---

IN THE COMPETITION TRIBUNAL

REPUBLIC OF SOUTH AFRICA

Case Number: 18/CR/Mar01

In the matter between:

The Competition Commission Complainant

and

South African Airways (Pty) Ltd Respondent

REASONS AND ORDER

Introduction

1. This case concerns the legality of two incentive schemes, which the respondent, South African Airways (Pty) Ltd (“SAA”), the country’s largest domestic airline, has with travel agents. The Commission brings this complaint referral pursuant to a complaint brought by the Nationwide Airlines Group (“Nationwide”) a domestic rival of SAA.¹ The Commission alleges that the incentives constitute an abuse of dominance designed to exclude or impede SAA’s rivals in the domestic airline market. The Commission seeks an order declaring that the schemes constitute prohibited practices and the imposition of a fine of R 100 million. SAA denies liability and has put all the issues in dispute. We have found that SAA has contravened Section 8(d)(i) and our reasons for this conclusion follow.

Background to the case

2. On 13 October 2000 Nationwide lodged a complaint with the Competition Commission against SAA.² In brief the complaint alleged that SAA was trying to exclude it from the domestic airline market by engaging in a number of practices that were prohibited under the Competition Act (the ‘Act’). Four anti-competitive practices were alleged – (i) SAA was engaged in predatory pricing (ii) SAA was poaching key staff (iii) SAA had concluded agreements with travel agents in terms of which they received commissions on an incremental basis that it alleged had an exclusionary effect (iv) SAA had a reward scheme for employees of travel agents, known as “Explorer”, which it was also alleged had an exclusionary effect.

3. These claims formed the subject matter of an interim relief application that Nationwide then brought unsuccessfully against SAA in October 2000. The reasons for the failure of Nationwide’s application are set out in our decision in Nationwide Airlines (Pty) Ltd and Others versus South African Airways (Pty)Ltd and Others and no more need be said of that here, although it suffices to say the incentive schemes which are at the heart of the present case were the subject matter of the interim relief application, although they were alluded to in passing.³

4. Subsequently, the Commission concluded its investigation into the Nationwide complaint. It referred the complaint to the Tribunal on the 18 May 2001. In its complaint referral the Commission relies on only two of the alleged restrictive practices that were in the original Nationwide complaint, viz. those that relate to the incentive schemes for travel agents and the Explorer scheme. The other practices complained of have not found their way into the case before us nor has Nationwide pursued them by way of a non-referral.⁴ It would appear that the reason for the Commission’s selection of certain practices to constitute the basis of its present referral, are that similar practices have been scrutinised in cases in other countries and the Commission has sought to rely on this jurisprudence in this case.⁵

5. It is worth noting at this stage that although this case is most commonly associated with Nationwide it is not confined to it. On 23 August 2001 the Commission amended its complaint referral to refer, inter alia, for the first time to the alleged exclusionary effects of the scheme on the complainant and “other competitors”.⁶ It is common cause that the only other competitor at the time was Comair Limited a company which operates a passenger service in the Southern African region. By virtue of a licence from British Airways, it uses the name BA/Comair. We will for this reason refer to this firm as BA/Comair.

6. According to the Commission the abuse that is alleged in this case commenced in about April 1999 and by the end of the hearing (December 2004) was believed to still be continuing. Nevertheless the evidence we have had presented in this case has not always correlated with that period or with any consistent time period. The Commission has provided some figures for the period ending March 2001 (travel agents sales figures and sales of airline tickets at Johannesburg International), others for the period ending May 2001(Table B in figures bundle 2 a comparison of travel agency flown revenue and BSP), some for June 2001 (Table E2 which relates to passenger information on BA/Comair), and yet others until October 2004 (Table G figures bundle 2 which relates to passenger sales on Nationwide).

7. Perhaps the reason for this unevenness in selection is that information was collected at different times during the long life of this case and earlier information was not updated to conform to a common endpoint.⁷ We do not wish to exaggerate the difficulties in these inconsistencies as some of the information is of less probative value than others or the use of different periods has been appropriate because the data is being employed to illustrate different points.

8. Nevertheless from the point of view of fairness the case had to be pinned down to a finite period. For reasons that will become clearer later it is not the existence of the schemes in question that is pertinent but their nature, which has changed over time. For this reason, we have decided that for the purpose of assessing the duration of the abuse, we shall assume that the evidence of its existence commences in October 1999 and ends in May 2001, the latter date being the date, according to the Commission, when its investigative period ended.⁸ We will refer to this from now on in the decision as the ‘relevant period’. This is a period for which most of the more important information on effects is presented, although we will, for the purpose of interpreting information, make use of figures that come before and after that period. It is common cause however that the Explorer scheme ended in June 2002 and that the override scheme was still in existence at the end of this case.

9. Thus what we are saying is that it may well be that the effects of these two schemes may have been in existence for long after our reference period, but we believe that it is necessary to confine our findings to a finite period which corresponds with a period where evidence on market shares, sales of tickets through travel agents and effects on rivals’ sales of tickets can reasonably be correlated.

10. As will be evident from the date that this complaint was lodged by Nationwide (October 2000), and the date we heard final argument, (5 March 2005), this case has taken a long time to conclude. The blame for this delay has itself been a subject of contention between the Commission and SAA, and is discussed more fully below. It suffices to say at this stage that we find it highly undesirable for litigation to take so long to reach conclusion and that it satisfies neither the interests of complainants, consumers or respondents.

Synopsis of our approach

11. We first examine the operation of the two schemes at issue in these proceedings, as this is necessary to understand the case before us. We then examine briefly the theory of harm advanced by the Commission and SAA’s response to it. We then consider the various elements of the case that the Commission needs to establish in order to prove a contravention. We first, as is customary, analyse the relevant markets, then consider if the respondent is dominant in these markets and then move on to consider the abuse. The section on the abuse first considers arguments on what the legal test is and then examines the factual issues in light of that conclusion.

12. Finally, since we have found that SAA has contravened section 8(d)(i), we consider what remedy is appropriate.

PART I – THE MERITS

The incentive schemes

13. The three airlines that competed in the domestic market during the relevant period, all made use of travel agents’ services to sell domestic tickets, for which they paid by way of commissions.⁹

14. Initially it appears that the structure of the commissions was quite straightforward and travel agents received a standard basic commission. At some stage, and it is not clear from the evidence exactly when, but certainly well prior to the relevant period, airlines began introducing what is known as an override incentive scheme for paying commission.

15. The SAA override incentive scheme works in this way. Agents receive a flat basic commission for all sales up to a target figure that is set for them in the contract. The target figure is expressed in rand value. If they exceed the target they become eligible for two further types of commission that are paid over and above the basic commission, which continues to be paid on sales over the target. The first category is what is termed the ‘override commission’. This is an additional commission paid if the agent meets and exceeds the target. However, the override commission is not limited to the amount above the target, but is payable on the total of all sales achieved above and below the target. Thus assume the firm has a target of sales set at R 100 million. If it exceeds this target by R 10 million it will receive an override commission, typically set at 0,5%, on all it sales i.e. it will amount to R 550 000. Note because the firm continues to receive its basic commission of 7% as well, the average rate would now be 7,5%. Because the override commission is payable on all sales earned, even those below the target, it is referred to as the ‘back to rand one’ principle. In some contracts the override commission is set at a constant rate, in others, it is subject to a continual increase as the firm’s sales reach continually higher targets. Thus in the American Express contract the override rate is a constant 0,5%.¹⁰ In the Luxavia agreement, the override rate increases the more the agent exceeds its target. ¹¹ The rate starts at 0,5% when the firm reaches its target, but moves to as high as 1,55%, if it exceeds its target by 25%. The figures show that if it attains its peak override commission at this level, it would not only earn a base commission of around R 37 million, it would also earn an override commission on top of that of R 8 million.

16. But there is also a second category of commission for which the firm that exceeds its target is eligible and this is referred to as the ‘incremental commission’. If the travel agent earns a certain percentage of sales above target it then becomes eligible for this additional commission. This commission, unlike the override commission, is payable only on the amount above the target and is not therefore back to rand one, but is ‘back to rand base’. For this reason the incremental commission rate is typically much higher than the override and base commission rate, and in some agreements, is subject to escalation as well. Going back to our two models; in the American Express agreement, its incremental commission kicks in when it achieves sales in excess of 15% of its base target. The incremental rate commences at 14% for sales 15% above target, but rises steeply so that sales 35% above target are rewarded at a rate of 31%. Note the effect this has on the respective commissions. When American Express gets to 15% above target its basic commission is about R6,3 million while its incremental commission (14%) amounts to R1.6 million. When it reaches its peak at 35% sales above target its basic commission is R7,4 million, but its incremental commission (31%) is now R 8,5 million thus exceeding the basic commission.

17. The Luxavia agreement also has an incremental commission although it is structured differently. Here the incremental commission kicks in when sales are 5% above target, although the rate starts at only 5% and increases to a rate of 20% for sales that exceed the target by 25%. It thus differs slightly from the American Express structure in that the incremental commission starts becoming payable earlier, but at a lower rate and reaches a lower peak commission.

18. The target is a key feature in these agreements. Agencies do not face a common target that they must meet. Each firm is set a custom made target based on its previous annual sales figures with a percentage increment. Even the increments are not uniform and appear to be the subject of negotiation between the firm and SAA. In the agreement with American Express, the agency is required to achieve an annual growth of 25% from its target in the preceding financial year.¹²

19. It is not clear from the evidence as to whether SAA introduced the override scheme into the market or whether they reacted to what British Airlines was doing internationally. Nothing turns on this. What we do know is that when SAA first introduced the scheme there was no complaint in the market place. Mr Viljoen, the chief executive officer of South African Airways at the relevant time, has testified that he found these types of agreements in place when he joined SAA and that an agent had told him that they had been around since 1980.¹³

20. In October 1999, SAA, according to the evidence of Viljoen, adopted a more aggressive approach to the override scheme. In Viljoen’s words, he was not prepared to reward agents for generic growth i.e. growth that came about from inflation, as opposed to an increase in sales. SAA remedied this by firstly reducing the basic commission from 9% to 7%. Then it made the attainment of override and incremental commission more challenging, either by raising targets annually over and above the rate of inflation, and/or raising the point at which incremental commission became payable. At the same time, it increased the rate of the incremental and override commission.¹⁴ It is common cause that in order to retain their previous level of profitability, agents would not only have to exceed their targets, but also exceed them by some margin to take advantage of the override and incremental commissions. However for those who could ascend the peaks the rewards were bountiful. It is this mechanism and the alleged incentives it entails that are at the heart of the Commission’ s case on the abuse.

Explorer Scheme

21. The Explorer scheme rewarded individual travel agent consultants with a free international air ticket based on their achieving SAA’s sales targets. Conceptually it resembles frequent flyer schemes for passengers. The crucial distinction between the override scheme and Explorer is that Explorer is targeted at employees of the agency while the override scheme is aimed at the firm and hence its shareholders benefit.

22. SAA was the only airline offering this type of incentive to local travel agents. Mr Venter, the Financial Director of BA/Comair, testified that his airline could not match such a scheme because BA/Comair’s volume base of sales was too low in relation to that of SAA, to make it viable.

23. A second aspect of the Explorer Scheme was a bonus pool that allocated points to an agency as whole, based on the sales of all its consultants. More points are obtainable depending on the agency’s share of total SAA sales. According to Viljoen, this aspect of the scheme incentivises not only the individual consultant, but also the staff of the travel agency as a whole.¹⁵

24. The Commission state that the Explorer Scheme was in full operation during the relevant period and, in conjunction with the override incentive scheme, aggravated the anti-competitive effect of the override incentive schemes.

The Commission’s case

25. The Commission’s case is that the abuse of dominance relates to two relevant markets. The first is the market for domestic scheduled airline travel and the second is the market for South African travel agency sales of domestic scheduled air travel in South Africa.¹⁶ SAA is alleged to be dominant in both. It alleges that SAA has used its dominance in the travel agency market to impose on travel agents a system of compensation that not only rewards them in terms of an unobjectionable basic commission, but also rewards them additionally by means of commission calculated on the override and incremental incentives we referred to above. For the sake of convenience we refer to these latter two additional methods of compensation, collectively, as the override incentive scheme. The Commission argues that the effect of the override incentive scheme is to induce travel agents to sell more SAA tickets and less of those of its rivals when the agent has an opportunity to do so. The reason is that as the agent sells more SAA tickets its, rewards increase significantly. These rewards would be foregone if the agent instead sells tickets of SAA rivals. Thus agents have a compelling financial incentive to prefer to sell an SAA ticket to a customer over that of a rival. Crucial to the Commission’s case, is the fact that SAA re-designed its override compensation in 1999 by reducing the basic commission and increasing rewards via the override and incremental incentives, but travel agencies needed to achieve much higher levels of sales before these additional rewards became payable to them. Once attained however, the rewards became increasingly lucrative.

26. How does the travel agent manage this if the customer wants to fly with a rival? The Commission argues that the travel agents are not always able to influence the customer’s choice, but can do so frequently enough for their intervention to matter. The reason agents have this ability, is that airline ticket prices are so volatile that they are not transparent to customers and hence they are willing to accept the agent’s advice. The increased business that the scheme brings to SAA does not come so much out of new business, but rather at the expense of SAA’s rivals who, because their compensation schemes are less lucrative, since they sell less tickets, will never be in a position to reward the agent in the way the SAA scheme does. Thus the agent earns more by selling its next ticket on an SAA flight than on a Nationwide flight.

27. This, says the Commission, is what makes the conduct exclusionary in nature. The imposition of the Explorer scheme on top of the override scheme serves to enhance its exclusionary nature as the Explorer scheme operates at the level of individual consultants and employees.

28. The Commission goes on to argue that because travel agents can and do distort consumer choices to accomplish their own commercial objectives, this leads to two competitive harms. The first is that consumers in the short run will be flying on more expensive tickets and at less preferable times than if the ticket offering had been unbiased. Secondly, that SAA is able to perpetuate its existing dominance and to restrict new entry into the market and to inhibit its existing rivals from expanding in the market.

29. The net result of an anti-competitive exclusionary strategy is a less competitive market in which there are higher fares, less choice for consumers and less innovation.¹⁷

SAA’s case

30. SAA embarked on a war of attrition against the Commission’s case. Not only does it dispute the Commission’s approach to market definition as we later discuss, but it goes on to deny that it is dominant, even if these definitions are accepted. Secondly, it contests the notion that travel agents either have the inclination or the ability to move passengers away from their airline of choice to SAA. Thirdly, SAA disputes the Commission’s case on the effects of the schemes. SAA argues that not only has the Commission failed to establish a causal link between the schemes and the expansion of SAA in the market, and the corresponding demise of its rivals, but that it has also failed to demonstrate harm to consumers.¹⁸

31. SAA does not deny having re-designed its override scheme in 1999, but claims that this was not done to introduce exclusionary incentives, but rather to lower the costs of its distribution system by making agencies more efficient and not to reward them for sales growth that they had no part in i.e. increases due to ticket price inflation.

32. Finally SAA invokes the efficiency defence contemplated in sections 8(c) and 8(d). It argues that even if the scheme is found to be anti-competitive, it nevertheless enhances the efficiency of its distribution network because travel agents are incentivised to become more familiar with SAA products and hence better able to guide consumer choice.

The Relevant Markets

33. As this case concerns an alleged abuse of dominance, it is trite law that the Commission needs to establish that SAA is dominant in respect of some market for the conduct alleged to be abusive to be unlawful. In most abuse cases only one potential market is implicated and the relevant market debate turns on whether it has been defined with sufficient precision with respect to potential substitutes. In this case the debate is somewhat different, as we are dealing with the relationship between two possible relevant markets. These are not alternative market definitions, as SAA in its heads of argument suggests, but interdependent markets – without the one, the abuse could not be effected, without the other, the exclusion would be ineffectual.¹⁹

34. In our view this is an aspect of the case that SAA has failed to grasp and hence its counter-definition of the relevant market as one premised on specific routes at specific times ignores the possibility of the relationship.

35. That more than one market can be implicated in an abuse case is not novel in competition law. Cases in the European Union have dwelt upon these possibilities. Whish states that in the EU, it is not required that the abuse, dominance and the effects of the abuse all occur in the same market. For instance, in the Commercial Solvents case, Commercial Solvents ceased supplying its downstream customer with a raw material to manufacture a particular drug, since it wanted itself to enter the downstream drug-manufacturing market.²⁰ It was found to have abused its dominant position in the market for the supply of the raw material in order to better its own position in the downstream drug market, in which it had no presence, let alone being dominant. In the European cases of tie-ins, it is common for the abuse to be perpetrated in one market, while the effect is felt in another market.²¹

36. There is nothing in our Act that suggests that an abuse of dominance cannot be perpetrated in one market and the effect thereof be experienced in another related market. Any contrary interpretation would mean that a dominant firm could leverage its market power from one market into another, with impunity.

First relevant market – travel agency services

37. The Commission first alleges that there is a relevant market for “South African travel agency sales of domestic scheduled air travel in South Africa”. ²²

38. Let us see how they get to this conclusion. Airlines use travel agents to sell their tickets. To do this they remunerate travel agents for their services by way of a commission. This is not the only model for their remuneration, but was during the period of complaint.²³ Airlines have various options besides travel agents for selling their tickets. They may do so directly themselves, or use some other method such as the internet. The Commission argues that the latter two options are not adequate and hence not competitively relevant substitutes for the services of travel agents. In any event, the evidence is that during the relevant period 70-85% of domestic airline tickets, depending on the airline, were sold through travel agents.²⁴ It is also the evidence of Viljoen that if travel agents did not provide this service, each airline would have to set up satellite offices to provide these services. He states:

“This of necessity would require enormous capital expenditure and overheads which could only be recouped by airlines by way of possible fare increases. Such a result is clearly detrimental to passengers.”²⁵

39. There is a prior question – is the travel agent, the agent of the customer or the airline? The question is answered by understanding what we are analysing. Where the practice complained of relates to the effect of a remuneration scheme for travel agents, then the appropriate definition is that of a market in which airlines purchase ticket distribution services from travel agents.

40. The best evidence for the centrality of travel agents as a distribution mechanism is that of Viljoen who, in an effort to justify the override scheme, has argued that for airlines to duplicate these services by creating networks that replicate those of travel agents would be prohibitive.

“If I had to replicate that overnight, I don’t know how I will, we will have to treble, as I said our IT platforms. We would have to employ staff at a huge cost and train them.”²⁶

41. The reason for this is fairly clear, as the Commission argues. Internet sales, at least at this time, did not account for a significant number of sales during the relevant period. This is evidenced by the airline’s own Internet sales data: in 2001, SAA sold less than 0.3% of its ticket revenue through its own website.²⁷

42. Direct distribution channels are not an alternative for consumers who want to examine their choices. Thus although these channels are alternative means for airlines to sell tickets to consumers, they were not during the relevant period, satisfactory substitutes for consumers shopping for the best available options on domestic flights and the preponderance of consumers choosing travel agents over the other options speaks most powerfully to this point. Nor indeed is it likely that airlines would need to bother with incentive schemes if this outlet was not of such centrality. The best evidence for their centrality during the period is the significant percentage of each of the three airlines’ tickets sold through travel agents during the relevant period. (See Table 1 that appears below)

43. The figures show that all three airlines relied, during the relevant period, on travel agents for the sale of the bulk of their domestic airline tickets. By comparison, other vehicles for ticket sales cannot be regarded as competitively significant substitutes.

44. We find that first relevant market is the market for the purchase of domestic airline ticket sales services from travel agents in South Africa.

Dominance

45. Having defined this market we now turn to the question of whether SAA is dominant in this market.²⁸ Table 1 below sets out the respective sales of SAA, BA/Comair and Nationwide for one year during the relevant period, July 2000 to June 2001. These figures come from tables prepared by the Commission based on documents discovered by the three respective airlines. SAA has not challenged the veracity of the figures, insofar as they purport to represent the respective ticket revenues from sales through travel agents, and they constitute a useful proxy for the market shares in the travel agent market during the relevant period.²⁹

46. SAA’s dominance as a seller of tickets emerges from these figures. Not only does it account for 65,7% of total sales but also 69% of sales through travel agents.

Table 1: SAA’s Market Shares Relative to its Competitors *

	SAA	BA/Comair	Nationwide
Sales Domestic	65.7%	27.6%	6.6%
Airline’s Sales through travel agents	69%	25.3%	5.7%
Proportion of sales through travel agents relative to total sales	85%	74%	70.2%

Source: Tables A.1-A.3 Pages 1-3 Figures Bundle 2- * (July 2000 to June 2001)

47. As Table 1 shows, SAA’s sales dwarf those of its two domestic rivals. This difference in volumes is relevant to the theory of exclusion that the Commission advances. It is not only SAA’s absolute size that matters, but also its size relative to that of its rivals.

48. SAA sales constitute over 45% of sales of domestic airline tickets through travel agents and hence it is presumed to have market power in terms of section 7(a).³⁰

Second relevant market - Market for domestic airline travel

49. The Commission’s second relevant market is the market for “scheduled domestic air travel in South Africa.”³¹

50. This relevant market definition was the subject of great contestation. SAA disputes the way the Commission has defined the market not, as is usual, in a debate over possible substitutes, but at the conceptual level as to what market is relevant given the nature of the complaint. Secondly, even if the Commission’s definition is accepted, SAA disputes the method that the Commission has adopted to count market shares.

51. SAA has not presented a consistent position on the relevant market. In its answering papers, Viljoen argues for a market for the provision of domestic air travel services in the conveyance of passengers on particular domestic airlines routes, on particular flights, at particular times on particular days.³² Its experts, whose report came later in the proceedings, had a different view. They defined two types of relevant market. Firstly, that of all domestic flights on three city-pairs where SAA and Nationwide compete, with a separate market for business travel and another for leisure travel in each of these city-pairs.³³ Secondly, the market for distribution of domestic airline tickets.³⁴ In final argument, SAA adopted the stance of Viljoen and contended for the market for the provision of domestic air travel services in the conveyance of passengers on particular domestic airline routes on particular flights at particular times on particular days. SAA also argued that the channel to market whether by means of travel agents, internet or direct booking is immaterial to the service being bought and the relevant market.³⁵

52. We agree with the Commission that this approach to the relevant market is flawed because it fails to properly account for the abuse being advanced. In short, the relevant market is wrong because it is not relevant to the theory of harm being advanced. Were the case one of excessive pricing in respect of a category of airline tickets, then finding sub-markets in respect of particular classes of passengers on city to city pairs at particular times, may be feasible. It is not however pertinent at all to the market in which the abuse is being experienced (the travel agent market) or the market where the exclusion finally takes place, domestic air travel.

53. In our view the Commission has correctly defined the second relevant market as being the market for scheduled domestic airline travel, as this is the market where, if the behaviour is exclusionary, the final effect will be experienced. It will not be experienced in any narrower sub-market, as the exclusionary effect, if it exists, is experienced across the range of city-to-city pairings, passenger classes and flight times. Although SAA’s rivals are not present on all domestic routes, and at all times, they are potential competitors in respect of all scheduled flights where they do not already have a presence. The evidence is that the number, destination and time of flights is continually changing so that we cannot view the market as a static model where rivals only contest route times and schedules that they are on at a particular moment in time.

54. The override scheme applies to all domestic tickets sold by SAA uniformly, not to specific classes of tickets. If it has the effect the Commission contends for, namely that it incentivises agents to sell SAA tickets at the expense of its rivals, then this behaviour will be felt in the related market of domestic airline travel as whole. It can affect the sale of any of its rivals’ tickets on any route, at any time or on any class. SAA’s distinction may be of relevance if the abuse was perpetrated in respect of customers in the domestic airline travel market, but on these facts, it is not. The abuse occurs in the related market and its effects are experienced across the domestic airline travel market as a whole.

55. This market-wide effect was conceded by Viljoen:

“MR PRETORIUS “So in other words, do you agree with me that the same effect that is in the whole market will be in any single city pairing that we may speak about, if there is such an effect?

MR VILJOEN: Yes, there is. In that case, yes.” ³⁶

56. We find that the second relevant market is the market for domestic scheduled airline travel.

Dominance

57. Perhaps the most contested terrain in this matter has been the Commission’s figures and methodology in arriving at its conclusion that SAA is dominant in this market.

58. Dominance in a market may be calculated using various determinants. Most commonly the method is based on the relative sales revenues of the firms in the particular market. Whilst sometimes other figures are used, number of goods sold etc, this if often because sales revenue figures are not available, rather than the fact that they are not considered a reliable statistic for the purpose of determining market share.

59. In this case, the Commission has sought to establish the fact that SAA is presumptively dominant in the market for domestic airline travel by making use of data on ticket sales revenue. This is not an uncomplicated task as the discussion will show.

60. The first question before we examine the data is whether to attribute the figures of SAX and SAL, two other regional airlines that operate in the domestic market, to SAA. SAX refers to South African Express Airways (Pty) (Ltd). At the relevant time, Transnet owned 76,01% of SAX, with the balance held by Thebe Investment Corporation. SAL refers to South African Airlink (Pty) (Ltd). At the relevant time, SAA had a 10% interest in SAL with the balance held by private shareholders, unconnected to SAA or the government.³⁷

61. The Commission argued that since SAX and SAL have some common ownership links with SAA, share ground-handling facilities, code sharing, common sales outlets and common branding, we should conclude that they acted in concert with SAA. More importantly, the evidence of Viljoen suggests that these airlines operated as satellite services complementary to those of SAA and hence did not directly compete with it.³⁸

62. The Commission further makes the point that, insofar as the market for travel agency services is concerned, many of the override agreements concluded during the relevant period, including those of the largest agencies, included sales of SAX, SAL, or both, within the target figures. This means, the Commission argues, that from the perspective of these agents, the differences between these firms were irrelevant and hence it is proper to count them as part of SAA’s share for the purpose of the travel agency market.

63. In the market for domestic air travel, since we find that SAA is dominant without including the market shares of either SAX or SAL, we need not decide this particular issue, although we supply figures for both in Tables 2 and 3 below. In relation to the travel agency market, attributing their market shares to SAA seems correct, given that for the purpose of achieving their compensation levels, travel agents could include sales on SAX and SAL in their SAA totals.

64. The first source of data on ticket sales is the revenue generated by travel agents on tickets that they sell. The method by which this is done is called the BSP or bank settlement plan. When an agent sells a customer a ticket they take the first coupon in the ticket pack and send it to the BSP, which computes an amount owing to the airline. This amount, which reflects gross sales by travel agents, is referred to as ‘BSP revenue’. When the passenger crosses the gate at the airport and boards the aircraft for the flight in question, the coupon that is in the boarding pass is collected and is processed. That information is captured in an information system and is referred to as ‘flown revenue’.

65. BSP revenue is always higher than flown revenue for the equivalent number of sales. This is so for a number of reasons. In the first place, flown revenue excludes revenues owing to travel agents for commission. It also excludes cancellations and interline revenue, where part of a particular flight booking makes use of the services of another carrier.³⁹

66. Flown revenue is therefore a more reliable gauge of market share. However although the Commission, cognizant of this distinction, prepared its market share figures based on flown revenue as opposed to BSP, its approach was still criticised by Viljoen. Indeed, so many difficulties did he foresee that he seems to view the exercise as futile.

67. Here is a sample of what he said on this point:

“So the difficulty we have as an industry is there is no easy comparison of market share… If we use capacity, which is what I used, that is the maximum that it can be if you fill every seat providing you split it between point of sale. If you use BSP it is highly flawed because it is sold and not flown. There are cancellations. There are other airline sectors in there. There are yield management effects. If you use ACSA, again it is just feet crossing the threshold into an aircraft and it does not take into account the point of sale. The only true measurement of market share is actual carried or flown passengers of each airline, which statistics we do not have for our competitors…(p 535)…So are no accurate figures. (p529). You need to get the flown, carried passengers of each of the airlines taking into account where the tickets were bought, if you want an accurate measure of the true domestic market shares. There is nothing else. And we don’t have it. So we guess”.

68. Let us deal with these objections of Viljoen’s, and others not recorded in this extract to see if they have any substance.

i)Tickets may have been sold in another country not South Africa.

69. Viljoen says a flight may have been reflected as flown revenue even when the ticket was sold by an overseas travel agent. He stated that depending on the time of the year, this figure could constitute as much as 20–40% of the number of passengers. No documentary evidence was offered to support this however.⁴⁰ Nevertheless the Commission responded to this criticism and through us, requested further documentation from SAA.⁴¹ It requested SAA to provide figures of flown revenue of SAA on domestic services sold in South Africa, broken down into the following categories: Interline, online, direct sales, travel agent sales, and finally, total sales including interline and total sales excluding interline.⁴² It therefore was able to compute SAA’s market shares by excluding sales outside South Africa, thereby representing only domestic sales of domestic tickets, and also to exclude interline data.

70. Secondly, the Commission obtained data from each of the airlines and combined this with data from ACSA to calculate the number of domestic passengers passing through Johannesburg Airport who purchased tickets in South Africa. ⁴³

71. The revised figures submitted by the Commission at the resumption of hearing in November 2004, which we find in the record in Figures Bundle 2, cure the defects listed by Viljoen in that firstly, they excluded domestic air tickets sold abroad. Secondly, they addressed the concern that the figures didn’t account for cancellations since, being flown revenue, these figures excluded refunds, cancelled tickets, lost tickets, i.e. any ticket that are included in BSP that’s not contained in flown revenue. Finally, the figures excluded interline sales.⁴⁴

72. When the hearing resumed three months later, the Commission had had a chance to present these figures to Viljoen in cross-examination. The Commission showed, that in fact, there was no substantial growth in the tickets sold outside South Africa, but in fact, there was a decline.⁴⁵ Though denied by Viljoen, this was not seriously contested by SAA.

73. Viljoen also appears to be have doubted that the SAA figures could be cleansed of foreign sales as, according to him, it did not keep records in this way. Counsel for the Commission explained that the order requesting further information from SAA had required a breakdown that made the distinction and that this information had then been supplied. It is worth noting here that when Viljoen resumed his evidence he was no longer an employee of SAA, having left a month earlier and had, it appears, no part in the compilation of the document. Be that as it may, this fails to explain the discrepancy between what he says SAA could supply and what they did. SAA itself did nothing to attempt to fill this lacuna.

74. We accept that the order was validly complied with and that the figures supplied purport to be what they say they are and that the Commission correctly relies on this information to calculate market shares.

ii) The figures do not take into account ticket swaps between airlines.

75. Viljoen testified that SAA and BA/Comair have an arrangement that each will honour the other’s tickets. A passenger thus with a ticket issued by BA/Comair and who then flies on an SAA flight, would be reflected on flown revenue as an SAA sale. The Commission has not been able to correct for this feature, but Viljoen is not able to explain why it is statistically significant given that this is on his evidence a reciprocal arrangement.

iii) SAA figures will always be higher because it has a better yield per ticket sold.

76. Viljoen says that because SAA has a better yield management system it gets a better yield per seat than do its rivals.⁴⁶ Accordingly its revenues will always be higher than the other two even if they have sold the same number of seats. Flown revenue does not correct for this. Again this criticism is made abstractly and no exercise has been performed by SAA to show how statistically significant it is. BA/Comair and Nationwide provided the Commission with their average ticket yields from July 1999 to June 2001.⁴⁷ SAA was asked to provide the same figures, but stated that it did not have them. This came as a surprise to Viljoen, now testifying as an erstwhile chief executive, who said that in his day the data was available and that perhaps the new people did not know how to access it.

77. This is a most unsatisfactory response. Again, as with the flown revenue figures, it does not redound to the credit of SAA that it cannot have resolved anomalies in respect of documentation between the erstwhile chief executive officer and those responsible for answering the Commission’s requests.

78. Nevertheless the Commission still performed an exercise to show what the figures were using passenger numbers only. The Commission obtained from ACSA figures that showed the number of domestic passengers passing through Johannesburg International Airport who purchased tickets in South Africa. This data yielded the following set of market shares---SAA 66,8% BA/Comair 22,4% and Nationwide 10,8%. When the figures included SAL and SAX, then broken down they were as follows: SAA 56,9%, SAL 5%, SAX 9,7%, BA Comair 19,1% and Nationwide 9,2%. Viljoen still questioned whether these figures had the foreign sales extracted and more generally whether the ACSA figures were reliable. The Commission had called a witness from ACSA who testified that these figures came from the airlines themselves.

79. T he Commission also argued that revenue, as a basis of calculating market share is an accepted practice worldwide. Furthermore, revenue is used by all airlines, SAA included, to calculated commission and incentive payments.

80. We would agree. Even though the Commission has done its best to correct for this alleged distortion by extracting passenger figures which reveal much the same figures, it is not clear, as a matter of antitrust economics, that Viljoen’s objections on this point have any substance. The fact that a firm’s revenues are higher per unit than those of its rivals, does not mean that sales revenue has to be adjusted to correct for this. It is precisely because it is able to attain these revenues, that is of interest.

81. In our view the Commission has done a more than thorough task in assembling and analysing the sales revenues and has correctly come to the conclusions on market shares that SAA has. Let us now consider the figures.

82. We set the various figures that the Commission has extracted in their various permutations in Tables 2 and 3 below.

Table 2: Domestic market share of SAA using various data sources (%)

Airlines	Flown revenue for domestic travel agency sales1,2	Domestic passenger numbers via Johannesburg that purchased tickets in South Africa3
SAA only	69	56.9
SAA and SAX	n/a	66.7
SAA, SAX and SAL	n/a	71.7

Notes: ¹ Calculation excludes SAX and SAL.

Sources: ² Table A.1, Figures Bundle 2, p1. ³ Figures bundle 2, Table C, p5.

Table 3: Market Shares of SAA based on revised estimates

Description of market	Data	Market Share (%)
Market share of SAA for travel agency sales	Flown Revenue	69
Market share of SAA (by value)	Flown Revenue	65.7
Market share of SAA by passenger numbers (with SAX and SAL included) (by volume)	Passenger Numbers	SAA 56.9 SAL 5 SAX 9,7
Market share of SAA by passenger numbers (with SAX and SAL figures excluded) (by volume)	Passenger Numbers	66.8

Source: Figures Bundle 2 pages 1, 5

83. What most of these figures suggest is a market share figure close to 66%. Even if one excludes the SAX and SAL and uses only passenger numbers, the most favourable figure in the tables for SAA, the market share is still 57%. Is it likely that there remains in this lowest possible figure, methodological gremlins that the Commission has not ironed out that affect SAA shares more disproportionately than its rivals that could bring this figure below 45%? We are satisfied that there are not.

84. We are further comforted by the fact that other proxies for market share (i.e. other than revised flown revenue and passenger figures) exist, which tell a similar story. We have the figure offered by Ms Harris, of Rennies Travel, SAA’s witness, that between them, SAA, SAL and SAX manage on a daily basis, approximately 70% of the domestic flights or departures in this country.⁴⁸

85. Then we have the consistency between these figures and internal travel agency estimates. For instance one travel agent in correspondence remarks in a letter to SAA that “the nearest other airline barely achiev[es] one third of SAA sales.” ⁴⁹

86. Not only has SAA not come up with figures of its own, its objections to the Commission’s figures have been speculative, trivial and ultimately, unhelpful. Never did SAA say, “ you have got it wrong by making this mistake this is what the figures are.” Rather it adopted the pose of a caviller without offering a version of its own, a task which could not have been difficult, given the small number of players in the market and its own market intelligence, being the largest airline. The Commission by contrast has done a thorough and diligent job in establishing the figures and has during the course of the hearing met all of SAA’s objections to the extent it has been possible to do so.

87. In our view the Commission has demonstrated that SAA’ s market share is well over 45%. Because we find that SAA is presumptively dominant we need not deal with a good deal of the evidence raised by SAA’s expert witnesses to the effect that it does not in fact have market power. This evidence is irrelevant, because once we find a firm’s market share exceeds the 45% threshold it is presumed to be dominant in terms of section 7(a) which states categorically that a firm is presumed dominant if it has 45% of the market. This is to be contrasted with section 7(b) where the presumption of market power is rebuttable. SAA is not just dominant but overwhelmingly so.

88. We therefore conclude that in the second market for domestic airline travel, SAA is dominant in terms of 7(a).

Conclusion on market shares in both markets

89. Although we have examined at some length whether SAA has market power in the domestic airline market it is not necessary for the Commission to show that this is the case. The Commission’s case depends on proving market power in the primary market where the abuse occurs and that is in the first market for travel agency services. Because the abuse occurs here, and its effects are only felt in the secondary market of airline travel, it is only really necessary for the Commission to establish that it has market power in this first market.

90. Nevertheless the Commission has, as we have discussed, done more than this and shown SAA is presumptively dominant in both markets. Given the relationship of the markets, it would be highly unlikely for SAA to dominate the market for travel agents services if it was not dominant in the adjacent market for domestic airline travel.

91. But the Commission has gone even one step further than relying on legal presumptions. It has shown that SAA enjoys market power in relation to travel agents. Recall this is the market power that is relevant in this case otherwise the abuse could not be perpetrated. The evidence in this case of the negotiations with travel agents and how SAA was able to impose a remuneration model on them against their will, is an example of this. In the words of Viljoen, who was never afraid of the blunt answer:

“MR VILJOEN: No, the fight was about the fact that I’m paying them less, because even if they make the targets, they are getting less, much less.

ADV PRETORIUS: But I’m saying they weren’t happy with what you offered them.

MR VILJOEN: Not at all.

ADV PRETORIUS: Why did they accept it?

MR VILJOEN: Why did they accept it?

ADV PRETORIUS: Ja

MR VILJOEN: Because I said I’m not paying more. That’s it.” ⁵⁰

92. Mortimer for his part stated that:

“ Our market power as travel agents is weak. We are not able to dictate our revenue model in the South African market. In South Africa suppliers determine the average level of commissions and we as agents look to our profitability form our ability to be successful in implementing the override agreements.” ⁵¹

93. SAA has not led any convincing evidence to rebut the notion that it has market power in relation to the travel agency market. Its experts queried if the market should not be analysed as a dual sided market. That is a market where two or more distinct sets of customers have an interdependent relationship with one another by virtue of a relationship with a common intermediary. For antitrust purposes, such a relationship, some recent literature suggests, may be important because the market power or alleged market power of the intermediary may be affected by its relationship with not just one set of customers, but perhaps two or more.⁵² Thus a credit card company may need to be looked at not only from its relationship to its cardholder customers, but also the stores which accept its cards. The point about this theory, as Holt has correctly noted, is that it relates to the market power of the intermediary between two sets of customers. In this case the travel agent is an intermediary between the supplier airlines and its customers. But the travel agents market power is not at issue in this case and therefore the theory, interesting as it maybe, has no application.

94. Now recall that a firm that enjoys a market share between 35% and 45% is presumed dominant unless it can show otherwise. This means that even if SAA’s market share in the travel agents market is below 45%, SAA has not rebutted the presumption of market power. If it was below 35%, a figure no one, not even the most optimistic SAA witness has seriously contended for, the Commission would have the onus of proving market power and that, on the present evidence, it certainly has done.

The Abuse

95. Having found that SAA is dominant in the two relevant markets we now consider whether the Commission has established that SAA has abused its dominant position.

Legal standard

96. The Commission, as noted, alleges SAA’s Explorer Scheme and the incentive scheme amount to a contravention of section 8(d)(i) and in the alternative, a contravention of section 8(c).

97. We set out below the whole of section 8 with the relevant sections underlined because the structure of the section becomes relevant to the interpretation of arguments to be considered.

Section 8

It is prohibited for a dominant firm to-

1. charge an excessive price to the detriment of consumers;

2. refuse to give a competitor access to an essential facility when it is economically feasible to do so;

3. engage in an exclusionary act, other than an act listed in paragraph (d), if the anti-competitive effect of that act outweighs its technological, efficiency or other pro-competitive gain; or

4. engage in any of the following exclusionary acts, unless the firm concerned can show technological, efficiency or other pro-competitive gains which outweigh the anti-competitive effects of its act –

   1. requiring or inducing a supplier or customer to not deal with a competitor;

   2. refusing to supply scarce goods to a competitor when supplying those goods is economically feasible;

   3. selling goods or services on condition that the buyer purchases separate goods or services unrelated to the object of a contract, or forcing a buyer to accept a condition unrelated to the object of a contract;

   4. selling goods or services below their marginal cost or average variable cost; or

   5. buying-up a scarce supply of intermediate goods or resources required by a competitor.

98. Both these sub-sections refer to ‘exclusionary acts ’ a term defined in the Act as:

“..[an] act that impedes or prevents a firm from entering into, or expanding within, a market;”⁵³

99. We have previously observed in York Timbers that 8(c) places a considerably greater burden on a complainant than 8(d). The first major difference between the two sub-sections is the treatment of the evidential burden in respect of the efficiency justification.⁵⁴ In terms of section 8(d), the onus of proof of the efficiency justification is on the respondent, whereas in section 8(c), the onus to negate it is on the complainant. Another difference is that in terms of 8(d) an administrative fine is competent for a first contravention. It is thus treated in the same way that the Act treats the other so-called per se contraventions.⁵⁵ In contrast, section 8(c) is treated in the same way as other ‘rule of reason’ contraventions for which a fine is not competent, unless the conduct is substantially a repeat by the same firm of conduct previously found by the Tribunal to be a prohibited practice.⁵⁶

100. There is a third difference. Paragraph (d) makes specific reference to five exclusionary acts, whereas paragraph (c) refers generally to an ‘exclusionary act’. It would follow that any exclusionary act not captured in the list in paragraph (d) would fall to be considered in terms of paragraph (c).

101. It would seem from the manner in which the section is drafted, that conduct in (d) is presumed to be exclusionary, whereas conduct not in the list would still have to be established as exclusionary. It is established under (c) only if it meets the definition of an exclusionary act.

102. The reason for these differences in treatment is that the exclusionary acts in 8(d) are listed, presumably evidencing the legislature’s view that these are the more egregious of the exclusionary acts and so firms who are dominant are on notice that they must behave with due caution in relation to this conduct, whereas in 8(c) no acts are listed and hence the complainant would have to prove that the conduct sought to be impugned is indeed exclusionary. Here the dominant firm has no advance notice that the conduct is deemed exclusionary in nature, and hence may be in some danger zone. This explains the policy choice to shift the onus in (c) to the complainant in respect of negating efficiencies, and secondly, not exposing firms to a fine for first time offences.

103. SAA asks us however to give a different reading to section 8(d). SAA argues that it is not enough to prove that the respondent did one of the acts listed in 8(d). The Commission must still prove that it is exclusionary i.e. that it impedes or prevents a firm from entering into, or expanding within, a market. Thus one would have to read into the text of 8(d) the revised words, so that it would now read: “engage in any of the following [exclusionary] acts, if they are exclusionary…”. SAA argues that the legislature did not intend to create a per se offence in 8(d), it merely meant to signal that if exclusionary conduct of the kind listed, is found to be the exclusionary conduct in question, the consequences for the respondent are the reverse in onus and the prospect of a first time fine. SAA argues that although its interpretation requires reading in ‘ if they are exclusionary’ it does account for the presence of the words ‘exclusionary acts’ in 8(d). If the listed conduct were per se exclusionary, there would be no need for the legislature to have referred to the words ‘exclusionary acts’, hence, on the SAA interpretation, we avoid redundancy.

104. SAA’s approach is not supported by the language of the section, which states, quite unambiguously, the “.. following exclusionary acts”. In section 8(c) where the Act does require proof that the conduct is exclusionary the language is different and refers to “an exclusionary act other than act listed in paragraph (d”). SAA’s approach is also inconsistent with the way in which the per se sections are treated elsewhere in the Act where the listing of specified conduct has created the offence. The whole point of the legislature’s approach is to treat certain conduct in this fashion so that its exclusionary nature does not have to be established each time. The interpretation of SAA would create some species of ‘middle ground’ contravention neither per se nor rule of reason⁵⁷. The creation of this unique species to support this reading seems wholly unwarranted. The legislature must be seen to be taking a view that either the conduct is exclusionary in the sense meant by the Act, or not. If not, why is 8(d) then worth the candle? If the point of the specified list is to warn firms in advance what conduct puts them into the danger zone, how is that purpose served if the conduct is only susceptible to ‘consideration as’, (SAA’s reading) rather than being ‘deemed’, exclusionary?

105. We find that as a matter of law if the Commission proves that SAA's conduct “requires or induces a customer not to deal with a supplier” then it has proved an exclusionary act. It is not necessary for the Commission to go on to prove that the conduct “impedes or prevents a firm from entering into, or expanding within, a market”. This is consistent with our finding in Patensie where we held that section 8(d) does not require the showing of an exclusionary effect:

“However, as already noted, in terms of Section 8(d) the complainant does not have to establish that the act complained of has an exclusionary effect, that is, that it prevents a firm from expanding in the market - if it is established that one of the acts specified in the various sub-clauses of Section 8(d) has been perpetrated and that the perpetrator is dominant, then the exclusionary nature of the act is presumed. We find that the Commission has discharged this onus.”⁵⁸

106. Perhaps a more fruitful area of debate is what consequences flow from the fact that an act is ‘exclusionary’? The question is then whether conduct deemed (8(d)) or found to be (8(c)) an exclusionary act , is, for that reason, impugnable (subject of course to the conclusion on the efficiency justification). That is not a simple answer. In York Timbers, an interim relief application, we stated that this was not sufficient:

“99. We conclude then that even if the facts had established a refusal to supply by SAFCOL, it would not have been possible to impugn this practice under the Competition Act. It is not enough to show that a given practice is a product of market power. It must also be shown that the act complained of actually extends that power or creates new sites of power. This has not been established and, accordingly, the application for interim relief in respect of the alleged violation of Section 8(d)(2) is denied.”  ⁵⁹

107. One approach is to say that if the act is exclusionary, it is deemed to have an anti-competitive effect. On this approach the only issue that remains to be decided is the balancing of the efficiency justification against this deemed anti-competitive effect.

108. The problem with this approach is that it can lead to the outlawing of conduct that has no anti-competitive effect. The definition of an exclusionary act is very broad indeed. Discussing not dissimilar language, Areeda and Hovenkamp, in their treatise have this to say:

“In defining undesirable conduct, we are concerned mainly with exclusionary behaviour, that which prevents actual or potential rivals from competing or impairs their opportunities to do so effectively. But this term and the root idea are also too broad, for they embrace all competitive behaviour: All successful competitive moves tend to exclude, particularly in oligopoly markets.⁶⁰

109. The same observation by the authors can be made in respect of our Act’s definition of an exclusionary act. The term is not a useful filter for determining whether conduct is competitive or anti-competitive; both can sensibly be included in the definition. If, however, we do not regard the notion of anti-competitive effect, referred to in both paragraphs (c) and (d), as inferentially linked to an exclusionary act, this danger can sensibly be averted. It means that that the notion of what it anti-competitive is something different to an exclusionary act.

110. At a purely textual level they appear to be notionally different. If they were congruent notions, then the legislature need not have troubled itself with introducing the language anti-competitive effect into the paragraphs, but would instead have referred to exclusionary effects. We suggest that there is a difference between an exclusionary act as defined and the inference that it has an anti-competitive effect. Without some notion of what the anti-competitive effect is it would be impossible to perform the weighing with the efficiency justification that both (c) and (d) require. In order to perform a weighing of the anti-competitive effect on the one side of the scale to the efficiency gain we need to have some notion of its quantitative effect. But the definition of an exclusionary act is descriptive of a conduct’s ‘type’, not its ‘gravity or extent’. Thus by way of example the refusal to supply one customer with a de minimus market share and the refusal to supply a substantial number of customers, representing a large proportion of the rest of the market are both exclusionary acts in terms of d(ii), but they have very different competitive consequences.

111. For this reason the Act requires a finding both in terms of paragraphs (c) and (d) that the complainant not only establishes that there has been an exclusionary act, but also that it has an anti-competitive effect.

112. We have to answer two questions if we follow this approach. Firstly what do we mean by anti-competitive effect? Secondly, if it means something different to the notion of an exclusionary act, what purpose does reference to the latter term serve in section 8? Why does reference to an anti-competitive effect not suffice? If it is after all anti-competitive, does it matter if we call it exclusionary first?

113. To answer this we must appreciate the purpose of the abuse of dominance prohibition in our Act.

114. Modern antitrust law identifies two species of abuse of dominance.⁶¹ The one kind, termed exploitative, focuses on the effect of the abuse on the consumer, who, in consequence of the output decisions of the dominant firm, may be facing output constraining behaviour and hence higher prices. Section 8(a), which prohibits a dominant firm from charging an excessive price to consumers, is an example of this kind.

115. The other kind is an abuse that has an exclusionary effect, and this is what concerns us in terms of sections 8(c) and (d). It is exclusionary in the sense that it is conduct which excludes or impedes the growth of rivals in the market. The theory of harm that explains why exclusionary behaviour may be anti-competitive is usually explained in this way. An act that is exclusionary of rivals can lead to the creation or enhancement of a dominant firm’s ability to exercise market power. By its exercise of market power it can adversely effect consumer welfare by output limiting decisions. As we see this consists of a set of stages of assumptions before we lead to a conclusion of welfare loss. Should an abuse of dominance provision that seeks to proscribe exclusionary behaviour require there to be evidence of each part of a chain of causation establishing the links from the act of exclusion to the loss of consumer welfare? Areeda and Hovenkamp caution against such an approach:

“However because monopoly will almost certainly be grounded in part in factors other than a particular exclusionary act, no government seriously concerned about the evil of monopoly would condition its intervention solely on a clear and genuine chain of causation from exclusionary act to the presence of monopoly”⁶²

116. It is comforting to note that even in United States law this is still a matter of controversy. Two positions have emerged in the case law. Sometimes this eclecticism manifests itself in the approach taken in the same decision. Fox notes in her gloss on the Court of Appeals decision in Microsoft that:

“ While it may be sui generis in many respects, it is typical in its

ambivalence regarding seriously exclusionary practices that may

not have output effects.”⁶³

117. And later Fox asks rhetorically:

“What did the court mean by “anticompetitive” and did it use that word

consistently.”⁶⁴

118. The first approach in the case law favoured by some writers notably those close to a Chicago School, requires a showing of harm to consumer welfare in order to make conduct unlawful. Absent such a showing these writers contend there is a danger that courts will “mistake protecting competitor profits for consumer welfare”.⁶⁵ These writers find authority for their position in decisions of the Supreme Court such as Brooke Group a case dealing the standard of proof required in predatory pricing claim. The Court noted:

“Mistaken inferences are especially costly, because they chill the

very conduct the antitrust laws are designed to protect.” ⁶⁶

119. Muris, who is representative of the views of this school of thought, argues that:

“Both the history of the Supreme Court cases, as well as an analysis

of the weak empirical foundation of much modern economic theory,

suggest that so-called exclusionary conduct can be condemned as

monopolistic only after a full analysis, including consideration of

whether the practice in fact has an anticompetitive effect.” ⁶⁷

120. For Muris anticompetitive effect is the ”ability to raise price and restrict output.” ⁶⁸

121. The second approach is to find liability if there is evidence that the exclusionary behaviour will lead to substantial market foreclosure. Writers who support this approach are concerned that if harm to consumer welfare is the standard, antitrust law will be under-deterrent, because evidence of harm to consumer welfare is very difficult to come by. Support for this approach is found in earlier Supreme Court decisions, such as Lorain Journal and Otter Tail Power Co. In Lorain Journal the Court held that it is:

“not necessary to show that success rewarded appellants attempt to

monopolise”⁶⁹

122. In other words what the court is saying is that proof of anti-competitive effect, in the sense that Muris understands it, is not necessary.

123. The second approach is best articulated in Fishman v Wertz a decision of the Seventh Circuit. The Court held that the antitrust laws protect competition and the competition process, not results. It was no defence that consumers were not hurt. Thus:

“[Defendants] assert that “ the antitrust laws do not apply where

there is no consumer interest to protect.” … The dissent makes the

same argument about consumer effect. “ Antitrust law condemns

results harmful to consumers….” “We agree that the enhancement

of consumer welfare is an important policy – probably the paramount

policy – informing the antitrust laws… Some Supreme Court cases

indicate that effect on ultimate consumers is, in an appropriate

context, a significant consideration in analysing a business practice to

see whether there has been an antitrust violation…The antitrust laws

are concerned with the competitive process, and their application does

not depend in each particular case upon the ultimate demonstrable

consumer effect. A healthy and unimpaired competitive process is

presumed to be in the consumer interest.”⁷⁰ (Our underlining)

124. Areeda and Hovenkamp in their treatise have suggested a test that has influenced later decisions. They suggest that:

“In sum, ‘exclusionary’ behaviour should be taken to mean conduct

other than competition on the merits or other than restraints

reasonably necessary to competition on the merits, that reasonably

appear capable of making a significant contribution to creating or

maintaining market power.”⁷¹ (Our underlining)

125. In Microsoft we hear echoes of this approach when the Court of Appeals commenting on one aspect of Microsoft’s behaviour in excluding rivals from the boot-up sequence said:

“ Because this prohibition has a substantial effect on Microsoft’s

market power, and does so through a means other than competition

on the merits, it is anticompetitive.” ⁷²

126. In Europe, Fox argues, the approach to exclusion cases is “often phrased as a dynamic one: the right of market actors to enjoy access to the market on the merits.” ⁷³ She goes on to argue that in European Court of Justice cases it is clear that:

“harm to competition is a wider concept than the result –oriented

output limitation. Use of dominant power to procure significant

advantages not on competitive merits, thereby pre-empting

competitors’ opportunities, is a harm to competition under Article 82.” ⁷⁴

127. What is happening in this debate in the case law and academic writing takes us back to the chain of causation referred to by Areeda.

128. What are courts doing when they find behaviour to be anti-competitive in the absence of evidence of harm to the consumer? Essentially they are consciously, or sometimes unconsciously as Fox suggests, making inferences of fact and law and sometimes, mixed fact and law, to arrive at findings of competitive harm by way of proxy. Courts may find that, as a matter of fact, a particular business practice is exclusionary. As a matter of fact, a court may also find that the practice has the potential to foreclose the market for competitors of the dominant firm. As a matter of inference on the facts the court may find that this is likely to foreclose competition. As a matter of inference the court may determine that if competition is foreclosed, there will be an adverse impact on competition.⁷⁵ This latter type of inference, as Fox has pointed out, is not factual, but legal. It is based on an assumption that :

“.. markets are more likely to reward merit if they are not clogged by substantial unjustified exclusions. Moreover, although we may not know the direction in which ‘open’ competition will take us, we may prefer to let the chances of competition –rather than the strategies of the dominant firm- take us there.” ⁷⁶

129. What this excursion into the case law and commentary suggests is that there is respectable authority for the notion that exclusionary practices should not require evidence of actual competitive harm for a finding of abuse. A finding is still possible if there is evidence that the exclusionary practice is substantial or significant, or expressed differently, has the potential to foreclose the market to competition. If it is substantial or significant it may be inferred that it creates, enhances or preserves the market power of the dominant firm. If it does the latter it will be assumed to have an anticompetitive effect.

130. It is however necessary not to limit that competitive harm to the notion of consumer harm because of the danger, as the writers we have quoted have suggested, that the Act would then have introduced an elaborate scheme to regulate exclusionary abuses of dominance, with deemed exclusionary conduct and reverse onuses, only to find that it is under-deterrent. If on the other hand we inferred harm to competition simply by the finding that conduct was exclusionary, that would risk the danger of outlawing aggressively competitive, but non-predatory behaviour and so would be overly deterrent. ⁷⁷

131. Authority for the fact that harm to structure suffices to show an infringement of the Act, rather than requiring direct evidence of harm to the consumer welfare, is evident in the approach of the Competition Appeal Court. In the Patensie case where the court was dealing with the implications of terms in a firm’s articles of association that restrained its members from contracting for the services of its rivals, Selikowitz JA held:

“There can be no doubt that the restraint provisions, with or without the reinforcement by the penalty provisions in Appellant’s Articles of Association enable Appellant to prevent all its members from offering their produce to other packing houses indefinitely and for so long as they are members of the Respondent. Conversely, the members are deprived of the opportunity to select a competitor to pack and market their citrus fruit. Furthermore by tying more than 70 per cent of the farmers to the exclusivity agreement, other potential competitors who may wish to compete in the relevant market are effectively excluded. The effect of the offending Articles of Association is to hinder the maintenance of the degree of competition which exists and to hinder the growth of competition.

The Tribunal’s finding that Appellant’s “conduct that is complained of is in clear violation of section 8(d)(i)” is, in my view unassailable.” ⁷⁸

132. In summary, we find that the Act sets out the following approach to exclusionary practices. In the first place we examine whether the conduct in question is exclusionary in nature. In terms of section 8(c) that would be conduct that fits the definition in the Act for what constitutes an exclusionary act. In terms of 8(d) it is conduct that meets the definitions set out in the sub-paragraphs of that section. If the conduct meets the requirements of the definition, we then enquire whether the exclusionary act has an anti-competitive effect. This question will be answered in the affirmative if there is (i) evidence of actual harm to consumer welfare or (ii) if the exclusionary act is substantial or significant in terms of its effect in foreclosing the market to rivals. This latter conclusion is partly factual and partly based on reasonable inferences drawn from proven facts. If the answer to that question is yes, we conclude that the conduct will have an anti-competitive effect. Whichever species of anti-competitive effect we have, consumer welfare or likely foreclosure, we have evidence of a quantitative nature and hence we can return to the scales with a concept capable of being measured against the alleged efficiency gain.

133. Thus far the onus of proof in terms of both sections is on the complainant. Here the treatment of the onus in the two sections now diverges.

134. In terms of 8(c) we then consider whether the anti-competitive effect outweighs any efficiency justification for the conduct. If it does we can find that there has been an abuse of dominance. Here again the onus is on the complainant.

135. In terms of section 8(d) the burden of proof now shifts to the respondent who must prove that the efficiency justification outweighs the anticompetitive effect. If the respondent does not, then the conduct will be found to be an abuse.

136. It is now appropriate to answer our prior questions. An anti-competitive effect is something different to an exclusionary act. This does not make the reference to an exclusionary act somehow superfluous. It firstly signals that we are analysing an exclusionary as opposed to an exploitative abuse. Because we know we are dealing with an exclusionary as opposed to an exploitative abuse, it helps guide our analysis of the alleged anti-competitive effects of the conduct. More importantly, because some forms of exclusionary act are for the legislature more commonly associated with egregious behaviour by dominant firms these are signalled out for special mention, so that dominant firms are on their guard to be especially careful when embarking on this form of market behaviour. Finally, we would suggest that the use of the word has a “characterising “ function. it signals the legislature’s intention to view competitive harm as structural in nature as opposed to a test of abuse of dominance that is based solely on consumer harm.

137. A cautionary word needs to be said about the use of terms such as exclusionary and anti-competitive. These are labels which are not always used uniformly by writers or, as Fox indicates, by courts. Reliance on foreign authority should be premised on a clear understanding of how the terms are being used. The danger of uninformed borrowing of comparative jurisprudence that the Competition Appeal Court has frequently warned against is no more apparent than here. One person’s notion of what constitutes an exclusionary act may be more extensive than what is meant by the same term in our Act. (Unlike in our law the term is not defined in U.S. law by statute but by judicial interpretation, which has led to inconsistency.) One person’s anticompetitive harm can mean harm to consumer welfare only, whilst another’s embraces harm to the structure of markets as well. We as the Tribunal may likewise have been guilty of not making ourselves clear on this point in the past. In York Timbers we held that establishing the existence of an exclusionary act was insufficient to prove liability, absent a showing that it enhances monopoly power. That approach is followed in this decision albeit with a more detailed analysis of why anti-competitive harm includes harm to the structure of a market. In Sappi Fine Papers, however, we held that an exclusionary act was presumptively anti-competitive. This holding, made obiter, is not as the reader will see, the position that we subscribe to now with our further experience of working with the Act.

The abuse – factual issue.

Proof of an exclusionary act

138. We have now defined the legal test for section 8(c) and 8(d). The Commission must prove an exclusionary act has taken place and secondly that it has an anticompetitive effect. We must now examine whether the Commission has met this standard. Since a showing under 8(d)(i) will not require us to examine the conduct in terms of 8(c), the alternative count, we start by applying the test of proof of an exclusionary act to the facts in terms of 8(d)(i).

Does the conduct require or induce suppliers [ travel agents] of the dominant firm [SAA] not to deal with its competitors [Nationwide, Comair]?

139. A crucial part of the SAA defence is that override agreements are not unique to SAA nor are they of recent invention as they were around long before the relevant period. Factually this is true. SAA has had incentive schemes for travel agents for some years. No one is sure when they started but they appear to have been in operation during the 1980’s i.e. well prior to the relevant period.⁷⁹ Nor were they the brainchild of SAA. They had been in vogue overseas for some time. By January 1999, SAA had override agreements with all the large travel agents.⁸⁰ Both Nationwide and Comair also had override agreements.⁸¹ It is also true that from a legal perspective the SAA override agreements in issue in this case are identical to their predecessors in operation prior to the relevant period.⁸² There is also evidence that for the period of April 1998 up until July 2000 Nationwide experienced a continuous increase in its passenger numbers. There is also no evidence that Comair had any concerns about the effect of the SAA override scheme prior to the relevant period.

140. SAA relied on this to assert that the scheme could not have affected the behaviour of travel agents so as to induce them not to deal with SAA’ s rivals. Thus, it argues that its rivals’ complaints of harm during the relevant period cannot be related to any anti-competitive effect of the scheme, but the pro-competitive behaviour of SAA in the market place.

141. What this ignores is that it is not the existence of the scheme that is in issue but its nature, and its nature, on SAA’s own evidence, changed significantly in late 1999. Thus while legally the scheme may have looked similar to its predecessors, economically it was a different creature. On SAA's own version it had become more ‘challenging’ for travel agents.⁸³

142. Let us examine why this is so. We noted earlier that the override agreements provide three forms of compensation by way of commission: the basic, the override and the increment. Prior to 1999 the basic commission was set at 9%; this appears to have been the industry standard for a very long time. Sometime in 1999 British Airways changed its international commission from 9% to 7%. SAA, in October 1999, followed suit. In terms of the new agreements the drop in the basic commission was accompanied by an increase in the commission payable in terms of the override and the increment, but only if the agent reached a more demanding target. Its effect on agents is described by one of the witnesses called by the Commission, Mr Willem Puk, of SureTravel.

“ But then.. used the philosophy that we don’t actually want to save the 2%, we actually want to give it back to you and in fact will give you back more. We will give you the 8%, 9% and 10%,providing you can move market share towards us and that was there [their] philosophy. It is a scheme they adopted in the UK just prior to introducing it in South Africa and I think from that step, things snowballed very quickly in South Africa.” ⁸⁴(Our underlining).

143. The snowballing that Mr Puk refers to presumably came about as a result of SAA’s adoption of BA’s more aggressive approach to the implementation of the override scheme. SAA did so in October 1999.⁸⁵ SAA is very explicit about its reasons for changing the nature of the scheme. As the CEO at the relevant time, Andre Viljoen, testified:

“..we then went about redeveloping those override structures to.. not reward or pay them for what we call generic growth in revenue, because inflation pushes the price up anyhow and to make it challenging for them to earn a reward for distributing our tickets.” ⁸⁶ (Our emphasis)

144. SAA soon redesigned the agreements to reflect this change in policy. In the agreement with Luxavia, one of the largest groups, the intention of the scheme is expressed clearly in clause 2.1.3:

“SAA wishes to incentivise and reward the agent by paying a cash override incentive for achieving certain total SAA flown revenue targets, all on the terms set out in this agreement. Accordingly this override agreement is established to stimulate and encourage the increase in total SAA flown revenue in return for the cash override incentive payments.”⁸⁷

145. The agreement goes on to record that the agent understands that the standard (basic) commission is subject to reduction from time to time in line with worldwide airline industry trends.

146. With the change in the compensation system shifting the emphasis from basic commission, now lower, to the override and incremental incentive commission, now higher, and with targets becoming more ‘challenging’, what is it that agents were being incentivised to do? The answer is to direct more business to SAA if they were to retain their profit margins, let alone improve upon them. But it does not necessarily follow that if this was how the agreement incentivised them that its effect was to induce agents to sell SAA tickets at the expense of those of its rivals, which is the conduct required to make the act exclusionary.

147. In order for the agreement to have this effect the Commission has to show-

- that agents have a financial incentive to move customers who purchase tickets from them away from rivals and towards SAA; and

- that agents have the ability to do so.

148. The Commission argues that it has shown both. To prove the first proposition the Commission’s experts, Oxera, devised an economic model to show what travel agents’ incentives are in terms of the override scheme.

149. The model itself has not been the subject of any serious criticism by SAA and hence need not be examined in any detail. Rather SAA contends that it is so theoretical, that it is of little value.

150. Oxera developed a model of the incentives facing travel agents based on a selection of the override agreements. The model assumes a similar agreement exists between a competitor of SAA that is one –third of its size. We are told by Mr Holt, the expert from Oxera who testified, that the model holds true even if the competitor is two-thirds SAA’s size. Now the model assumes that travel agents have control over which airline their customers wish to travel on. Oxera concedes that this exaggerates the influence that travel agents have. (Note that the Commission’s case is not that agents always influence consumer choice, but they can often enough for it to be significant.) Nevertheless, the importance of the model is to demonstrate what the incentives of agents are, it is not meant as evidence that agents have this control. The Commission argues that the incentives in practice are even more stark for travel agents than the model shows, since in practice agents are dealing with more than one rival to SAA and those rivals have smaller market shares than the rival in the model. (The rival in the model was assumed to have 25% of the market to SAA’s 75%). This means that the additional commission generated by each individual competitor will be lower.⁸⁸

151. The model demonstrates that travel agents maximize their commission by allocating 100% of their customers to SAA and none to a rival airline. Oxera then use the model to examine how the travel agent could still maximise their commission if they shifted business to a rival of SAA. Using the American Express agreement as an example, Oxera shows that if they were to cut SAA’s market share by 5%, they would need to need to receive an average commission rate of 14,4% from a rival airline, which would so raise the rival’s costs as not to be viable. Secondly, Oxera shows that reducing SAA’s growth by 5% would mean a 32% growth by a competitor, which it argues, while not implausible, is generally the kind of growth rate associated with cut-rate, no- frills airlines who do not use agents for the distribution of their tickets.⁸⁹

152. The reason why the model shows these incentives is because of the marginal incentives of agents. Oxera explains that if all tickets were sold for a flat rate of 10% with no override scheme in place, the marginal incentive rate to travel agents would be 10%. With the override scheme, the incentive of the travel agent is greatly changed. Assume an agent needs to sell 100 tickets to get to the base target. Until the target is reached the agent gets only the basic commission. However above the base target the override and incentive commission’s change the compensation to the agent in two ways. Firstly, the commission increases above the basic incrementally for further amounts of tickets sold. Secondly, and perhaps most importantly, is that once the agent is above the base target the override also kicks in, which means that the override rate is applied not just to the extra ticket sold above the target, but to the whole value of tickets sold including those sold to reach the target. This is called the ‘back to rand one’ principle. According to Oxera:

“As a result, in some cases the marginal commission rates within the contracts can be well above 30% of additional sales for SAA.”⁹⁰

153. Oxera prepared a table illustrating the change in these marginal commission rates, which we have included in this decision for illustrative purposes as appendix 1. Oxera argues that the table shows that the function is often ‘discontinuous’ by which they mean that there are sharp increases in the marginal commission rate paid. This demonstrates incentives not only to achieve sales above the base target, but also at later targets down the line. In appendix 1, we see that in the case of American Express when it achieves sales in excess of 15 % of its base target( what Oxera have termed “out performance”), its marginal commission rate moves from 7,5% to 217,5% .

154. The best illustration of how dramatic this effect is comes not from Oxera, but one of the agents whose correspondence with SAA the Commission introduced into the record. The following extract is from an e-mail from Bradley Jay, then MD of Seekers Travel, to SAA CEO Andre Viljoen:

“As you are aware our domestic percentage is linked to the achievement of certain levels of international turnover, at this stage we are forecasting to be between 1.9 million to 600 000 short. 1.9 million being very conservative, if we hit the target, this is by selling at most R1.9 million worth of tickets, we stand to earn R2.7 million in tickets, in incentive. It obviously means that in the worst case scenario we can spend the 1.9 million including giving away free tickets, if necessary, in order to achieve a net revenue to Seekers of 800 000”.⁹¹

155. In other words, even if they gave away at that point in time R1.9 million worth of tickets for nothing, they would still have made R800 000 clear profit.

156. Thus at least one of the travel agents, without the benefit of Oxera’ model of marginal commission rates, has come to the same conclusion. At a certain level of sales, the benefits of selling another SAA ticket are overwhelming. Without achieving similar compensation from a rival, the agent has little incentive to sell the rival’s ticket and a compelling incentive to sell SAA’s.

157. That agents had a keen appreciation of this fact is borne out not only by this letter but also one from Mr Puk to agents within his group in which he berates them for not achieving sufficient sales through SAA.⁹²

158. Holt has also made certain other observations about the nature of the scheme by reference to some of the economic literature which he says tends to suggest its anticompetitive nature. Firstly, he suggests that the override and incentive commissions cannot be regarded as targeted discounts, the latter sometimes being treated as pro-competitive in the literature. He states in this case we have not a targeted ‘discount scheme,’ which is a payment made to the consumer, but a targeted ‘premium scheme’ which is where the payment is made to an intermediary, in this case, the travel agent. He says that a targeted premium scheme (i.e. the override incentive scheme) tends to be more anticompetitive than a targeted discount scheme because the benefits are not passed on to the final consumer.

159. Next he identifies the fact that the pricing structure is non-linear. He explains it thus:

“Essentially what this means is that there is a break in the relationship between the amount of the commission paid and the amount of sales earned by the travel agent for that particular airline; that there is effectively a jump or a bonus level, which is paid once you meet a particular threshold. Now it is clear from the general economic literature ..that the greater the degree of non-linearity in the scheme, the greater the anticipated anticompetitive effects.”⁹³

160. Holt identifies the discontinuities or leaps in the marginal commission rate that we referred to earlier as examples of the high degree to which these agreements are non-linear.

161. He also makes an observation in relation to the length of the reference periods, by which he means the duration of the contracts. The longer they are, the greater the base on which bonuses are paid, and the less possible it is for competitors to sign up such agreements with travel agents. He considered a period of one year, the typical duration of the agreements in question, as being sufficiently long for them to have an anticompetitive effect.⁹⁴

162. Another aspect of concern with the nature of the agreements is their lack of transparency. Because, as we have seen, compensation is based on flown revenue, a lesser figure than BSP, the figure that travel agents have in their books, travel agents do not know at any given time between reporting back from SAA (typically every quarter) whether they are making their targets or not. This says Holt will make them more conservative in their approach.⁹⁵ He goes on to say that as a result they:

“will tend to want to focus traffic even more on SAA in order to be absolutely sure or as sure as possible of getting these bonuses by directing as much traffic as possible to SAA.”⁹⁶

163. This aspect of his evidence is supported by one of the travel agents who testified, Mr Mortimer, who stated:

“Flown revenue… is exclusively held by the airline. So we don’t exactly know where we are in relation to the flown revenue at any particular point in time. We are benchmarking. We are getting quarterly sales and performance reports and review meetings are taking place with the preferred partner where the airline presents our performance results, lets just say at the end of the quarter. We are looking at our invoices. We are applying an adjustment factor to ballpark what we think the flown revenue is and as we start getting towards the end of an agreement and lets just look at the dates of this e-mail. This e-mail is the 27^th February. We are approaching 31 March. .. It’s almost the end of the financial year. We are going to just make it or just miss it. The MD is concerned that he may have just missed it, and I can remember it because there was the possibility for me of an around the world ticket and I was disappointed.” ⁹⁷

164. If the management of a travel agency operate under this cloud of uncertainty for much of the time, all the more so the individual consultants who interact with the consumer on a daily basis and who know what pressures their agencies are under to meet target.

165. Although Holt was the subject of much criticism from SAA, in our view unwarranted, this criticism apart from some carping technical complaint from SAA’s expert, which Holt adequately answers, did not extend to the way in which his model works.

166. Rather the criticism is that the model is purely theoretical and dependent on assumptions that are not valid, such as the travel agents’ ability to move business to SAA. We deal with this issue below. He is also criticised because he fails to take into account that the override agreements were in effect since the 1980’s. On this point, the Commission correctly counters when it points out that it is not the fact of the agreements, but the way their nature changed in 1999, that is crucial. He is also criticized for assuming that the other airlines did not have override agreements at the time, which they did. Whilst Holt had been incorrectly instructed on this point it does not affect his model. His model assumes a “similar agreement” between the rival airline and travel agents.⁹⁸ Furthermore, as the rivals are not dominant firms, their schemes whilst similar to SAA’s, are always going to be ineffectual – they simply do not have the market share to change the incentives of travel agents unless they drastically increased the compensation to agents. Holt argues that this would have to be to a level that is unaffordable to them.

167. The other criticism of the model is that if it were true, then rivals would stop making use of travel agents and find some other avenue to sell tickets. This is best answered, not by Holt, but by Viljoen, who, in explaining the centrality of travel agents, has pointed out the increased cost of ticket distribution if an alternative was considered. If it is unattractive for SAA, with its volumes, to forsake travel agents as the optimal sellers of tickets, all the more so for its smaller rivals.⁹⁹

168. We now turn to consider the most contentious factual issue in the case and that is whether travel agents can influence customers’ choice of airline. If this is not the case, then SAA is correct and Holt’s model, for which this is a key assumption, must fail, and hence an essential pillar of the Commission's case on exclusion.

169. At first blush the SAA critique seems the more plausible. SAA argues that if travel agents steer customers away from their preferred airline to a more expensive one, then they risk being caught out and the consequences of being exposed in this way could be irreparably damaging to their businesses. SAA says that Holt’s model does not take this into account. However the Commission has not based its case on this aspect on Holt’s evidence. Holt does not attempt to say that this is the case – he says if agents could, the model has the consequences he outlines.

170. The Commission relies instead on other sources. Firstly, what agents have said themselves about their ability to do so and secondly, on past statements by SAA, where agents’ ability to move customers appears to be assumed.

171. Mr Bricknell, the chief executive of Nationwide, testified that in mid 2000 he had noticed a drop in their sales and he asked his staff to investigate the problem. The feedback he got was that travel agents:

“..were so highly incentivised by South African airways. They also had this program called Explorer that the consultants were getting incentivised as well, and that we just couldn’t compete. So we were being sold off.” ¹⁰⁰

172. As a result of these reports he went to see Ms Lillian Boyle the chief executive officer of Rennies, one of the major travel agency groups and asked why they were not getting support. Boyle he said, replied that she was glad to hear that her instruction had been adhered to:

“ because SAA was their preferred carrier and she had instructed her entire group to only deal with South African Airways and not to sell Nationwide.” ¹⁰¹

173. Bricknell was not contradicted on this point. Although SAA called Ms Boyle’s colleague Ms Harris, she testified that she was not present at the particular meeting. Boyle could have been called to testify by SAA on what is an important meeting, but was not. The Commission attempted to call Ms Boyle, but despite a subpoena she did not attend, and we received a medical certificate from her doctor to indicate that she was indisposed for that entire week – the week when the Commission was calling its witnesses. Some months later, when SAA led its case, she was still not called, even though one presumes from the relatively minor illness indicated in the medical certificate, she would have recovered. Bricknell’s evidence on this meeting is thus uncontradicted.

174. The probabilities that Bricknell's evidence on this point is true, and can be accepted, is bolstered by correspondence between Ms Harris of Rennies and David James of SAA. In this letter dated 28 October 1999, Harris expresses an undisguised outrage at the new incentive scheme, which had obviously just been communicated to Rennies.¹⁰² What it clear from what Ms Harris writes is that she believes that Rennies can influence the movement of customers.

“For a number of years we have supported SAA through preferred supply policies – to the extent that we have actively either moved to or retained business on your carrier.”

175. And at the end of the same letter where she threatens:

“Given your one-sided attitude to the relationship we will now move to evaluate, in addition to our passenger volumes, our group cargo volumes handled by your carrier, namely from Renfreight and Safcor. Sadly, should your position not change and should we continue to find ourselves compromised by your stance, we will be forced to move our support to your competitors – perhaps South African Airways management does not appreciate that we add real value to the relationship. If that is the case then I fear will we need to demonstrate this capability to you which will unfortunately be to the mutual disadvantage of both of our organisations.”

176. Harris as noted, was called by SAA to testify. Cross-examined on the contents of this letter her answers were not satisfactory. She does not appear to deny the ability of the agent to move customers, but suggests that they would only do so with the customer’s best interests in mind. So on the credibility of the threat contained in the letter she testified:

” I said if you don’t retract what you have put on the table, which is not a viable model, then we will have no alternative but to move, obviously on the predetermined (sic) that it is right for the customer, away to another competitor.” ¹⁰³

177. Given that the letter was written in 1999, prior to the contemplation of these proceedings, we place greater reliance on it than the subsequent oral testimony of Ms Harris, given at a time when Rennies had an incentive not to be seen capable of influencing customer choice. Rennies also seems the one agency well disposed to assisting the SAA case.

178. But Harris is not the only travel agent whose correspondence indicates a belief that they have the ability to move sales from one airline to another. In a letter to SAA dated 25 February 2000, on the subject of deliberations over the new override agreement, Nigel Adams, the managing director of American Express Travel writes:

“Clearly Amex could offset its risk by drastically reducing its flown revenue with SAA, and directing a large portion of this R 170 000 000-00 worth of flown revenue to other Airline carriers where its current base of business is low, and attractive override deals are presently on offer.”¹⁰⁴

179. Note that prior to this, on 9 November, Adams, who wrote the above letter, wrote to his superior, Craig Bond at Tourvest, where he notes that SAA has told him commission is going to be cut from 9% to 7% on 1 November 1999.

180. He then goes on to state:

“I understand that Tourvest and ourselves are committed partners of the National Carrier, but at the end of this financial year, as an owner and a manager, I am on line for the profitability of this business. I have agreements with both British Airways and Comair where I can earn 10% and 5% back to Rand One respectively and to be honest the temptation of my line management is to really start pushing these partners even harder.” ¹⁰⁵ ( Our emphasis)

181. When Viljoen was examined on this letter, and on others from agents evidencing their ability to move passengers, his response was that:

“those letters are nothing more than posturing.” ¹⁰⁶

182. Whilst all commercial bargaining may have an element of what Viljoen refers to as “posturing”, it is unlikely that travel agents would make a threat that had no credibility whatsoever. Even if Ms Harris was exaggerating Rennies’ ability to move passengers, the fact that they have some influence in this respect is unlikely to be mere posturing. If travel agents had no significant influence in customer choice, SAA would be well aware of it and the bluff would be easily called. What SAA perhaps did not know was the extent of a particular agent’s influence and, in that respect, Viljoen may have considered this an exaggerated claim of influence.

183. Furthermore, if this is just posturing by travel agents in their bargaining with SAA, this does not explain why Adams, in an internal letter to the CEO of his parent company (the one quoted above), would refer to his agency’s ability to push other airlines if both he and Bond did not believe that they had some capacity to do so. This is not posturing for the benefit of SAA, as it is not a letter they would have seen.

184. But we also know that SAA itself considered that agents had this ability. This emerges from an exchange of letters between American Express and SAA in which the terms of a proposed override agreement are being negotiated. In a letter dated 5 November 2001, Nigel Adams of American Express writes:

“American Express will commit to a 25% year on year growth for the period November to March for both domestic and international.”

185. And later on in the same letter Adams adds:

“American Express will be forced to disregard other airline agreements entered into to achieve 25% growth in these difficult times.”¹⁰⁷

186. Fay Mhlanga from South African Airways writes a letter to Adams responding to the suggestion of achieving a 25% growth:

“Achieving 20% domestic growth and 25% international growth for the full year will be difficult without actively shifting business to South African Airways” ¹⁰⁸ (Our emphasis)

187. Thus Mhlanga, like Adams, recognises the ability of travel agents to influence and to shift business between carriers.¹⁰⁹

188. The Commission also relies on both the correspondence and oral testimony of other travel agents. The first travel agent called to testify was Mr Willem Puk, the Managing Director of the Sure Travel Group. Sure operates as a franchise of independently owned agencies.

189. Prior to this complaint, Puk, like Harris, was less reticent about expressing his views on travel agents’ potential to move passengers in correspondence, than he was in his testimony before the Tribunal. A monthly e-mail bulletin from Puk to Sure travel agency members dated 11 December 2001 is most instructive. Some select quotes, evidencing the Group’s attitude to SAA and Nationwide are¹¹⁰:

“saa’s (sic) second quarter’s results are in and our support looks good our domestic group market share support for the six months (apr-jun) is up from 62% last year to 68% this year. While international support average for the group is now 31% compared with 27% a year ago. please compare this to your own figures when you receive them later this week.”

190. And

“3) a reminder regarding your allocation of free tickets with saa

1. Included in our contract as per last year is the ability for each suretravel agency to claim free and promotional tickets for their use.each agency is entitled to receive …the first two tickets are upgradable to business class if your saa market share support exceeds 35% for international & 50% for domestic..”

191. And

“4) we are still giving too much of our domestic business to nationwide, R40 million last year alone. we cannot hope to keep both saa and ba comair satisfied if we can give a non-preferred so much business. SOME MEMBERS have even signed deals and receive overrides from nationwide. ..this is completely unacceptable and where this is discovered, both the domestic saa and ba/comair deals will have to be pulled. Members who are also receiving an override from nationwide are an embarrassment and liability to our group. .. we are well aware of the very competitive pricing policy of nationwide and you will shortly see a preponderance of tactical specials from both saa/ and ba (in fact please view matchmaker today) for the first of SAA’s specials to counter this.”

192. And emphasising this message again:

“however if there is one thing that will make us stand out head and shoulders above any other group it is our individual commitment to our preferreds.. please reinforce this commitment on a daily basis with your staff and challenge ALL sales on non-preferreds”

193. In his testimony Puk however, contrary to the message he sends to his members as quoted, adopted the stance that to “mislead “ customers would not only be unethical but bad business practice and contrary to the ASATA charter to which all agents subscribe.¹¹¹

194. When re-examined by the Commission this high moral tone slips slightly:

“ADV PRETORIUS: I ask you again Mr Puk, what do you mean by channel it to your preferred partner? What does that mean in practice?

MR PUK: It means, where you have a discretion, where in the case of our fare is identical , giving the first shot to a preferred partner is directing the business, where possible to a preferred partner.”¹¹²

195. Puk’s position becomes more pragmatic later when he concedes:

“ADV PRETORIUS:.. So I am asking you again, which one is really paramount to the Managing Director of a firm, the consumer or the incentive , reaching the incentive threshold? Mr Puk from your point of view, what is most important?

MR PUK: From my personal point of view, if you are asking for the paramount, I am employed to make sure that the group achieves its preferred agreements.”

196. Puk was not a witness sympathetic to the Commission’s case. Sure Travel did not disclose the e-mail referred to above to the Commission despite a subpoena, and the Commission appears to have obtained it from another source.¹¹³ Indeed, during the course of the investigation Sure Travel had assisted SAA by writing a letter to SAA, to be forwarded to the Commission, in which Sure Travel denies that agents can move passengers.¹¹⁴ The concession by Puk, that for him group interests are paramount over consumer interests, is all the more significant because of this.

197. The one travel agent who was consistent with what his firm had written in correspondence prior to the hearing and in his testimony at the hearing was Mr Conrad Mortimer, the commercial director of Tourvest Holdings (Pty) Ltd.

198. Again, we start with the correspondence. The Commission’s counsel led Mortimer on several letters from Tourvest companies to SAA indicating their continued commitment to supporting SAA, in one case expressing it as “despite lucrative British airways and Comair override agreements”. ¹¹⁵

199. Mortimer also refers to various incentives to consultants on his agency’s staff to increase revenue from SAA sales. Mortimer confirms a circular to staff regarding a competition to win a car. According to the circular:

1. ” Remember all you have to do to be in line for the car is SELL SAA!!! Not difficult, so with one month to go, let’s give it all we’ve got.” ¹¹⁶

200. Mortimer confirms that the incentive worked and that Tourvest achieved R10 million above its flown revenue target.¹¹⁷

201. But perhaps the most important aspect of Mortimer’s evidence is his completely frank, albeit measured, opinion on agents’ ability to move customers:

“…this is not simply a matter of an agent being able to turn on or turn off a tap. The tail doesn’t exactly wag the dog. The end customer does and the market does have its own dynamic. But within that market dynamic the agent is able to, from time to time, exert a greater or lesser influence.”¹¹⁸

202. Cross-examined on this point he stuck to his position:

“We may. We may and it certainly would be in our commercial interest to promote our preferred. It’s very simple. We are not going to make any profit out of selling a non-preferred’s ticket. We’re going to basically break even on trading. If we’re going to make profit, we’re going to make profit because we sold a preferred carrier.” ¹¹⁹

203. Mortimer indicated that where agents have this opportunity they promote their preferred supplier:

“..wherever we have the opportunity we promote our preferred supplier and that can and has been at times highly lucrative and it is on that basis that we are able to achieve our volume incentives and generate profitability in our business.” ¹²⁰

204. There is no motive for Mortimer to exaggerate this evidence – if anything, a travel agent would be inclined to claim the moral high ground and assert its independence and professionalism despite commercial pressures to the contrary. Nor does Mortimer show any inclination to please the Commission. The Commission indicated that although there had been a consultation with Mortimer, he had declined to make a statement. Some of the extracts from the evidence that we have quoted was elicited from cross –examination by SAA’s counsel. For these reasons we believe that we can place great reliance on his testimony.

205. SAA’s rival, BA/Comair, seems to be in no doubt on travel agents ability to influence as this exchange between SAA’ s counsel and Mr Venter illustrates:

ADV SERRURIER: Just explain to me would you? Is it your thesis that a travel agent is in a position to influence the customer, the passenger the flyer?

MR VENTER: Yes very much so.

206. We are satisfied that there is sufficient evidence of travel agents’ ability to influence customer preferences. Whilst the evidence is at the level of their potential to do so, and we do not have data of actual customer movements, which would be extremely difficult to obtain, the evidence of the potential is so widely perceived, not only by the different travel agents as emerges from the correspondence and the testimony, but also the airlines, including, as we have shown, SAA.¹²¹ This widespread faith in the ability of agents to influence suggests that they in fact do so. This is not to say that customers are putty in the hands of agents who bend them to their will. Mortimer is careful to give a very nuanced view of this ability. What does emerge, is that all players consider that agents can exercise a significant influence on choice and that this has commercial implications for airlines. If not, as we shall point out in the conclusion, there would be no logic to the incentive schemes in their current form and in particular, the override and incremental aspect of those schemes. The logic of the override and incremental incentive is premised on the ability of agents to direct a non-trivial amount of revenue to their preferred. The same of course can be said of the Explorer Scheme. Why should it exist at all if travel agents cannot move customers?

207. Of course an important part of SAA’s defence, through all its witnesses, has been to assert that even if agents had an incentive and inclination to move passengers to SAA, they would fail as consumers are not that easily duped. A customer who had paid more for a ticket as a result of the travel agent’s machinations would soon find out and the agent would suffer a reputational loss that may exceed the short- term gain of the marginal commission earned.

208. The problem for SAA is that information on ticket prices is asymmetrical because of their volatility and complexity. Viljoen in his evidence gives a prime example of this complexity. He says that there are ten levels of discount between business class and economy.¹²² If this is so, it must inhibit the ability of consumers to monitor prices that agents quote them. It is also clear, that at least during the reference period, the level of internet usage for ticket bookings was still in its infancy, and hence the best tool for making the market more transparent to consumers was not yet fully functional. Consumers are also aware that ticket prices are a function of demand over time. Since consumers have no access to cycles in demand from time to time, informational asymmetries are easily maintained. Consumers need not suspect they are being duped simply because they hear different ticket quotes for a similar service at different times. They may assume the difference is simply a function of demand movements over time.

209. Furthermore, ticket availability is also never a constant. Were agents to suggest that a rival airline had no tickets available, particularly if they are SAA’s rivals who are known to have less capacity, consumers may have no reason to suspect that this may not be so and given that occupancy, like price, is so transient a notion, it would also not alert suspicions if the consumer later became aware that seats were available. In essence, the consumer’s ability to police opportunist behaviour by agents, is seriously constrained in this type of market, because of the informational asymmetries that exist.

The Explorer Scheme

210. Although much the bridesmaid during the course of this hearing, in terms of attention afforded to it, there is still evidence of the exclusionary nature of this scheme. Perhaps it got little attention because SAA’ own experts were sceptical about its pro-competitive value.

211. Professor Du Plessis testified that the scheme’s impact at the individual travel agent consultant level reinforced that of the more general incentives created by the override schemes signed at a group level for the travel agents. In fact, in his view, explorer incentive scheme would in all likelihood have had a far greater anti-competitive effect, since a points-based system for rewarding loyalty to the consumer at least has some positive impact in terms of benefits to that consumer. However since the Explorer scheme was applied at the travel agent level or the consultant level, there were no welfare-enhancing benefits for the consumer.¹²³

212. This theme is echoed in the testimony of the Commission’s expert. Holt testified that the Explorer Scheme was more likely to have an anti-competitive effect, as the reward is not granted to the consumer but to the intermediary in the form of the travel agent and its employee the consultant. He thus distinguished it from a frequent flyer program, which at least rewards the consumer.

213. The language contained in the scheme document is instructive as to its intentions. In explaining the bonus pool arrangement, an additional feature of the Explorer scheme, SAA states:

In order to further reward travel agents who proactively support South African Airways, in addition to rewarding individual consultants, travel agents can earn a special bonus pool of South African Airways Explorer points bases on South African Airways share support of individual agent location in the monthly BSP reports.” ¹²⁴ (Our emphasis)

214. When the interesting use of the word ‘proactive’ was put to Viljoen in cross-examination, he suggested that proactive means:

1. “Just doing it a bit smarter”. ¹²⁵

215. Viljoen’s answer seems to run contrary to the ordinary definition of the word, which according to the Concise Oxford Dictionary means:

“creating or controlling a situation rather than just responding to it.”

216. This we would suggest is precisely what the scheme was all about. Travel agents, and in particular consultants, who were after all the workers at the coal-face who might not always have the same incentives as the shareholders who own the agencies, some of which are large corporations, were being given the incentive to be ‘proactive’ about how consumers decided between airlines. It was a useful complement to the override incentive scheme and filled the gaps left in that scheme to resolve the principal -agent problem in SAA’s favour.

217. We find then that the Commission has established that the override incentive scheme provides a compelling commercial inducement to agents to prefer selling SAA tickets to those of its domestic rivals and secondly, that to a significant extent, agents are able to influence customer preferences so as to give effect to these incentives. The Explorer scheme serves to enhance these exclusionary effects.

218. The Commission has thus, on a balance of probabilities, established that the practical effect of the schemes is that they induce suppliers not to deal with competitors of SAA and hence constitute an exclusionary act in terms of section 8(d)(i). Having made this finding it is not necessary for us to consider whether the Commission has made its case in terms of section 8(c).

Does the exclusionary act have an anti-competitive effect?

219. Recall that we said an anti-competitive effect could be manifested in two forms. Either there is direct evidence of an adverse effect on consumer welfare or evidence that the exclusionary act is substantial or significant in terms of its effect in foreclosing the market to rivals.

220. Our present enquiry will be limited to examining whether the Commission has established evidence of the latter proposition. We say this as it appears to be common cause that the Commission has not established an adverse effect on consumer welfare except by way of inference. Expressed less technically we have no direct evidence that consumers are paying more for their domestic airline tickets or have experienced less choice in flights or inferior service. That does not mean that there have not been these effects. The Commission asks us to infer them rather than putting them to the burden of proving what the market may have been but for the exclusionary nature of the override scheme.

221. We therefore examine the other test for anti-competitive effect we identified earlier in the legal section as to whether the exclusionary act is substantial or significant in terms of its effect in foreclosing the market to rivals.

222. The Commission’s evidence here is of the following nature. Firstly, direct evidence of the extent to which SAA had imposed its override scheme in its post 1999 form on the travel agents market. Secondly, evidence of a decline in the performance of Nationwide and BA/Comair during this period – this constitutes circumstantial evidence of the effect of the alleged foreclosure. This must be coupled with some evidence from the travel agents of their increased business with SAA. Thirdly, the economic theory, derived in part from the number of the agents partaking in the scheme and the extent of SAA’s dominance that suggests, by way of theoretical inference, that significant foreclosure would be inevitable and that the cost to rivals of compensating agents to meet the exclusionary effect of the scheme would be prohibitive.

223. Travel agents. The evidence is that travel agents accounted for approximately 75% of sales of domestic airline tickets during the period April 2000 till March 2001.¹²⁶ SAA had 19 override agreements by the end of March 2001. All four major groups, were covered by agreements as well as smaller agencies. This represents, if one disaggregates the groups into their individual agencies, 683 agencies.¹²⁷ Holt testified that the “agreements appear to cover a very large percentage of the industry.” Although we do not have an exact figure for this, his conclusion was not challenged so we can accept it. Thus agents subject to the scheme represented a significant number of agents who in turn constituted a significant vehicle for ticket distribution. We know that 85% of SAA’s ticket sales were through travel agents.¹²⁸ Although rivals of SAA had a lower proportion of their sales through travel agents, the figures remain high - BA /Comair - 74% and Nationwide - 70,2%.¹²⁹

224. We can thus conclude that a significant portion of the travel agents market, itself a significant channel for ticket sales, was subject to override agreements.

225. The Commission led the evidence of executives of both Nationwide and BA/Comair in support of its case that these firms had suffered adverse effects on their businesses pursuant to the implementation of the scheme.

226. The evidence of Mr Bricknell, the chief executive of Nationwide, was that his airline commenced business in 1995 and for its initial three years growth was considerable. In about 2000 the airline noticed a decline in growth. Since he had expectations that Nationwide should still be growing he asked his financial director to investigate the cause. The feedback that he received from his staff was the travel agents were telling him that they were so highly incentivised to support SAA that Nationwide could not compete. As he termed it “they were being off-sold.” Pressed in cross-examination he said he attributed this trend to the SAA override incentive scheme and Explorer Scheme.¹³⁰

227. The Commission has produced a table of the number of passenger flown by Nationwide during the period December 1995 until October 2004.¹³¹ The table shows that from November 2000 Nationwide started to experience decline in passenger numbers when compared with numbers for the equivalent month in the previous year. This decline persisted until August 2001, declines again in January 2002 and is thereafter positive. Prior to this period, one already observes a decline in the rate of growth as early as August 2000 when one sees a growth of only 2,9% after three previous months where growth was 61%, 59% and 35% respectively. For the period November 2000 until January 2002, the monthly moving average (i.e. the sales recorded in the immediate proceeding 12 months) is in continual decline. The figure at the beginning of the period was 526 888 passengers and declines to 458,549 passengers in January 2002.

228. This is the way Holt interprets this data:

“MR HOLT: Well the first point to note is that the overall market growth was less than the 10% that we had for the purposes of our model assumed. So therefore, our growth rate in that model was, if anything, higher than one would have required to reflect these particular years.

I think the second point, and perhaps even more important point to note here, is that if you look at the overall market growth and compare the BA Comair and Nationwide performance to those market levels of growth, what you typically find is that before the period of the override incentive agreements, BA and Nationwide were experiencing above market growth and for the period later on, once the override agreements were in effect and during the period in question, their above market growth turned into below market growth or even decline.¹³²

229. SAA does not deny that the figures show a period of decline in Nationwide’s figures that coincides with the relevant period. However it denies any causality between the override scheme and the decline in performance. SAA says, in the first place, since Nationwide had grown from a nil base it could hardly expect its rate of growth to be maintained once its market share matured. Secondly, SAA argues that the override scheme was in operation at the time that Nationwide experienced its rapid growth. It points out that Nationwide’s performance improves later at a time when overrides are still in place. If this is the case there must be some factor other than the overrides that accounts for the decline and rise in fortunes of Nationwide.

230. Through the evidence of Viljoen and the cross examination of Bricknell, SAA suggests an alternative theory.

231. Bricknell under cross-examination conceded that in June/July 2000 Nationwide had increased the price of its tickets, erroneously anticipating that the rest of the market would do as well.¹³³ As a result of this, Nationwide lost market share although he expressed the view that the loss attributed to the price hike was temporary, as they re-adjusted prices to the competitive level soon thereafter. (Note that his financial manager Mr Griffiths testified later that the hike had lasted only two days).¹³⁴ It was also suggested in cross examination to the Nationwide witnesses that their firm was believed to be close to bankruptcy at the time and this might have had an effect on public confidence in the airline.¹³⁵

232. Viljoen testified that in late 1999, SAA embarked on an aggressive strategy to improve its performance as they had lost market share in previous years. This included better pricing, more reliable take off times and better overall service. In this he was supported to some extent by the evidence of Harris. This explains the improvement in SAA’s performance at the expense of its rivals.

233. SAA also relies on board minutes of BA/Comair dated 28 December 2001. The board was discussing the reasons for BA/Comair’s poor performance on the Johannesburg /Durban route in the months preceding the meeting. The board identified four possible causes (1) the increased frequency of SAA flights (2) lack of economic development in the Durban area (3) the SAA override scheme and the incentive bonuses paid to travel agents (4) the power of the SAA Voyager program. The board discusses the lodging of a complaint with the Commission in respect of the override scheme. The minutes also record the Chairman as noting that regardless of whether BA/ Comair lodges a complaint it would still have to find new ways of competing with SAA as it could not compete with it on frequency. ¹³⁶

234. Mr Venter, BA/Comair’s financial director, testified that while there is a relationship between the size of an airline and the market share that it can realistically gain, they were seeing a decline in their market share to which he attributed the influence of travel agents:

“ But the fact that we started to see a decline in market share ..there we believed there was strong influence from the travel agents.”¹³⁷

235. The figures produced by the Commission show a decline in Comair’s rate of growth in respect of ticket sales through Johannesburg airport from 11,97% for 96-97, and 14,5% for 97-98, to 4.05% for 98-99 and for 2000 to 2001, 0,2%.¹³⁸

236. What do we make of this conflicting evidence? It appears, and SAA does not seriously challenge this, that both Comair and Nationwide suffered a decline in performance during the relevant period. We have contemporaneous evidence from BA/Comair through its board minutes that the company was concerned with this at this time. We also know that Nationwide was too, which led to the interim relief case and the complaint that gave rise to this matter. What SAA disputes is whether its rivals misfortune can be attributed to the effects of the scheme. It argues that in the case of Nationwide, it had grown rapidly during 1995-8, when the evidence is that override agreements were in place and that it also showed renewed growth after the relevant period, again at the time that override agreements were in effect. This proves, SAA argues, that the scheme could not have been the cause of Nationwide’s downfall. SAA’ s theory of a combination of its own renewed competitiveness in the form of better service is the explanation, coupled with some poor business decisions made by rivals. Its rivals therefore suffered as a result of its competitive response, not the override scheme.

237. The Commission argues, convincingly in our view, that the performance of Nationwide in the period before and after the relevant period misses the whole point of the case. It was not that override schemes themselves are a problem, but the nature of the override scheme, as SAA changed it in late 1999 by lowering the standard commission and increasing the override commission and hence materially altering the incentives of travel agents. The decline is therefore causally consistent with the advent of the new override scheme. Because the figures for the later periods fall outside of the relevant period there may be other explanations for Nationwide and Comair’s recovery in their respective fortunes, but as we have insufficient data for this period, we must postpone making conclusions. For instance we don’t know whether bookings through travel agents were not as central to the later period as during the relevant period nor if they were, how much of the market was subject to override agreements and whether they were in the same form. Mortimer’s evidence suggests that post March 2001, American Express did not continue with their override agreement.¹³⁹

238. We also have to accept the evidence that SAA had improved its performance during the relevant period. Whilst the evidence is not inconsistent with the Commission’s case, we simply cannot be sure which was the more preponderant cause of the decline - competition on the merits or the override scheme. But this difficulty is not unique to this case. All cases of exclusionary anti-competitive conduct create the dilemma that the counter-factual, namely what the market would have looked like absent the alleged prohibited practice, is impossible to construct. An exclusionary act is not confined to causing the decline of rivals, it is equally exclusionary if their opportunity to expand in the market is retarded or constrained. As the Court of Appeals in Microsoft observed:

“ To require that section 2 [of the Sherman Act] liability turn on a plaintiffs ability or inability to reconstruct the hypothetical market place absent a defendant’s anticompetitive conduct would only encourage monopolists to take more and earlier anticompetitive action. We may infer causation when exclusionary conduct is aimed at producers of nascent competitive technologies as well as when it is aimed at a producer of established substitutes. Admittedly, in the former case there is added uncertainty, inasmuch as nascent threats are merely potential substitutes. But the underlying proof problem is the same – neither plaintiffs nor the court can confidently reconstruct a product’s hypothetical technological development in a world absent the defendant’s exclusionary conduct. To some degree, “the defendant is made to suffer the uncertain consequences of its own undesirable conduct.”

239. The fact that SAA made use of an anti-competitive scheme at the same time as it behaved more pro-competitively does not immunise it from any consequences for its misdeeds in terms of the Act, rather it means that it must suffer the ‘undesirable consequences’ of the former co-existing with the latter.

240. We find that anti-competitive effects are likely for the following reasons:

1. At a theoretical level, Holt has demonstrated convincingly the anticompetitive nature of the agreements in question. Whatever their effect was in isolation, the Explorer scheme grafted on top of them puts their nature beyond doubt;

2. The evidence is that Nationwide and BA/Comair experienced a period of decline after strong periods of growth that coincided with the periods in which the override agreements became more challenging;

3. The evidence of BA/Comair and Nationwide not only during this hearing but more importantly their thinking at the time and before litigation had commenced. Recall the board minutes of BA/Comair and the visit by Bricknell to Boyle at Rennies.

241. We find further that the effect of the anticompetitive conduct on the structure of the market was to inhibit rivals from expanding in the market whilst at the same time reinforcing the dominant position of SAA. Therefore the exclusionary act is substantial or significant in terms of its effect in foreclosing the market to rivals.

242. Although, because of this finding, we do not need to make a finding that there was actual harm to consumers, despite the lack of direct evidence on this point, it is highly likely that this foreclosure has had adverse effects on consumers. As Holt has put it the conduct would lead to allocative inefficiency. This means that consumers are likely to have made wrong choices of airlines, chosen the wrong prices and essentially, it has led to the wrong set of outputs.¹⁴⁰

Efficiency justification

243. Having found that the Commission has established that the override scheme, in conjunction with the Explorer scheme, is an exclusionary act as contemplated by section 8(d)(i), and that it has an anticompetitive effect, we must now evaluate the efficiency justification raised by SAA. If we find an efficiency justification exists, we then evaluate whether it outweighs the anticompetitive effect. On both issues the burden of proof shifts to SAA.

244. SAA relies for its efficiency defence on evidence led by Viljoen, Harris and its economist, Professor Du Plessis.

245. Viljoen was, according to his testimony, central to SAA’s decision to change the nature of the remuneration provided by the override scheme. If anyone can testify to the efficiency gains of the scheme it would be him. Yet his evidence on this matter fails to establish the connection between the nature of the scheme and the benefits he sought from the travel agents thereto. Viljoen explains that the relationship between the airline and the travel agent is subject to the contradiction that the airline wants to minimise costs whilst the agent want to maximise commission. SAA was busy exploring alternatives with agents. He mentions that SAA gave 15 million rand to ASATA to look at alternatives and that working groups had formed on these issues but that after 18 months they had not yet reached agreement.¹⁴¹ Viljoen testified that SAA wanted travel agents to be incentivised to undergo training to learn about the SAA product. What this meant in practice was never very clear in his evidence, but the clearest articulation emerged when he was being pressed on the incentive behind the Explorer Scheme:

“Yes so how do you sell. You must know what you are talking about. You must know the product. You must know the service. You must know the network. You must know the aircraft. You must know the seating, know the food. You must know all that stuff.” ¹⁴²

246. Yet when pressed to explain the relationship between these goals, laudable as they may be, and the manner in which the schemes functioned, he was a lot less clear. During cross-examination he was asked how he expected American Express to achieve incremental growth of 35% he says:

“ Well, in the negotiations they put on the table to us that they were training their staff. They were expanding the infrastructure countrywide, that they believed that they would achieve those sorts of growth figures, but not by not selling on [off] other airlines but by stimulating the market, by having more offices.”¹⁴³

247. The evidence of Harris is no more helpful in establishing this link. Harris testifies to the benefits of the relationship describing being given direct access to key people in the event of problems so that they could be ironed out. Yet her evidence also does not explain why this type of service from SAA is necessarily linked to the override scheme.

248. On this shaky factual premise Professor Du Plessis has had to construct a theoretical basis that links the scheme to the claimed efficiency. He, following his witnesses, makes as the first claim that the scheme incentivises agents to improve their knowledge of the airline’s products and that as a result there is a better match between airline and customer. He describes these as trade creating, rather than trade diverting gains. It is by no means clear why this is so.

249. There is nothing about the mechanism of the schemes to show how they would be attractive to the consumer and hence trade creating in the sense that these would be attracting clients to SAA and thus growing the market.

250. The second benefit mentioned is that the agent has an interest in improving the airline’s service by regularly informing the airline of improvements needed, and complaints by customers. To some extent this benefit has a greater relationship to the override scheme than the previous one. It may well be true that if agents’ fortunes are tied to the fate of SAA, they have an interest in advising SAA on how to improve shortcomings so that they can meet their targets. The scheme could therefore encourage efficient information exchange between agent and principal, which the agent, absent such an incentive, may be unwilling or uninterested in communicating. However this is not the only method of achieving this type of benefit, nor is it so overwhelming as to outweigh anticompetitive effects. Nor is it likely that SAA, which sells tickets directly to customers and must also presumably get its share of customer feedback, needs the override scheme to achieve this end.

251. There is of course a fatal paradox in the manner in which SAA advances the efficiency argument. If agents cannot influence customer choices, as is SAA’s central contention, why embark on an arrangement to influence their behaviour when they are passive recipients of consumer’s predetermined choices? It can only be a rational incentive if one acknowledges that agents do influence choice to some extent. Du Plessis ‘ answer is to suggest that they help grow the market. But it is not clear how they do this without influencing consumer choice. There is no evidence of how, armed with their superior knowledge of SAA product, they put more people on to planes. The most obvious method by which they might do this is to provide lower prices on SAA tickets so as to promote air travel – but the evidence suggests that they cannot do so and indeed are discouraged from intra brand competition between agents as, according to Mortimer, SAA does not want to lose control over its pricing. ¹⁴⁴

252. If they grow the market by marketing promotion for instance by announcing specials that are available, granted this may bring in new passengers who might otherwise not have chosen to travel or to fly, but this kind of promotion does not require an incentive scheme. SAA can find numerous other ways of rewarding agents from this type of promotion of its products without the exclusionary nature of the current scheme.

253. The disjuncture between the efficiency benefits claimed and the mechanism of the scheme was well articulated in our view by the Commission in its heads of argument. The Commission points out that, because of the target set in the schemes as the basis for reward, a travel agent that undergoes all the training but fails to meet its target receives no reward, whilst another who does not, but makes its target, will.¹⁴⁵

254. The Commission also relies on several European cases where similar target based incentive schemes were examined.¹⁴⁶ The cases indicate that the nature of these schemes is not to promote efficiency because they are not volume driven, but rather to promote loyalty. As the European Community observed in the BA/Virgin case:

“ A travel agent that sells an inefficiently small number of tickets can earn the maximum commission provided its small sales represent a 25% increase over its sales in the previous year. Equally, a high volume travel agent will not get extra commission in return for the economies of scale it realises for BA unless its sales increase over the previous year. .. Travel agents are encouraged to remain loyal to BA rather than to sell their services to competitors of BA by being given incentives to maintain or increase their sales of BA tickets which do not depend on the absolute size of those sales.”¹⁴⁷

255. In the SAS case the Swedish authority points out that one of the objectionable features of the scheme it was considering was that the bonus scales were subjective:

“The bonus scale is thus not constructed on objective grounds but is wholly and completely adapted to the individual customer’s previous purchases.”¹⁴⁸

256. The objectionable features identified in these cases are present in the SAA override scheme. There is thus in our view no logical nexus between the efficiency gains that SAA claims the scheme seeks to achieve and its mechanism. If as Holt has pointed out it wanted to achieve these objectives, the scheme would have been constructed differently. The far more plausible explanation of the scheme is the one advanced by the Commission that it is the mechanism for inducing loyalty to the dominant carrier to the exclusion of its rivals.

257. We are not satisfied that SAA has proved that the override scheme or the Explorer scheme, provides any technological, efficiency or other pro-competitive gains that outweigh their anti-competitive effect.

258. Accordingly we find that SAA has contravened section 8(d)(i) of the Competition Act by the implementation of the override scheme. We find further that use of the Explorer scheme has contributed to the anti-competitive effects of the scheme. We find that, although the override incentive scheme on its own would constitute a prohibited practice in terms of the Act, the same cannot be said with the same certainty of the Explorer scheme. This is because we do not have sufficient evidence to know if the latter scheme, taken in isolation, would have constituted a prohibited practice. We do know however that it served to complement and to enhance the anti-competitive effects of the override incentive scheme.

APPENDIX 1

PART II - REMEDY

259. We must now consider what remedy is appropriate given our finding that the respondent has contravened section 8(d)(i).

260. The Commission initially sought four remedies; a behavioural remedy, a declaration that the conduct constituted a prohibited practice, an order declaring that the relevant provisions of the Explorer and override schemes be declared void, and an administrative penalty.

261. At the hearing on 7 December 2004 the Commission abandoned its proposal for a behavioural remedy and we need not consider this issue any further. We deal with the remaining forms of relief.

Declaration and Voiding

262. The Commission seeks an order that the provisions of the Explorer scheme and the override scheme are declared prohibited practices. The Commission further seeks an order that the relevant provisions of the override agreements and the Explorer scheme be declared void. The Commission does not detail which agreements these are and which provisions should be voided. The Commission states specifically that it does not propose alternative formulations for the voided clauses. ¹⁴⁹

263. SAA’s approach to the declaratory relief is twofold. It firstly opposes the relief on the grounds that it has not contravened the Act. That part of its defence is now academic given our finding that it has. It then specifically opposes any voiding of the clauses in the agreements, as that would create practical problems for it until it had a viable alternative.

264. Apart from arguing that we should not find against them SAA does not make any specific comment on the merits of a remedy declaring the Explorer Scheme and override scheme restrictive practices. We presume that this is because SAA accepts the logic that if we find against them such a declaration would automatically follow. It is accordingly only necessary for us to consider the terms of the declaration. The Commission says no more than that we should declare the practices to be prohibited. We find however that the declaration needs more flesh put on it. Given that on the evidence, the mere existence of an override scheme may not on its own constitute a prohibited practice, but that it is the nature of the scheme that counts (that is after all the Commission’s case on the merits) we narrow our finding to the period in which the override scheme was found, in its present form, to be an abuse of dominance.

265. We noted at the beginning of this decision that the evidence before us related to different time periods. Thus although in many instances data from the airlines has covered longer periods, evidence of the content of the agreements has covered a shorter period. The Commission states that the ‘challenging’ override agreements commenced in April 1999, but according to Viljoen’s evidence the date is October 1999. Then to add to the confusion, June 2001 is the end date for much of the data presented in the Commission’s most recent set of figures. Despite this the Commission in its affidavit on remedies states that the investigation and evidence led “ focussed on the period June 2000 to May 2001. ¹⁵⁰

266. We have thus confined the declaration to a period where evidence of both for the most part coincided and we felt confident of making a finding for that period. We have taken a conservative view of whether these findings could be extended beyond these periods, even though we have had some evidence that extends beyond this period. The period of October to May 2001 is thus a cautious determination of conduct one suspects continued for much longer, on either side of these dates.

267. In relation to the Explorer Scheme we have no evidence when it commenced, it appears at the very least to have been in operation by October 1999. We know however from SAA that the scheme stopped in June 2002. We could have declared that the Explorer scheme was a prohibited practice for this period. However, given our finding, that on its own, there is insufficient evidence that the Explorer scheme constituted a prohibited practice, we confine our declaration to the period when we know that it coincided with the period in which the override scheme was in operation in an unlawful form.

268. For this reason we have confined both declarations to the period from October 1999 until May 31 2001.

269. We do not consider it appropriate to void the specific clauses in the agreements. In the first place as we noted, the Commission has failed to give any particulars. We don’t know for instance if any of the agreements are still currently in force, it is, after all, four years or more since the relevant period. Secondly, we do not know whom they are with and thirdly, which are the relevant clauses that should be voided in each of the agreements. We are also concerned about the issue of joinder. We have previously found that we cannot void an agreement unless both parties have been joined. ¹⁵¹ In the present case, the Commission would have us void agreements with travel agents where none of the agents have been joined to these proceedings. Without hearing them, we cannot void the agreements. For all these reasons we deny this aspect of the relief.

Administrative penalty

270. As we have found that SAA has contravened section 8(d)(i), it is competent although not mandatory, for us to impose an administrative penalty. In terms of section 59(2) of the Act an administrative penalty may not exceed 10% of the firms annual turnover in the Republic during the firm’s preceding financial year.

271. As we noted in Federal Mogul it is not clear from the Act what the preceding financial year or ‘base year’ refers to, as there are several possibilities.¹⁵² In that case we did not need to decide the point as the parties had agreed what the base year should be.¹⁵³ In this case the Commission has suggested that the base year should be June 2000 to May 2001. SAA’s position on this issue is far from clear. In his affidavit in respect of the remedies, Mr Chavarika, SAA’s executive legal counsel, criticises the Commission for its arbitrary selection of the base year and then in the same breath, accuses the Commission of arbitrary selection in respect of the investigation period of its case.¹⁵⁴ Yet later on, in the affidavit when asking in the alternative for a lesser fine than that sought by the Commission, Chavarika makes use of the same base year. Nor has SAA, in its heads of argument on remedies, put in issue the choice of base year. We have decided to follow the Commission’s proposal for the base year as despite SAA’s criticism of arbitrariness, it has neither questioned the legal basis for the Commission’s choice of base year nor offered its own alternative.

272. It is common cause that SAA’s total annual turnover for that year was R10 396 096 000.¹⁵⁵ However, this is not the figure that the Commission relied on to constitute the base revenue for the purpose of calculating the fine. Rather the Commission has relied on a lower figure that it says represents SAA’s turnover in the ‘affected market’ described as SAA’s flown revenue through travel agents in the domestic market. This figure says the Commission is R2 022 124 775. It is this figure that the Commission argues should serve as the base revenue.¹⁵⁶

273. Sometimes a restrictive practice may have no relationship to a firm’s total annual turnover, as the relationship between the contravention and the total business to which that turnover may be attributed may be remote. In Federal Mogul we held that we would exercise our discretion to set the threshold for the calculation of a fine at a level lower than total annual turnover, where, in the appropriate circumstances, it might be more correct to select as the base, the turnover in the line of business where the infringement took place.

274. In this case the Commission has done just that. It has argued that as the abuse took place in the market for the sale of domestic airline tickets through travel agents, that should be the appropriate affected turnover for the purpose of reaching a threshold for the penalty. This means the figure of R 2 billion as opposed the R 10 billion. SAA does not dispute that this figure represents the domestic turnover of flown revenue sold through domestic travel agents. However, SAA argues that not all the tickets were sold through travel agents who were party to the impugned agreements and hence the R 2 billion figure is too high. (As has been its approach to other evidence in this case, SAA has queried a figure, but not given one of its own when it was in the best position to do so.) In our view this complaint does not raise a material issue. The affected turnover as represented by this figure is lower than the one that the Commission could legitimately have relied on, namely the total domestic turnover, given that the abuse on our findings had effects ultimately in this market. ¹⁵⁷The Commission’s choice to confine its base figure to the turnover in the smaller market is in SAA’s favour.

275. Even if we are wrong on this, the evidence suggests that the amount of turnover not covered by override agreements of some sort was insignificant.

276. For the purpose of this decision we will confine ourselves to the figure that the Commission has selected.

277. We will regard the figure of R 2 billion as an appropriate base figure on which to calculate the penalty.

278. We regard the circumstances of this case as one where an administrative penalty is the most appropriate form of remedy, particularly given the absence of a behavioural remedy and the legal difficulties with voiding provisions of the agreements.

279. Following our approach in Federal Mogul we now examine the factors that affect the level of the penalty as they are set out in section 59(3).

Nature, duration, gravity and extent of the contravention

280. The evidence is that SAA has had override agreements with travel agents since 1980.¹⁵⁸ But given that we have found that it was not the existence of override agreements, but the material change in their nature that constitute the abuse of dominance the period of the contravention is much shorter. The Commission in argument suggests that the challenging override agreements came into effect in April 1999. On the evidence of Mr Viljoen and the figures supplied by the Commission it would appear that the abuse commenced in October 1999. In any event we cannot go back any earlier than 1 September 1999, as that is the date on which the Act came into effect.

281. The evidence of how long the abuse persisted is more difficult to analyse. The Commission alleges that the abuse has continued for more than five years and was still continuing at the time we heard argument in March 2005. SAA does not deny this, but goes to great lengths to explain why it could not simply terminate the scheme until alternatives had been found. Indeed it suggests that rather than being admonished it should be commended for the reasonable steps it has taken to find a workable solution. As this relates more to the conduct of SAA than to the nature of the scheme, we consider this line of defence more fully below when we come to consideration of that factor.

282. However, although the evidence is that the scheme was still in effect at the time of the hearing, the only evidence we have of its effect is for the investigation period, which ends in mid-2001. We do not know for instance if the nature of the contracts changed in any respect after the investigation period ended. Recall that this has been an important part of our finding on the contravention that it is the nature of the override, not the fact of an override being in existence, that is of central concern. For this reason we give SAA the benefit of finding that the abuse in this case is only shown to have lasted a period of approximately 20 months.¹⁵⁹ We examine later, when we view the behaviour of the respondent, the significance of the fact that the schemes are still in place.

283. As far as the Explorer Scheme is concerned Chavarika, in his affidavit on remedies, alleges that the Explorer scheme was terminated on 1 June 2002.¹⁶⁰ The Commission has not led any evidence to the contrary so we accept this fact.

284. In our view the nature of the override scheme during the relevant period was of such a nature that the real intention must have been to make use of it in an anti-competitive, exclusionary manner, rather than to introduce efficiencies into the ticket distribution system, evidence, which we have already found to be wholly unconvincing. Its use in conjunction with the Explorer scheme at the same time aggravates its effect. The combined conduct whilst brief in duration, in terms of the evidence, was nevertheless of a seriously anti-competitive nature.

Loss or damage suffered as a result of the contravention

285. Two types of loss are possible here: loss to competitors and loss to the ultimate consumers by way of higher prices or less choice and inferior service.

286. The Commission argues that it has established that, at least during the investigation period, the fortunes of Nationwide and BA/Comair declined whilst those of SAA soared. Given that this change in fortune coincided with the more robust implementation of the override scheme, post October 1999, the Commission seeks to draw the inference that it was the unlawful conduct that caused the loss. SAA seeks to refute this. It relies partly on testimony of Mr Viljoen, who testified that at the same time as the override scheme was changed, SAA had introduced a number of other wide-ranging changes that made it much more competitive and that these changes explain the improvement in its performance. SAA also refers to evidence that Nationwide had, at one stage during the period, increased its fares in the erroneous anticipation that others would follow. When they did not, Nationwide experienced a decline in sales. Whilst Nationwide have acknowledged the setback attributable to the increase, its financial manager, Peter Griffiths alleges that this did not account for any substantial decline as his firm corrected its pricing soon after the market response.¹⁶¹

287. The BA/Comair board minutes are the best evidence of the response of that firm at the relevant time. But this evidence, as we have noted is equivocal. It notes the improved performance of SAA, but also shows concern about the effect on its fortunes of what it regarded as anti-competitive activities by SAA, inter alia, the override and Explorer schemes.

288. This leaves us with an uncertain picture of cause and effect. SAA argues that because of this we should not assume that the decline in Nationwide and Comair’s performance during the investigative period is attributable to the override scheme and Explorer.

289. To some extent, SAA is correct. We have a situation where more aggressive competition on the merits coincided with conduct we have found to be anti-competitive. We also know from the figures that the Commission has produced, and which we analysed in our decision on the merits, that both Nationwide and Comair showed a marked decline in performance during the investigative period. Does this mean that SAA must be given the benefit of the doubt and that in the absence of evidence of harm, uncontaminated by harm from competition on the merits, make no conclusion at all? To do so would be as dangerous as to conclude that mere harm to competitors means by inference the presence of anti-competitive conduct. Seldom will one find a situation where a respondent firm’s behaviour in the market place is either all good or all bad. To abstain from attempting to make a finding because we find the presence of both simultaneously too hard to call, would be to do a disservice to the enforcement of the Act.

290. What we must do is to test whether on a balance of probabilities the harm proven may to some or more extent also be attributable to the contravening conduct.

291. In this case the evidence, if not the cause, of harm is common cause. So is its coincidence with the more aggressive implementation of the override scheme. We also have the evidence of Mr Holt that the override scheme was likely to affect the behaviour of travel agents and this would impact on rival firms’ sales. We further have two items of contemporaneous evidence by the rival firms. In the case of Nationwide we have the evidence of Mr Bricknell that on observing a decline in his firm’s sales he ordered his financial manager to conduct an investigation as to the cause.¹⁶² At first, this evidence sounds strange – why order an investigation in this formal way, surely the firm knew what was going on in its own market? That might be more plausible if the reasons were solely the pro-competitive conduct of SAA. The fact that he suspected something more below the surface, a suspicion that turned out to have some foundation, is an indication that the override scheme, not obvious in its effect to outsiders, was playing some role in Nationwide’s recent decline.

292. But his suspicion that something more than vigorous competition on the merits was lurking, is more than the mere whim of a single disappointed competitor. We know from the Comair minutes that its board was also of the view that the improved performance of SAA, which they acknowledge, was not the sole cause of the losses Comair was experiencing at the time and their suspicion was that certain practices of SAA, the override scheme among them, were prime suspects as well.

293. By examining this evidence cumulatively, we are satisfied we can find that the override scheme and Explorer were the most probable explanation for some of the decline of SAA’ s rivals during the relevant period. Granted we cannot be certain how much, but in our view it was not a trivial cause of this loss.

294. The problem of assessing the loss as we noted in Federal Mogul is that the counter–factual, namely what the market would look like without the prohibited practice having taken place is always unknown. For it is not merely a case of how much worse off competitors are shown to be, the issue we have examined thus far, but also how much better off they might have been absent the abuse. Nationwide, we know, showed a spectacular rate of growth in its early entry into the market, which ended during the reference period. Granted, as SAA would have it, the rate of growth would decline as their market share increased after entry, but the conduct may well have arrested the fortunes of a very effective entrant. We also know from the internal correspondence of some of the travel agencies, which we quoted earlier, the strong pressures that the override scheme was imposing on the industry and it is highly probable that this must have injured Nationwide and BA/Comair.¹⁶³

295. In our view, the evidence shows some loss to the competitors of a not insignificant nature although more precise quantification is impossible. To some extent, this gives SAA the benefit of the doubt.

296. On the other hand, it needs to be noted that the effect of loss on the present evidence was for a limited period and that none of the competitors was forced out of the market.

297. As SAA are quick to point out, there is no evidence of loss to consumers. The Commission concedes that while there is no direct evidence on this point the loss to consumers may be inferred from the loss to competitors. By increasing the dominance of SAA, the scheme has contributed to making the market less responsive to rivalry and hence to the detriment of consumers.

298. We find again that it is highly likely that the scheme did have an impact on consumers because of its effect on the performance of SAA’s rivals. However we do not know what the extent of this is because the counter factual is unknown. We find that although loss to consumers is highly probable, the extent of that loss is uncertain.

Behaviour of the respondent

299. SAA argues that not only is the override scheme a practice known world wide in the industry, but it has been used in the domestic market by the complainant Nationwide and BA/Comair. The Commission argues in contrast that it was well known, or should have been to SAA, that override schemes had fallen foul of competition law elsewhere, but that SAA nevertheless continued to implement them. The Commission is referring to the cases in the European Union and some member states.¹⁶⁴ SAA however point to a case in the United States where the scheme was not found to contravene section 2 of the Sherman Act.¹⁶⁵ SAA also argues that the fact that a scheme may fall foul of one jurisdiction’s competition regime does not mean that the similar conduct is necessarily unlawful in our own.

300. Whilst evidence of unlawfulness in some jurisdictions does not give rise to any certain conclusion that a similar practice will be impugned here, a firm of SAA’s sophistication, which of its own admission has followed international practice in implementing the scheme, should have been more alert to the fact that from international developments its conduct was at least in the twilight zone of legality. Far from leading the firm to becoming more cautious in this regard it chose, coinciding with the introduction of the present Act, a fact alone that should have induced greater circumspection, to advance further toward the darkness of illegality by its introduction of the more ‘challenging’ scheme in October 1999.

301. Thus we have a firm uncertain as to whether its conduct is lawful choosing to exacerbate it instead, as it could have done, going to the Commission for an advisory opinion. Nor did SAA do anything to make this enquiry when its conduct was challenged by Nationwide in its interim relief application in 2000. Whilst the Tribunal found against Nationwide on this point, this was because the factual issues had not been sufficiently canvassed not that the theory advanced in the Virgin case was wrong in law. Indeed we expressed the view that we should not be surprised if SAA was a dominant purchaser in the market for travel agent for travel agent services. ¹⁶⁶ This observation too had no salutary effect on SAA’ s conduct.

302. The European Court of Justice has held that a dominant firm is considered to have:

“a special responsibility not to allow its conduct to impair genuine undistorted competition in the Common Market.” ¹⁶⁷

303. Whilst SAA may argue that it did not know it was a dominant firm, and hence lacked an appreciation of its ‘special responsibility’ it ought, at the very least, have been alert to the dangers inherent in a scheme that was, to put it euphemistically, controversial with other competition regulators. Nor could it have been ignorant of the size of its market shares whatever its quibbles may have been over the Commission’s methodology - even on its own construction, it was in a high temperature zone. Yet despite this, it not only showed no special responsibility, it behaved irresponsibly.

304. SAA, as we noted earlier, argued that it could not simply abandon the override scheme, as this would have been irresponsible. This appears to be based on the assumption that without the override scheme in its post October form, there would be no other means of compensating travel agents short of developing a new form of compensation requiring complex negotiations with the industry. That, according to the evidence of Ms Harris, they were in the process of doing. Notwithstanding these negotiations, by the conclusion of these hearings no new model had been arrived at.

305. Had an attempt to reinvent agents’ compensation been a response to allegations that the current practice was possibly a contravention of the Act, this would have been laudable and influenced the quantum of the fine favourably. But this is not the evidence.

306. Neither Mr Viljoen nor Ms Harris suggested that the reason for the negotiations arose from these concerns. Harris testified that two years ago ASATA and SAA had formed a remuneration team to look at alternative remuneration for travel agents because:

“..from a global perspective travel agency commission was being done away with and the entire remuneration model was being altered.”¹⁶⁸

307. Thus the negotiations appear to be commercially, not compliance, driven and therefore cannot avail SAA as mitigation. The length of time this has taken would seem to bear this out. By way of contrast the evidence shows that SAA, in a matter of months, implemented the new “challenging” override schemes and as the record of the correspondence shows, against the wishes of the travel agents.

308. Further had SAA engaged the Commission at the time and asked for its input on this matter this evidence would have been more credible. Nor indeed is it particularly impressive that SAA has taken more than five years since the lodging of the complaint to arrive at a new compensation scheme for travel agents.

309. It appears that SAA was not prepared to abandon the override scheme at any time soon despite serious questions over its legality nor did it show any caution in this regard.

310. We therefore find that the behaviour of SAA in the market place since the complaint does not warrant any mitigation of the fine but on the contrary aggravation.

The market circumstances in which the contravention took place.

311. We know from the evidence of Viljoen that the October changes to the override scheme were introduced at a time when SAA was losing market share to rivals and had embarked on a number of strategies to rectify that situation, amongst which was the more aggressive nature of the override scheme. Other changes introduced at the time were certainly pro-competitive and even if they hurt rivals that would be part of competition on the merits, which the Act seeks to encourage not suppress.

312. What is regrettable is that in response to a loss of market share, SAA did not confine itself to pro-competitive responses, but included conduct that was to have an anti-competitive exclusionary effect on rivals. That anti-competitive conduct particularly impacted upon the business of the newest and fastest growing entrant at the time, Nationwide.

313. We find that the prohibited practice was introduced at a time when the dominant firm was losing market share, that a new entrant was showing promising growth and hence served not only to attempt to regain market share by competition other than on the merits, but also to stifle the entry of a new rival and contain an existing one. Recall, Holt’s evidence was that prior to the relevant period, Nationwide and BA/Comair had grown more rapidly than the market as a whole and that afterwards they had declined and their growth was less than that of the market.

314. We must also bear in mind that the domestic airline market is central to the life-blood of our economy. Were SAA’s abuse of dominance successful and one or more of its domestic airlines exited, we would have faced the prospect of a duopoly or even a monopoly on the major domestic routes. This is a market littered with the corpses of failed entrants. Had more firms failed as a result of the override scheme, the prospects for other new entrants would have been even bleaker.

315. The market circumstances therefore aggravate the conduct of SAA.

Level of profit derived from the contravention

316. The Commission concede that this is impossible to establish and we agree with them. We can make no finding in this respect.

317. The degree to which the respondent has co-operated with the Competition Commission and the Tribunal

318. We now deal with whether SAA has co-operated with the Commission. The Commission accuses SAA of adopting tactics throughout the course of this litigation that were designed to delay or obstruct the hearing of the matter to its finality. The Commission cites a number of instances.

319. It states that SAA first embarked on a High Court review of the complaint referral, a review that was later abandoned, but delayed further prosecution of the matter by one year. It then complains of an incident where it alleges compliance with one of the Tribunal’s orders for discovery of documents was deliberately frustrated by Mr Chavarika, who had, wrongly, it is now common cause, alleged that certain documentation was not in existence. The documents were subsequently discovered by SAA after the Tribunal had ordered that an affidavit be produced, confirming this be produced from Viljoen. On the contrary, what was produced was an affidavit, this time from Mr Viljoen to state that they did exist and the documents were then discovered.

320. Then the Commission cites an instance in April 2004, when the matter was due to commence, when SAA applied for a postponement on the basis that it wished to consolidate the present matter with a matter that at that stage the Commission were investigating, also related to the override scheme. This Tribunal refused the postponement and SAA unsuccessfully took the decision on appeal to the Competition Appeal Court which dismissed the application imposing a costs order.

321. The Commission has also relied on the fact that Mr Viljoen, whose cross-examination could not be concluded during the first period for which the matter had been set down, had failed to appear at a later date arranged for his convenience and thus the matter had to be postponed again.

322. The Commission also catalogues various other instances, but we will not burden this decision further by mentioning each one. SAA for its part alleges that it has a valid explanation for each instance of apparent delay. Thus in explaining the review, SAA says that the law on this point was uncertain and it was entitled to pursue this action until the Supreme Court of Appeal Court had decided otherwise, at which point it abandoned the review. On the application for postponement SAA alleges that it has been vindicated by time as that case has now been referred to the Tribunal i.e. subsequent to the CAC determination.¹⁶⁹

323. SAA explains Chavarika’s apparent difficulty in obtaining the documents that Viljoen was later to procure, as a bona fide error.¹⁷⁰SAA also goes on the attack and points out, correctly, that some delays have been at the instance of the Commission.¹⁷¹

324. Each of these incidents of delay at the instance of SAA, taken in isolation, has been accompanied by a reasonable explanation. What remains for us to decide is whether taken cumulatively, we can come to the conclusion that SAA has deliberately embarked on a strategy of delay. In the absence of this point being put to an SAA witness in cross-examination, we must be careful of coming to such a conclusion interesting as it might have been to hear them in reply.

325. What we can conclude is that SAA has certainly done nothing to indicate it wanted to help the Commission in its investigation nor to expedite the matter nor to reduce the issues in question.

326. What SAA did do was to litigate this matter to the last. No point was conceded or not taken. Even on issues in respect of which it was plainly unable to come up with a credible alternative version, it resolutely refused to concede an inch. Recall Viljoen’s long war of attrition with the Commission over market share figures that we detailed earlier. That, of course, it is fully entitled to do for to draw an adverse inference from such conduct may risk chilling litigants fully exploring their rights. However such an approach means the litigant cannot rely on sub-section 59(3)(f) as a means of mitigating a fine. While we do not regard the conduct as having aggravated the fine, as we give SAA the benefit of the doubt, there can be no benefit to them either.

327. Whether the respondent has previously been found in contravention of the Act

328. The respondent has not previously been found in contravention of the Act.

Conclusion

329. The Commission has sought a fine of R 100 million. This constitutes 5% of the base figure of R 2 billion that we referred to earlier. SAA’s response is that this figure, if a fine is deemed appropriate, is exceedingly high. It states that in an internal memorandum to its executive committee the Commission’s enforcement and exemptions department had recommended that the fine be set at 0,5% of turnover. This, says SAA, is one tenth the fine sought now and was recommended at a time when the Commission had completed its investigation and had a complete conspectus of the whole case.

330. We do not have this document in our record, but assuming that we can accept that such a document exists its relevance escapes us. What the Commission may have at one time discussed internally has no bearing on what we have to consider today, namely, what is an appropriate level of penalty on the evidence before us. It is that issue at the end of the case that the Commission has addressed in its recommendation, what one of its departments may have said once upon a time long before this case ended, is irrelevant.

331. SAA has also raised as an additional mitigating factor its financial plight at the time argument was heard. We do not need to decide whether such circumstances justify the mitigation of an administrative penalty. The evidence has been that in the past when it was in financial straits, the State, its shareholder, has come to its aid. We have had no regard therefore to SAA’s financial circumstances.

332. SAA also belatedly raised a constitutional point that we were not entitled to impose an administrative fine on a respondent where the standards of proof of a criminal trial are not observed.¹⁷² SAA states this violates its rights to just administrative action in terms of section 33 of the Constitution.

333. Section 33(1) states:

“Everyone has the right to administrative action that is lawful, reasonable and administratively fair.”

334. The Competition Appeal Court in the Federal Mogul case, where a similar constitutional point was taken by the appellant, has decided that the administrative fine contemplated by section 59 is civil in nature as its purpose is not to punish criminals rather its context is corrective and non-criminal in nature. ¹⁷³

335. SAA attempts to escape this difficulty by asserting that the CAC reasoning was flawed. Since we are bound by CAC decisions this argument does not assist SAA in this forum. Secondly, they argue that the CAC was not dealing with a challenge to section 59 in terms of section 33 of the Constitution and hence this point is not yet decided. Besides this bald assertion the argument is taken no further and hence it is difficult to predict what form it might take that makes the issues novel. Whilst is correct that the CAC decided the matter on the basis of section 35 of the Constitution and not section 33 it is difficult to believe that the CAC would come to any other conclusion since they stated expressly that the procedure must be fair, yet upheld the imposition of a penalty in that case where the standard of proof was civil and not criminal.¹⁷⁴ We accordingly find the constitutional argument devoid of merit.

336. The remaining part of the constitutional argument was raised in oral argument and not in the heads of argument. SAA argued that the Tribunal could not impose an administrative penalty on a respondent where evidence led in relation to the penalty was by way of affidavit not viva voce testimony. SAA argues this way because we allowed the Commission and SAA to give evidence by way of affidavit and not oral testimony. The Commission furnished the initial affidavit to which SAA was able to answer. SAA argues that it would be unfair to impose an administrative fine on a respondent relying on untested evidence.

337. Whether this point is good is not something we have to decide. As our approach to the consideration of the factors indicates, we have not merely relied on the affidavits on remedies to come to our conclusion and indeed we have considered the viva voce evidence as well. Where the affidavits have raised disputes of fact we have either accepted SAA’s version or, as is the case with the dispute over whether SAA co-operated with the Commission, given SAA the benefit of the doubt. Accordingly the constitutional argument, whilst interesting, is of no application in the present case.

338. Our discussion of the various factors above has indicated that we have found that the conduct was, on these facts, of short duration, albeit of a serious kind. In a market with high barriers to entry and a long history of failed entry, conduct of an exclusionary nature by the dominant incumbent must be viewed seriously indeed. The behaviour of SAA in seeking not to minimise the effect of its conduct in circumstances when it knew or ought to have known it was operating at the cusp of legality is an aggravating factor. Nor can we ignore either the centrality of a competitive domestic airline industry to our economy and the fact that if the exclusionary conduct was successful it could have led to a monopoly or duopoly on certain domestic airline routes. Nevertheless, we have found that no competitors were forced out of the market although we consider them to have been adversely affected albeit to an extent unknown. The effect on ultimate consumers would have been adverse, but again we cannot determine the extent. We can make no conclusions on the level of profit made by SAA as a result of the prohibited practice. We cannot find any evidence in aggravation of the respondents’ co-operation with the Commission and Tribunal but it has not established any mitigating evidence either. We regard the market circumstances in which the abuse occurred as an aggravating factor.

339. All these factors point to us considering the Commission’s recommendation as being too high, but SAA’s 0,5% suggestion as too low. Our view is that an appropriate level would be 2,25 %, which amounts, on a base figure of R 2 billion, to a penalty of R 45 million.

340. We indicate below how we have come to our finding on the basis of our factual and legal conclusions.

Method used to set the level of the penalty

341. We first took the factors we have to consider in terms of section 59 and gave them a weighting in terms of one another so that they add up to 10%, the maximum permissible level for a fine in terms of section 59(2). We explain what we took into account in performing the weighting in Table 4 below. Then, having established these weightings, in Table 5, we applied them to our findings in the case and gave them a score relative to the total weightings in Table 4. This has served as the basis for the administrative penalty that we have imposed.

342. Note that the weightings are not just a matter of adding up strikes against a respondent. Where we find mitigation, we would credit the respondent with a score, again weighted by reference to Table 4. Thus we score co-operation with the authorities at 1,5%. A firm that failed to co-operate could find its fine increased by this percentage or part thereof. Conversely a firm that co-operated could find that its fine is discounted by this amount or part thereof.

343. Before we consider the tables, a few cautionary words are necessary. The approach that we have adopted in calculating the size of the administrative penalty attempts to lend rationality to an important decision. While the Act specifies a non-exhaustive list of factors that are to be taken into account, it does not weight these in any way. In general, and in contrast with many other jurisdictions, our legal system does not have clearly developed or widely used sentencing guidelines. It is our view that the size of the administrative penalty be argued and determined with as much attention to evidence and rational argument as the merits of the case itself and, to this extent, at least the approach adopted here is intended to act as a guideline for the future. However, further experience with the Act may indicate that either the weightings are inappropriate or that we have not exhaustively considered all the factors that may exist. In this decision we have set out our thinking in some detail in order to assist readers to understand how we approach the difficult task of allocating to legal and factual conclusions a rating that can inform the size of the penalty.

TABLE 4 RELATIVE WEIGHTINGS

Factor Percentage a) nature, duration and extent of contravention 3 % This factor is given the highest weighting. Firstly, because it deals with three separate issues, nature, duration and extent and thus as a matter of quantity it is the most wide ranging of the factors. Secondly, it needs to be weighted heavily enough to provide for a meaningful distinction between various types of contravention. Duration, for instance could refer to the act being perpetrated over a period of months or years. It is also important to differentiate sufficiently between types of prohibited practice. For instance, a section 4(1)(b) prohibition, the so-called hard-core cartel, would be considered the most egregious form of conduct and so would receive a higher allocation than a less serious form of prohibited conduct e.g. resale price maintenance. b) loss or damage as a result of contravention 1,0 % Here we would look at loss or damage to competitors and/or consumers as a result of the prohibited practice. This receives a lower weighting as the competitors/consumers can recoup this loss through a claim for civil damages. c) behaviour of respondent 1,0 % This deals with the behaviour of a respondent firm in relation to the market i.e. consumers and competitors as opposed to how it responds to the regulators which falls under sub-paragraph (f). This factor must be weighted sufficiently high to serve as both an aggravating factor for respondents whose behaviour in the market justifies, but on the other hand, is there to provide mitigation to those who attempt to redress the adverse effects of their conduct. d) market circumstances 1,0 % Here we deal with what the nature and dynamics of the market are at the time of the contravention. We examine here the type of market, its structure and history. We look at how materially the conduct impacted or could have impacted on the market structure. e) level of profit derived 0,5 % Here what we are dealing with is made quite specific. Nevertheless evidence of the level of profit derived as a result of the contravention is difficult to prove in practice and for this reason the factor, while not unimportant, is not given a high weighting. f) degree of co-operation with CC and CT 1,5 % This factor is given a high weighting because of the importance we attach to co-operation with the regulators. Those who co-operate should be able to score well in mitigation of the penalty whilst those who have not, should be penalised. g) found in previous contravention 2 % This factor needs a high weighting as a repeat offence is very serious and needs to be adequately deterred. Total 10%

344. In Table 5 we apply the above weightings to the facts of the case and come up with an appropriate level for the penalty. Note that in respect of some factors, where the evidence is not established, SAA is not penalised and received a nil allocation. However since we have found no mitigating evidence, it has not received any deduction.

TABLE 5 – APPLICATION TO SAA CASE

Factor Percentage SAA allocation a) nature, duration and extent of contravention 3 % (0,75%) Here we first take into account that an exclusionary abuse is not as serious a form of conduct as a hard- core cartel. The worst exclusionary abuse might therefore qualify for a weighting of 1,5% depending on the extent and duration which we then look at. Here the duration was found to be for a limited period. Bearing in mind the extent, a score of 0,75 out of a maximum revised total of 1,5% is appropriate. b) loss or damage as a result of contravention 1 % (0,25) Here we could make no finding on direct loss to consumers and we found some damage likely, but not quantifiable to competitors, but found that competitors had not been forced out of the market. c) behaviour of respondent 1,0 % (0,5) The respondent’s failure to ascertain whether its behaviour was unlawful in circumstances it should have, or to correct it over a period of almost five years, is a factor in aggravation. Found no mitigation in relation to negotiations for a new compensation scheme. d) market circumstances 1,0 % (0,75) Found that this was a strategic market with history of high barriers to entry and early exit. Found that if abuses had succeeded in excluding competitors it would have had very serious impact on the structure of the market. e) level of profit derived 0,5 % (0) No evidence on this hence 0% f) degree of co-operation with CC and CT 1,5 % (0) No finding in aggravation or mitigation, hence 0% g) found in previous contravention 2 % (0) No evidence on this hence 0% Total 10% (2,25)

ORDER

1. SAA is ordered to pay an administrative penalty in the amount of R 45 million (forty five million rands) to the Commission within 20 business days of this decision.

2. We declare the following conduct of SAA to be prohibited practices in contravention of section 8(d)(i) of the Act:

- the scheme known as the override incentive scheme, being a contract between itself and various travel agents between October 1999 and May 31,2001; and

- the scheme of travel agents’ compensation known as Explorer, from a date unknown until May 31 2001.

Costs

There is no order as to costs.

________________ 28 July 2005

N. Manoim Date

Concurring: U. Bhoola, D. Lewis

For the Commission: W. Pretorius, instructed by Roestoff, Venter, Kruse Attorneys

For the respondent: Adv. A. Subel S.C. and R Bhana, instructed by Knowles Hussain Lindsay Inc on behalf of Edward Nathan and Friedland

---