed, whether calling him would have considerably enhanced the weight to be given to the document. It is unlikely that Griffiths would have been able to confirm or deny the accuracy of every one of the thousands of cells in the route profitability spreadsheet. He may have been the author of the document but he compiled it from the diverse sources of information provided to him.

[89] To call every person who had an input into the route profitability spreadsheet  or  even to  produce  every  original  invoice  would  have  been  an unwieldy and not very productive exercise. While there may be some discrepancies, the margin of error is small. Perfection in these circumstances is unachievable. Viewed holistically, the conclusion must be reached that the route profitability is admissible on the basis of the evidence before court. Even if this were not the case, the route profitability should be admitted on the the basis of its reliability in terms of section 3 of the the Law of Evidence Amendment Act 45 of 1988. Because Noble conceded that the costs, when compared to the management accounts, could be considered slightly high, I am inclined to utilise his lower recalculated figures.

PwC's Approach

[90] Affuso's primary submission is that no damages have been suffered by Nationwide. Her opinion is that the incentive agreements could not have any impact other than in the sales through travel agents. She disregards sales via other channels such as direct sales through call centres and internet sales. She uses only the sales generated by travel agents which are recorded in the BSP revenue data provided by IATA. In addition Affuso only uses the data on those routes where Nationwide competed with SAA. Her reasoning is that passengers could only be diverted on routes where SAA, SAL and SAX overlapped with Nationwide.

[91] With reference to the BSP data on those particular routes, Affuso uses the ex post period - the period unaffected by the infringement - to compare the trend in Nationwide's market share. From this she concludes that no damages were suffered by Nationwide. The crux of her argument is if, as Nationwide claims, its market share was significantly reduced by the SAA override agreements then one would expect Nationwide's market share to decline in the infringement period, coupled with a corresponding increase in the ex post period. In fact, she says, the direct opposite occurred. Nationwide's market share increased during the infringement period and declined in the ex post period.

[92] Affuso contends that if one looks at the period after SAA had removed the overrides in April 2005, Nationwide's BSP market share does not appear to register an increase, as would be expected if the abusive conduct was having the alleged impact. Instead Nationwide's BSP share experienced a significant decrease following the removal of SAA's override agreement. She pointed out that from April 2005 until March 2006 Nationwide's average BSP share of the market decreased to 7.9%. From the period March 2006 to September 2007 it continued to decrease to 4.3%. On this basis Affuso concludes that the incentive agreements did not foreclose the markets.

[93] A diagrammatic representation of her argument is set out hereunder from which it appears as though Nationwide's BSP revenue dropped after the anti­ competitive conduct ended. It should be noted that this graph does not take into account the sharp spike at the end of 2004 when SAA's announcement was made that overrides were to be terminated. A fundamental problem with this hypothesis, and an issue which will be dealt with in greater detail later in this judgment, is the now accepted fact that the BSP data in the ex post period was corrupted. The further criticism is whether one can make a valid comparison between the abuse period and the ex-post period.

NB: Please consult the PDF version for the images

[94] Affuso's analysis reflected on the above diagram is that there was no increase in Nationwide's BSP share in the ex post period and therefore no damages suffered by Nationwide. She criticises Oxera for inflating lost revenue by accounting for other distribution channels and domestic routes unaffected by SAA's overrides. She accuses Oxera of making no adjustment to counterfactual passenger numbers in order to account for other factors like LCC's and internet bookings.

[95] It is the very criticism which Affuso levels at Oxera which reveals the fundamental flaw in her ex-post analysis, namely that she ignores the fact that the ex-post period was not a valid comparator. Instead of the reliance on travel agents as the primary distribution channel for airline tickets, the ex-post period was characterised by significant development in the LCC's. While Kulula and 1Time commenced operations during the abuse period in August 2001 and February 2004 respectively, it was in the ex-post period that there was a burgeoning growth in the LCC market. Mango also entered the market in the ex­ post period in November 2006. This is evidenced by LCC's gaining a market share of 11% during the abuse period as opposed to 23% after the abuse period.

[96] As a consequence of the unprecedented growth in LCC's, the rise of ticket sales via the internet grew from less than R40 million to just under R120 million in the ex-post period. To suggest that the market was the same in the ex­ post period, and accordingly a valid comparator with the abuse period, would be factually incorrect. The airline industry was in constant flux at the time. The ex­ post period is characterised by the waning role of travel agents' centrality in the purchase of airline tickets. Moreover because of having been foreclosed in the time-sensitive market, Nationwide's non-time sensitive passengers would be far more susceptible to the introduction of the LCC's than other airlines. The changes are best reflected in two graphs which show the growth of LCC's over the period and the growth of internet banking sales.

[97] A diagrammatic representation of these trends is best observed in the two graphs below. The first graph reflects LCC passengers based on ACSA data and includes Kulula, 1Time and Mango. The FSC (full service carriers) are shown as the darker blue. The second graph reflects internet sales as share of total sales in the market. The total market includes Nationwide, SAA, and Comair. SAX, SAL and LCCs are excluded. (The data for this graph is Nationwide sources of sale, Nationwide domestic revenues, SAA yields and Internet data, SAA domestic revenues, Comair point of sales data - Oxera analysis.) From the two graphs it is apparent that the abuse period and the ex-post period are not comparable. Due to changes in the market it was inevitable that there would be some decline in BSP revenue. To extrapolate from Nationwide's decrease in BSP revenue over this period that no damages at all were suffered is an inaccurate assumption.

[98] The growth in low cost carriers is particularly marked in the ex-post period.

NB: Please consult the PDF version for the images

[99] In light of the concessions that ex-post period BSP data is incorrect and because the ex-post period is not an appropriate comparator period, Affuso's analysis that there has been zero damages is not sustainable. As an alternative to her analysis on the ex-post period, Affuso applies the interpolation method to show that Nationwide's damages were minimal. It is difficult to marry Affuso's two methods, which on the face of it produce contradictory results: her interpolation shows some damages were suffered, although greatly reduced, whilst the ex-post analysis shows zero damages.

[100] Initially Affuso interpolated the counterfactual scenario between the ex­ ante market share of 6.4% based on the three relevant routes and the ex-post share of 7.9% based on an average of six relevant routes. This produced an initial figure of R5.9 million which increased by R2 million with the inclusion of SAL and SAX. It has been revised as high as R58 million based on her most recent calculations taking into account the misallocated BSP data.

[101] A diagrammatic representation of her interpolation method is seen hereunder. It should be noted that the BSP data used is for all relevant routes where Nationwide and SAA competed. On Affuso's interpolation, Nationwide's only damages are identified as those where the pale grey line (the actual BSP passengers on the relevant routes) dips below the pink dotted line (the counterfactual BSP passengers on the relevant routes).

NB: Please consult the PDF version for the images

Nationwide actual BSP share, relevant routes Nationwide counterfactual BSP share, relevant routes

[102] To understand the extent of divergence between the two experts using the linear interpolation method, the diagram below shows a comparison of Oxera and PwC's damages. The reason for this is that PwC have used BSP data in the relevant routes while Oxera have used passenger numbers for all routes.

NB: Please consult the PDF version for the images

[103] As the interpolation methods of Noble and Affuso produce such divergent results, it is evident that before any meaningful interpolation can take place this court has to make certain determinations firstly, regarding the correct data set to be used and secondly, the appropriate averaging methods to be applied to the interpolation. Both are a fundamental point of departure between the experts, and one which has a significant impact on the outcome of the calculation of damages.

Appropriate Data Set

[104] The first significant area of difference is whether to use data from only those routes where Nationwide competed with SAA or the entire domestic market. Affuso's model uses only those routes which both SAA (including SAX and SAL where appropriate) and Nationwide flew. This, she submits best gives effect to the Tribunal's and the CAC's findings that the effect of the overrides was felt predominantly in the travel market sector. On the other hand Noble advocates the use of total data for the entire domestic airline travel market which he says best adheres to the decisions of the Tribunal.

[105] The Tribunal found that:

"While the foreclosing effects of its conduct were greater in this [travel agent] segment of the market, competition in the overall  domestic  airline  travel  market was reduced by SAA's incentive scheme."[28] (my underlining)

[106] Whether to use only those routes where SAA and Nationwide competed was considered and rejected by the Tribunal, which found that the override agreements applied to all SAA domestic flights across the country. Nor were they limited to only those routes where SAA faced competition from rival airlines but applied to potential competition on routes where rivals had not yet introduced scheduled flights.  For this  reason the Tribunal made the finding that the two domestic air travel. It is noteworthy that Bricknell's evidence at the Tribunal was that when Nationwide wanted to open new routes they were advised against it by travel agents. What is clear is that on a correct reading of the decision of the Tribunal, and as confirmed by the CAC, it is not open to Affuso to merely use those routes where SAA competed with Nationwide.

BSP v Passenger Numbers

[107] Interrelated to the debate of relevant routes versus the total domestic market is whether total passenger data or BSP data should be used in the calculation of damages. Noble uses passenger numbers for the whole of the domestic airline travel market. Affuso's approach is to use the BSP data on all routes where SAA and Nationwide competed. Therefore PwC uses revenue data to estimate passenger numbers whereas Oxera is more concerned with volume of passengers lost. Having determined that the total domestic market is the relevant market, this court must then determine whether BSP or passenger numbers is the appropriate dataset.

[108] Because BSP data records all sales made via travel agents, Affuso contends this approach is in line with the findings of the Tribunal. She criticises Oxera's data as producing unrealistically high predictions of counterfactual passengers. By amending Noble's methodology to confine it to travel agents sales and the relevant routes, the damages estimate is drastically reduced.

[109] Noble contends for the use of actual passenger numbers to estimate Nationwide's counterfactual passengers. This method is said to be useful because it relies only on data regarding Nationwide and not the other airlines. SAA questions the reliability of this approach for precisely the same reason - it ignores the rest of the market.

[110] Affuso's criticism of the use of passenger numbers is that it fundamentally ignores the fact that revenue depends upon both price and passenger numbers (which, in turn, depend on price). Hence, by separating both passenger numbers and price (or yield), Noble is assuming passengers would not respond to prices. This, she says, is fundamentally wrong as it ignores the law of supply and demand. In addition, avers Affuso, like the argument regarding the relevant routes, it ignores the fact that the Tribunal concluded that any effect would be felt primarily in the  travel agents' market. By using BSP data which relates directly to sales deriving from travel agents, which is the segment of the market covered by the override agreements, Affuso claims to be adhering to the Tribunal's finding that the major effect of the overrides was to be felt in the travel agents sector.

[111] Nationwide provides several reasons why the use of BSP data is incorrect. The first argument is that using BSP data is impractical in view of the fact that Nationwide's BSP data is not available before January 2001. To add to this, BSP data is not on a route-by-route basis. Affuso divides the total BSP revenues by total flown revenue to calculate a percentage share and then applies this percentage to flown revenue on individual routes to estimate a BSP sales revenue figure on each route. As a result the missing Nationwie BSP data must be constructed on the basis of assumptions, which are not necessarily correct.

[112] Noble correctly points out that firstly, it assumes a 1:1 relationship between BSP and flown revenue for travel agents, which is not correct. Secondly, this calculation assumes that the flown revenue across all  routes flown by all the airlines is consistent, which is not necessarily true. For example a route such as the Johannesburg/Cape Town route is likely to have a higher ratio of travel agent sales than other routes. Comair's BSP data also only commences in July 2000, which causes further difficulty in correctly dividing the BSP data between the different airlines.

[113] Moreover, there is no BSP data on passenger numbers, only BSP aggregate revenues. These values need to be converted to passenger numbers which require assumptions about the relevant yield. In contrast, the passenger number data sets used by Noble are admitted and accepted as accurate. They are available for the entire period unlike the BSP data. They are disaggregated for each of the relevant airlines and on a route-by-route basis. Hence there is no need for assumptions.

[114] Perhaps the most compelling criticism of the use of BSP data is the anomalies in the BSP data for SAA, SAX and SAL, which emerged during the course of the trial. The discrepancies in the data, although apparently known to Affuso much earlier on, were picked up late in the day by Nationwide's legal representatives. It is not disputed that from July 2004 to March 2008 the SAL and SAX BSP revenues are lower than the flown revenue. This is clearly indicative of an error in the data. Over this period there is a significant decline in SAX and SAL's BSP data. As from April 2008 the BSP data is significantly higher than flown revenue (as one would expect).

[115] The BSP data sales of all the airlines operating at the time - SAA, SAX, SAL, Comair and Nationwide - is agreed. If the incorrect data from SAA, SAL and SAX is added to the data from Comair and Nationwide, an aggregate BSP data set is produced, which is very similar to the data sets provided independently by the five airlines. This means, submits Nationwide, that the corrupt data was adjusted to cure the anomalies in the SAX and SAL's data set. The only plausible explanation for the anomalies is that the SAL and SAX data was misallocated to SAA, which would have the effect of artificially inflating SAA's BSP data to the detriment of Nationwide, whose market share would be artificially suppressed.

[116] Affuso did not dispute that the SAA data was incorrect and attempted to cure the problem by reconstructing the SAL, SAX and SAA data sets. She estimated the missing SAL and SAX data by using the average proportion of BSP revenue to flown revenue for SAL and SAX in the months where the discrepancy does not exist. Alternatively, she interpolated the missing SAL and SAX data. On these two bases Affuso concluded that the damages suffered by Nationwide amounts to R16.9 million to R17.3 million. [29]

[117] It was pointed out by Nationwide at the time that while the SAL and SAX data was adjusted, there are no changes to the SAA data, which assumes that the missing data was not misallocated to SAA, contrary to all indications. This has the effect of erroneously increasing the total BSP, which is known and agreed. The correct approach would be to subtract the reconstructed SAL and SAX data from the SAA data, which results in the correct aggregate total BSP data from the five airlines.[30]

[118] During the course of a subsequent matter - Comair v SAA[31] (where Comair is similarly claiming delictual damages for anti -competitive conduct over the same period) evidence emerged which confirmed the stance taken by Nationwide. According to further submissions[32] received from Affuso in July 2016, Janaurieu D'Sa of IATA testified that IATA collates BSP data from travel agents and then remits this data to the individual airlines. The BSP data for SAA, Nationwide and Comair is correct. This means that SAA's BSP was incorrect. Any correction to SAL and SAX BSP data must be accompanied by a corresponding decrease in the SAA BSP data. This was exactly the contention of Nationwide. On the basis of this evidence Affuso recalculated the damages using the corrected data from January 2004 and January 2005, and using the start dates of the linear interpolation as being October 1999 and April 2000. The damages increased to amounts ranging between R28.9 million to R48.5 million.

[119] What the above indicates is that Nationwide's criticism of SAA's use of BSP data is clearly justified. Nationwide successfully made out a case that the SAA BSP data was corrupt, and the initial attempts by Affuso at curing the problem significantly underestimated the impact it would have on any final damages figure. Subsequent evidence in the Comair case has proven this to be correct. For this reason alone the BSP data is too unreliable to use as the appropriate dataset. However, and equally important, using BSP data on the relevant routes fails, in my view, to give effect to the findings of the Tribunal that while the effect of SAA's conduct was felt primarily in the travel agent sector, the conduct had the effect of reducing competition in the total domestic airline market.[33]

[120] Affuso was unable to provide a satisfactory answer as to how effect should be given to the Tribunal's finding, which is binding on this court. She repeatedly stated that she had found no evidence that there had been any impact on markets other than the travel agents' sector. In my view this is tantamount to suggesting that the findings of the Tribunal must be disregarded. While it may be that this court finds that no damages were suffered in the domestic market as a whole, it is not for an expert to decide this and merely omit the relevant data from her methodology.

[121] The Tribunal found that the effect of the override agreements was to shift passengers who might have flown with Nationwide, to SAA. The effect is one of volume. Therefore the obvious starting point is the passenger numbers. I am in agreement with Noble that passenger number data best reflects the Tribunal's finding that the total domestic market was affected. 

[122] Noble's alternative method, and one supported by SAA if BSP data were to be rejected, is the market share approach. This analysis also relies on passenger numbers to ascertain the market share. It involves estimating Nationwide's counterfactual share by focusing on the split in the total passenger volumes between the three main airlines rather than actual passenger numbers. Nationwide's counterfactual market share is then converted into counterfactual passenger numbers. This involves looking at the entire airline domestic market - Comair, Nationwide and SAA (including SAL and SAX). According to Noble this exercise was done across the relevant period for the whole of the domestic market in order to remain aligned to the Tribunal's decision. It did not include LCC's because they had a very different business model to attract customers who have never flown before. It looks at Nationwide's market share before and after the abuse period to estimate what share it would have obtained during the abuse period. Nationwide contends that the market share approach shows that the market in absolute terms continued to grow; Nationwide also grew but would have grown more absent the abuse.

[123] While based on passenger numbers, the total size of the market in the counterfactual scenario is assumed to be the same as the actual scenario. Therefore all market factors at play during the relevant period are considered. In light of the significant changes taking place in the aviation industry at the time, it is important that any methodology estimating counterfactual passengers must not only reflect seasonality but also properly reflect the changing trends in the airline industry.

[124] The distinct advantage of using market share is that it captures the impact of changes taking place in the market as a whole. For example, if the overall market increased over the holiday season this method would estimate the higher passenger numbers for all three airlines and not only Nationwide. Similarly, the impact of the entry of the LCC's such as Kulula and 1Time would be accounted for. In this manner, the outcome of market processes that affect people's choices are effectively embedded in the analysis. It would also take into account the impact of the price wars that took place in the travel industry between 2002 and 2004 as testified to by Mortimer.

[125] Having rejected the argument for the use of BSP data on the relevant routes, I am persuaded that Nationwide's market share would be a more appropriate dataset to utilise, rather than passenger numbers. My conclusion is therefore that any interpolation must be based on market share data on all routes, rather than BSP data on routes flown by SAA and Nationwide.

Lost revenue dispute

[126] The other area where the opinion of the experts diverged widely is how to calculate lost revenues. As alluded to previously, Oxera determines lost revenue in two steps. Noble uses actual passenger number data to estimate the counterfactual passenger numbers, which are then converted to counterfactual revenues by multiplying them by Nationwide's actual average revenue per passenger. Affuso disputes that it is necessary to identify the lost passengers because on her model the only relevant lost revenue is the BSP revenue. She estimates Nationwide's counterfactual revenue as the counterfactual BSP revenue for the relevant routes. Noble therefore uses two steps in his analysis which focuses on passenger data while Affuso uses actual revenues to estimate the counterfactual revenues in one step.

[127] Affuso prefers the one step approach for a number of reasons. Firstly, because yield masks the considerable differences in the ticket fares paid by passengers, ranging from business class to the cheapest economy fare. Secondly, the number of passengers depends on the fare and if the fare goes up the passenger numbers will go down. By looking at revenue only, in Affuso's view, all those interactions are taken into account. Whereas if the two components are separated and assumptions are made on yield and passenger numbers separately, the impact on yield cannot properly be determined. This was a matter of much debate during the trial.

[128] On the other hand Noble considers the single step approach inappropriate because revenue data is a combination of the revenue per passenger and the number of passengers. For example Nationwide could earn revenue of R1000 by selling 10 air-tickets at R100 each, but the same revenue could be earned carrying 5 passengers at R200 each. Yield and revenue do not necessarily follow the same trend. The yields were higher during the abuse period than they were before and after. This could therefore mask a drop in passenger numbers.

[129] It was further contended that using BSP data to interpolate can mask a decline in Nationwide's passenger numbers as a result of SAA's abusive conduct, thereby underestimating lost revenue. This is because the BSP data is the end result of a combination of the passenger numbers and the price of the ticket sold. It does not take into account that the price is subject to external variables such as inflation and fuel price. Simply put, if the price of a Nationwide ticket increases to compensate for the loss of passengers as a result of the abusive conduct, the BSP figure will mask the fact that Nationwide has lost passengers.

[130] What Affuso's approach overlooks is that the impact of the overrides was felt in the volume effect, namely that passengers who would have flown with Nationwide were diverted to SAA. The anti-competitive conduct influences passenger numbers rather than price. Therefore, I am in agreement with Noble that Affuso's approach,  by using revenue alone, would have the effect of masking a drop in passenger numbers. In fact, as pointed out by the CAC[34], it is possible that Nationwide would have obtained a higher revenue per passenger in the counterfactual without SAA's conduct. This is an indication that Noble's figures may be somewhat conservative.

[131] In summary, the correct method in my view, would be to utilise the 3 stages suggested by Noble and to interpolate on the market share of Nationwide. Having determined the correct dataset to be utilised and the approach to be adopted, the number of lost passengers has to be ascertained. This is estimated by looking at the period before, during and after the infringement to estimate what the passenger numbers would have been, absent the infringement. As mentioned previously, there is more than one approach that can be used to assess the lost passengers (and lost revenue as a consequence of the lost passengers) but the linear interpolation is the one method agreed upon by both parties.

Linear I nterpolation Method

[132] What the interpolation method does is place two points on a graph, the first representing the before period and the second representing the after period. A straight line between these two points will indicate what the growth would have been but for the abuse. It can be used to estimate the counterfactual passenger numbers or the counterfactual market share. The stylized diagram below demonstrates the lost passengers as being the difference between the actual passengers (represented  by  the  solid  black  line)  and  the  counterfactual passengers, which is shown on the area between the solid line and the dotted green line between points B and C.

NB: Please consult the PDF version for the images

[133] Nationwide's interpolation method applied to market share is illustrated below. The shaded area represents the counterfactual market share for each month.

NB: Please consult the PDF version for the images

[134] The linear interpolation method involves comparing the pre-infringement value to the post-infringement value. An important determining factor in arriving at the final damages sum is where the pre- and post-infringement comparator periods should begin and end. Both experts agree that the calculation of damages should be made with reference to a time period unaffected by the abusive conduct, a so-called 'clean' or uncontaminated period. This is in line with the European Commission which states that:

"Another simple technique for deriving a comparator value from a range of data observations is linear interpolation. Where a comparison over time has produced price series from before and after the infringement, the 'non-infringement' or "counterfactual price during the infringement period can be estimated by drawing a line between the pre-infringement price and the post infringement price."[35]

[135] Noble and Affuso agree that a 12-month averaging period should be used to account for seasonal variation in the airline industry.[36] However there is disagreement on when the starting point of the interpolation (the ex-ante period) should commence and when the lingering effects of the abuse of dominance will come to an end (the ex-post period). The figures change dramatically depending on when one starts and ends the interpolation. Affuso uses a trailing averaging method which calculates the 12-month average using the 12 months before the abuse period and the 12 months after the abuse period. Noble uses a centred averaging approach where the beginning and end dates of the the abuse are used as a midpoint and the 12 months before and after these dates are used to calculate the 12-month average The first diagram shows the averaging period for the start of the interpolation. By using the Oxera averaging period for the start of the interpolation, PwC's damages calculation increases to R62 million. The second diagram reflects the averaging period for the end of the interpolation. By using the Oxera averaging period for the end of the interpolation PwC's damages increase by R48 million.

[136] The dramatic difference the averaging makes to the final figure is set out in the diagrammatic representations. The visual representations are the PwC diagrams using BSP data.

NB: Please consult the PDF version for the images

[137] Noble uses 12-month averages centered around the beginning and end of the infringement period, namely September 1999 and April 2005. Data for the six-month period before and after these points is utilized which means that inevitably data within the abuse period is used. In respect of the before point, his reasoning is that Nationwide, being a new airline, was growing rapidly at the time. Therefore, taking the 12 months before the infringement as the averaging period would introduce a growth bias. On the other hand Affuso starts the averaging period 12 months before the infringement period and argues that the interpolation should not overlap at all with the infringement period. She calculates Nationwide's counterfactual share with reference to a comparator period unaffected by the abuse. Her initial calculations used a linear interpolation between Nationwide's average market share over the twelve months before the infringement (October 1998 - September 1999) and the twelve months after the end of the infringement (April 2005 - March 2006). A diagrammatic representation of the effect of the different averaging methods is set out hereunder.

NB: Please consult the PDF version for the images

[138] Affuso criticizes the approach of Oxera as intellectually inconsistent and illogical because it utilises data within the infringement period. This she states is "technically incorrect and invalidates the approach". Moreover it has the effect of inflating the result.[37] However, in her testimony Afffuso stated that the real effect of the abuse was felt from April 2000 when all the offending incentive agreements  commenced,  bar one  - American  Express  which commenced  in October 1999. She indicated that as an expert she would "have no difficulty" starting the abuse period from April 2000. This means that the start point of the interpolation correlates with the Oxera's 12-month averaging from April 1999 to March 2000. In light of this concession I am of the view that the averaging period for the interpolation starting point should be the 12 months prior to April 2000.

[139] As regards the end point, Affuso's 12-month averaging period commences when the abuse period ends in April 2005 to end March 2006. She excludes data from July 2005 as SAA staff went on strike during July and passengers had no choice but to fly on other airlines including Nationwide, thereby inflating its market share. This was a once-off event that would distort the overall picture. Oxera's averaging period is centred on the end date of the infringement and therefore includes the six months before and after April 2005.

[140] The undisputed evidence is that once SAA made its announcement of the termination of its override agreements in November 2004, travel agents almost immediately diverted their discretionary business from SAA. As with the starting point, it seems that the date from which the abuse of dominance ceased to have an effect must be the correct starting point for the end averaging period. This broadly coincides with the averaging period used by Oxera. In the light of this evidence, I am of the view that the centred averaging approach of Noble is more appropriate and the end point averaging period should run from October 2004 until September 2005. However, I am in agreement with Affuso that the July 2005 data, when the SAA strike was taking place, should be excluded.

Contingencies

[141] Noble testified that SAA's abusive conduct would have had an effect beyond the travel agent sector and that the override schemes would negatively affect Nationwide's brand. I am persuaded that once a passenger is repeatedly persuaded by the travel agent sector that SAA is a superior airline this will create a bias in favour of SAA even when that passenger is buying his/her ticket through distribution channels other than travel agents. Repeated exposure to SAA in a passenger's business travel under a corporate agreement, does have the potential to influence the choices that passenger makes  as a leisure traveller. Travellers, who have only been exposed to SAA flights due to directional selling of SAA tickets by travel agents in the past, are likely to remain loyal to SAA regardless of the distribution channel through which they may purchase air tickets in the future. Brands matter because people's choices are influenced by their past behaviour.

[142] Further, while the effect of loyalty programmes may be limited, the lure of earning Voyager points, SAA's highly successful loyalty programme, cannot be totally disregarded. SAA argues that because Nationwide continued to grow during the affected period, this is an indication the anti-competitive effect of the loyalty programme was limited.[38]  I do not agree with this proposition. Once a passenger's business travel has been diverted to SAA and a certain number of miles accumulated, it is probable that many passengers would remain with SAA for leisure and other non-time sensitive travel in order to accumulate further Voyager miles.

[143] That some allowance must be made for brand loyalty is self-evident, but sight should not be lost of the fact that while the Tribunal held that the effects of SAA's anti-competitive conduct may have had the effect of reducing competition in the total domestic market, its effect was greater in the airline market distributed through travel agents. In short, the effect was felt predominantly in the travel agent sector but not exclusively. Any estimation of damages must take this finding into account.

[144] The question is really how much should be attributed to brand. In other words what percentage of the non-travel agent market should be attributed to the abusive conduct. Both the Tribunal and the CAC stressed that the main effect of the abuse of dominance was to be found in that part of the domestic air travel which was distributed by the travel agent sector. Despite the growth in internet and direct sales the Tribunal found that the travel agent sector accounted for 70% of the total domestic air travel market at the time.

[145] The question of contingencies was not raised by either expert. However, it is in my view the only way to account for the fact that any damages must take into the consideration the whole market but must bear in mind that the effects of the abuse was felt predominantly in 70% of the market (the travel agent sector). The issue of a contingency deduction was suggested by Mr Pretorius in his heads of argument. When a possible contingency deduction was put to Mr Gotz in argument his response was that the effect would be to reduce passenger numbers, which would make no difference as the gradient would remain the same, an argument that I do not accept. On being pushed he said a maximum discount of 25% should be applied to their damages calculation.

[146] In 2 Travel Group PLC (in liquidation) v Cardiff City Transport Services Ltd[39] the Competition Appeal Court in England considered a claim by the liquidators of 2 Travel for a loss of profit and a loss of capital assets. In para 397 of the judgment the court said the following:

"Of course, it is absolutely right that the Tribunal can only determine this case on the evidence before it, and cannot have regard to factual material that was not adduced before it. Neither Mr Good, nor Dr Niels nor Mr Haberman adduced such factual material. They provided expert opinion evidence. In particular, Mr Good and Dr Niels sought to assist the Tribunal in what sort of revenue would have accrued to 2 Travel had the infringement not taken place. We have found the work extremely helpful, and have taken it fully into account, but we certainly do not consider that the opinion evidence in the reports must be used on a 'take it or leave it' basis. It is for the Tribunal - based upon the factual evidence - to make an assessment of what would have happened in the counter-factual scenario, and this may very well involve re-working calculations done by the experts or adopting an approach which - although it draws on the work of both experts - adopts neither approach completely. This is what has occurred in this case. Our approach is neither that of Mr Good nor that of Dr Niels but - based upon the factual evidence we have heard - represents our concluded view as to what would have occurred in the counter-factual scenario."

[147] It lies within the discretion of the court to determine what contingencies should be applied. In my view a 25% contingency deduction applied to Nationwide's figures would properly account for the non-travel agent sector unaffected by the anti-competitive conduct.

[148] What remains is to deal with two further defences raised by SAA. The first relates to the settlement of the civil claim arising from the anti-competitive conduct during the period September 1999 to May 2001. The second is whether Nationwide's loss of profit can be attributed to poor management.

Settlement of the First Action

[149] One of the points raised by SAA is that Nationwide issued summons for payment of damages after the First Tribunal in the maximum amount of R269 million and a minimum amount of R43.7 million. This action was apparently settled for a fraction of the amount without going to trial. The effect of this is that Nationwide has been compensated for the damages it suffered until May 2001 which include Noble's starting point of October 1999. It is contended that Noble's calculation ignores the legal consequences of the settlement of the first action. The net result of this, so it is argued, is that Noble has used the incorrect starting point for the calculation of damages, which has been inflated not only by including 6 months after October 1999 in the averaging period but also by using November 2004 as the end point averaging period.

[150] Noble in his first report makes it clear that the counterfactual was calculated for the entire period of the anti-competitive conduct, from October 1999 to March 2005. However, he goes on to state that the damages values relate only to the second period as damages for the first period were claimed in separate proceedings. Only the lost passengers for the period 1 June 2001 to 31 March 2005 are included in the calculations and the damages for the first period are specifically excluded.[40]

[151] Later in his report Noble at para 4.36 goes on to explain:

"The damages estimate for the first period included a claim for two weeks of dissipation or lingering effect following the end of the period-Le. 1 June to 14 June 2001 (this was included because SAA's override agreements are likely to have had a continued effect beyond the official end of the period). In light of this, the amount corresponding to the two weeks of dissipation has been deducted from the second period damages estimate calculated in this report to avoid double recovery for this period."

[152] From the above it is evident that the submission that there has been some duplication in the claim is factually incorrect. Insofar as it may be suggested that the starting point for the interpolation does not take into account the first claim, it bears mentioning that both experts agreed that the interpolation should begin at the end of September 1999.[41] By necessity it would cover the period of the first claim for damages but this does not mean that the damages are in fact claimed for that period. In fact the contrary has been shown to be the case.

[153] In their third set of heads of argument[42] SAA changes the goalposts and alleges that the issue is not whether Noble incorporated the damages of the first period into the second period but rather that whilst damages of between R34.7 million to R269 million were claimed, the action was settled for a fraction of the amount. This is said to be indicative of the fact that the basis for the calculation of the initial amount was fundamentally wrong and this was not factored into the calculation for the second period. This is a novel proposition. What calculations were presented and what factors were taken into consideration at the time the claim for the first period was settled are not before this court. This is not anything this court can take cognizance of.

Poor Management of Nationwide

[154] One of the defences raised in its plea is that Nationwide's loss of profit was as a result of poor business management. No direct evidence was led by SAA, but Mr Pretorius argued that the best evidence of this is the fact that Nationwide's liabilities exceeded its assets each year of its existence. In addition, once the abusive conduct ended it increased its passenger numbers while substantially increasing its losses. The accusation is levelled that Noble's counterfactual scenario is fundamentally flawed with no bearing on the real world. Mr Pretorius dismisses Noble's counterfactual world  where  Nationwide makes a handsome profit when the passenger numbers increase. However, in reality Nationwide's losses increased when its passenger numbers increased. Both scenarios cannot be true, it is argued.

[155] Ashton and Henry at para graph 8.094[43] summarise how the European Union deals with factors extraneous to the anti-competitive conduct in the following manner:

"8.094 As discussed in section Ill. A, when quantifying damages, all those factors have to be controlled that bear no causal relationship with the incriminated behavior. The extent to which the factors have an impact on the profits follows from the respective multivariate regressions. The quantification of damages is thus corrected for the effects of these factors. An important factor in this context could be the business model used by the firm, which may differ in situations with and without abusive conduct. In principle, the econometric methods allow for an estimation of the impact of these factors on cost, revenue and profits and the determination of the extent to which the changes in profits can be explained by these factors. Depending on the case at hand, however, this may be difficult. For an estimation of the damages that is as precise as possible, it is, however, of central importance to identify all these factors and to control them in an econometric analysis. Otherwise it is possible that changes in profits will be attributed to the abusive conduct even though the changes will have had - at least in part - different causes. "

[156] In the European Guide reference, is made to a finance-based analysis where the financial affairs of the company pre- and post the abuse period are compared, to identify whether the company has made a profit during the infringement period. However, as Mr Gotz correctly points out, no case was made out for such an analysis. This was not the model used by PwC, and in fact Affuso makes no reference to such a model, but rather submitted that the interpolation was the appropriate model to use. It was not canvassed between the experts and cannot be considered by this court.

[157] If the financial performance of the company was to be regarded as an indicator of the damages suffered, then this model should have been put up by Affuso and evidence led on this aspect. The court cannot be asked to make this finding on the basis of the financials alone when both experts have agreed on the interpolation method as the appropriate model to estimate damages. Moreover, it is not Noble's evidence that Nationwide would have been profitable absent the abuse but that it would have earned a greater profit. Whether this would have meant that the business as a whole would have made a profit (on its books) or perhaps only shown less of a loss is entirely irrelevant for present purposes.

Conclusion

[158] A summary of my findings is set out hereunder. In order to give effect to the decision of the Tribunal that the anti-competitive effects of SAA's conduct were felt in the market for domestic airline travel, the appropriate dataset to use would be that of passenger numbers rather than BSP data, which only reflects the sales by travel agents. In the same vein, the Tribunal found that the anti­ competitive effect of SAA's conduct was not confined to those routes where Nationwide and SAA competed but also on routes where Nationwide did not fly. Accordingly, I reject Affuso's use of BSP data on only those routes where SAA and Nationwide competed and conclude that the appropriate data set would be that of passenger numbers in the entire domestic market.

[159] In assessing which passenger data better reflects the above findings, I am of the view that market share data is a preferable option to passenger numbers. Although based on passenger numbers, market share also takes into account the various market factors at play at the time. These include not only seasonal variations but also the growth of the low cost carriers and other distribution channels like internet and direct sales. I accept the two-stage approach of Noble as a more accurate reflection of lost revenue than the use only of BSP revenue data, and am in agreement with his three stage method to determine lost profit or damages. Insofar as it has been argued that the avoided costs have not been proven, this is rejected for the reasons set out herein.

[160] In line with the consensus of both Noble and Affuso, I accept that the linear interpolation model using market share on all routes is the appropriate methodology to apply. As regards the start and end points of the interpolation, for the reasons set out in this judgment l accept the averaging periods preferred by Oxera with the caveat that the data for July 2005, the time when the SAA strike took place, be excluded.

[161] To account for the fact that the Tribunal and CAC found that it was in the travel agent sector, which accounted for approximately 70% of the market at the time, that the effects of the abuse were predominantly felt, it is appropriate to apply a contingency deduction to the final damages figure. I am of the view that a 25% contingency should be deducted from any figure arrived at in order to make allowance for those passengers (and therefore revenue) which were totally unaffected by the overrides and any brand loyalty emanating therefrom.

[162] According to the tables provided by Oxera[44] the figure for interpolation using market shares October 1999 to March 2005 (centred averaging) adjusted for the July 2005 SAA strike and using the profit margin based on Nationwide's management accounts, the final damages figure is R 139.5 million (Model 01). The total damages for each model are set out hereunder.

INTERPOLATION - JULY 2005 STRIKE ADJUSTMENT AND PROFIT MARGIN BASED ON NATIONWIDE'S MANAGEMENT ACCOUNTS

Model   Model description	ModelDl   Base model-	ModelD2   Base model with	ModelD3   Adapted start	ModelD4   Adapted start	Model OS   Adapted start
	October 1999 (centred) to March 2005	trailing averages- October 1999	date-April 2000 (trailing) to March 2005	date-April  2000 (trailing) to March 2005	and end date- April 2000 (trailing) to
	(centred)	(trailing) to March	(centred)	(trailing)	October 2004
		2005 (trailing)			(trailing)
Interpolation using pax	217.4	171.8	193.5	181.6	245.3
Interpolation using	139.5	92.5	122.0	103.3	164.1
market shares
Interpolation average	178.5	132.2	157.8	142.4	204

[163] To this figure a 25% contingency deduction should be applied. I am accordingly of the view that Nationwide should be awarded damages in the sum of R104.625 million arising out of SAA's anti-competitive conduct over the relevant period.

In the result I make the following order:

1.     SAA is to pay to Nationwide damages in the sum of R104 625 million

2.      SAA is to pay interest on the said sum at 10.25 % as from the date of judgment until date of payment.

3.      SAA is to pay Nationwide's costs on a party and party scale including the costs of two counsel and the qualifying costs of the expert witness, Robin Noble.

_________________________

C H Nicholls

Judge of the High Court

Gauteng Local Division, Johannesburg

APPEARANCES

PLAINTIFF :                    Adv. A Gotz

                                            Adv. M Le Roux

                                            Instructed by Bowman Gilfillan

DEFENDANT :               Adv. G Pretorius SC

                                            Adv. B Manentsa

                                            Instructed by Cliffe Dekker Hofmeyr

HEARING DATES:        01 to 23 February & 30 March 2016

JUDGMENT DATE:       08 August 2016

---