Nationwide & Comair v SAA
17 Feb 2010
IN THE COMPETITION TRIBUNAL
THE REPUBLIC OF SOUTH AFRICA
CASE NO: 80/CR/SEPT06
In the matter between
NATIONWIDE AIRLINES (PROPRIETARY) LIMITED Complainant
COMAIR LIMITED Intervening Complainant
SOUTH AFRICAN AIRWAYS (PROPRIETARY) LIMITED Respondent
_________________________________________________________________________Panel : Y Carrim (Presiding Member), L Reyburn (Tribunal Member),
and M Holden (Tribunal Member)
Heard on : 17,18,19,20,25,26,27,28 March 2008; and 2,3,5,6,9,10,11,12,13,16
18,19,20 March 2009; and 27,28 May 2009
Reasons issued : 17 February 2010
REASONS & ORDER
SAA’s incentive schemes with travel agents had their genesis sometime in the 1990s. Prior to 1999 it appeared that South African Airways utilised an incentive scheme (contained in agreements with travel agents) whereby it paid a standard commission of for example 7% to travel agents in respect of each ticket sold. We refer to these agreements as the first generation agreements.
During late 1999 SAA introduced new incentive agreements (the second generation agreements). These agreements together with SAA’s Explorer scheme became the subject matter of the first Nationwide complaint.
Sometime in 2001 SAA amended its incentive scheme on the basis of advice it had obtained. It did away with the Explorer scheme, and introduced new override incentive agreements and trust agreements. These override agreements (third generation agreements) and the trust agreements (collectively referred to as the 2001 incentive schemes) remained in the marketplace until 31 March 2005.
Both the second and third generation agreements provided for a commission structure which rewarded travel agents to achieve targets set by SAA, by revenue or volume, on an override basis. These override commissions were paid over and above the standard commission for each ticket sold. A more detailed description of these agreements is dealt with later in these reasons. Another feature of these agreements was that they were negotiated individually with travel agents. Hence each travel agent had its own targets to meet. At the same time the duration and commencement dates of these agreements differed from travel agent to travel agent.
In Competition Commission v SAA the second generation agreements together with the Explorer scheme were found to be in contravention of section 8(d)(i) of the Competition Act. In that case – which we refer to as the Nationwide case/decision – the Tribunal found that this scheme induced travel agents not to deal with SAA’s rivals in the domestic airline travel market and that the foreclosure of its rivals in that market was likely to be substantial. The relevant period identified by the Tribunal in that matter was September/October 1999 – 31 May 2001.
We are required to evaluate SAA’s 2001 incentive scheme, consisting of third generation override agreements and trust agreements for anti-trust scrutiny in the context of market conditions prevailing in the domestic airline market at that time. The most significant recent developments relevant to that period were the launch of low cost carriers such as Kulula and the growth of other distribution channels used by airlines such as the internet and direct sales (call centre or corporate agreements).
After considering the evidence and arguments in these proceedings we have concluded that SAA’s 2001 incentive scheme was in contravention of section 8(d)(i) in that it induced travel agents to deal with SAA at the expense of its rivals and led to foreclosure of the rivals in the market for scheduled domestic airline travel. We have found that the two relevant markets for this period are the market for travel agent services to airlines and the market for scheduled domestic air travel. We have found SAA to be dominant in both markets. Despite the recent developments in the domestic airline market, such as the launch of low cost carriers and the growth of alternative distribution channels, we have found that travel agents still constituted the most significant and optimal route to market for domestic airlines. While low cost carriers accounted for most of the growth in the domestic airline travel market, we have found that during the relevant period, the market for these developments did not warrant market segmentation into a low cost/time insensitive/price sensitive market and time sensitive/price insensitive market. While the effects of SAA’s conduct may have had a greater impact on that segment of the domestic airline market distributed through travel agents, we have concluded that the conduct had the effect of reducing competition in the total domestic airline market.
Note on nomenclature
For purposes of convenience we have adopted the references utilised by Comair in these proceedings in order to distinguish between the various species of SAA’s incentive agreements. The salient differences between the second generation and third generation agreements are discussed below. It is important to signal here that the nomenclature speaks to the general features of these agreements.
We have also included for the sake of brevity a summary of the procedural background to this matter as Annexure 1 to these reasons. However for the sake of convenience we set out in brief how this matter came to be finally heard.
In the Nationwide case, the complainant before the Tribunal was the Commission, with which a complaint had been lodged by Nationwide and other companies in the Nationwide Group in October 2000. The Commission had in the normal way made its own investigations before bringing the matter before the Tribunal. Nationwide and its associated companies in the Nationwide Group were not participants in the case. In that case the Tribunal imposed an administrative penalty of R45 million on SAA in respect of its anticompetitive conduct.
In the present matter the complainants are Comair and Nationwide, which filed separate complaints that were consolidated by an order of the Tribunal dated 7 November 2007. Comair lodged its original complaint on 13 October 2003 and was granted leave on 2 April 2006 to intervene and participate in the case after it had been referred to the Tribunal by the Commission on 12 October 2004. Nationwide was granted leave on 25 May 2006 to intervene and participate as an intervenor in the same matter. But Nationwide also lodged its own second complaint (not to be confused with the first complaint which gave rise to the Nationwide case mentioned above) on 22 May 2006. The Commission declined to refer this second complaint to the Tribunal, asserting in effect that it was otiose in view of the Nationwide decision, but Nationwide then referred its second complaint to the Tribunal on 21 September 2006 in terms of Section 51(1) of the Act.
An abrupt turn in this forensic labyrinth occurred when the Commission ceased on 4 December 2006 to have a role in the case referred to the Tribunal as a result of Comair’s original complaint. On that date the Tribunal confirmed a settlement agreement which had been concluded by SAA on 24 May 2006 with the Commission regarding Comair’s complaint. The Tribunal’s confirmation was given in terms of section 49D(1) of the Act. SAA agreed to pay the Commission an administrative penalty of R15 million as part of the settlement, but with no admission of liability. An undertaking was given by SAA in the settlement document that its agreements with travel agents no longer contained provisions of the kind which had been the subject of the complaints lodged by Nationwide and Comair, and SAA undertook not to conclude agreements in future with local travel agents containing a number of specified exclusionary provisions. SAA also initiated a compliance programme for its managers and other personnel which it had prepared with the assistance of the Commission. Comair and Nationwide opposed the confirmation of the consent order but their objections were dismissed by the Tribunal.
Following the decision of the Tribunal confirming the settlement agreement, and apparently in order to comply with an imperative attributed to that decision that a fresh start be made with its complaint (in the knowledge that the Commission would not take part in further action against SAA in relation to it), Comair filed an application on 13 March 2007 on notice of motion to the Tribunal under Section 49D(4)(a) of the Act for a declaration that the conduct of SAA which it had sought to impugn in its original complaint was a prohibited practice in terms of the Act. Comair’s original complaint was accordingly replaced by the notice of motion. (The original complaint, along with the complaint Nationwide had lodged in its role as an intervening party, had been in effect extinguished by the confirmation of the consent order.)
The prospect of two parallel cases dealing with similar subject-matter was averted when a consolidation of the cases was ordered by the Tribunal on 7 December 2007. The cases so consolidated were, of course, Comair’s application on notice of motion, dated 13 March 2007 and Nationwide’s self-referred complaint of 21 September 2006.
Although the complaints as consolidated contained prayers for other relief, what Comair and Nationwide effectively seek in these proceedings is a declaration by the Tribunal in terms of Section 49D(4)(a), read with Section 58(1)(a)(v) or (vi) of the Act, that incentive schemes implemented by SAA with travel agents in the period 1 June 2001 to 31 December 2005, comprising forms of ‘override’ payments specified in agreements in force with travel agents, in this period, and so-called ‘trust payments’ which were also paid to travel agents in terms of agreements in force in the same period, were exclusionary and amounted to prohibited practices in terms of section 8(d)(i), alternatively section 8(c), of the Act.
The other relief sought originally by either or both of Comair and Nationwide included declarations that the relevant agreements setting up the impugned incentive schemes were void, and interdicts prohibiting the continuation of the allegedly exclusionary practices represented by these schemes. However, the opportunity to claim this relief fell by the wayside because of the effect of the consent order referred to above, since it has been clear from the settlement agreement that SAA has no extant agreements with travel agents containing the provisions in question. The formal abandonment of claims to these forms of relief was confirmed during the hearings and in the closing arguments of counsel for the two complainants.
At one stage in the hearings Nationwide’s counsel, Mr Gotz, sought to persuade the Tribunal that it should impose a further administrative levy on SAA in respect of transgressions of the Act the period following the confirmation order referred to above. No such relief had been proposed in Nationwide’s particulars of claim. However, that issue was not followed up in Nationwide’s heads of argument and we do not propose to deal with it in this decision as it was not properly before the Tribunal at the time of the hearing. However, the complainants continued to seek an order of costs against SAA.
SAA opposed the relief sought, asserting that its incentive arrangements as in force at the relevant time were lawful. Further, SAA contended in an objection in limine that the complaints of Comair and Nationwide should be dismissed because of the provisions of Section 67(2) of that Act, which states in effect that a complaint cannot be referred to the Tribunal if the respondent has been the subject of prior completed proceedings in the Tribunal relating substantially to the same conduct. We deal below with the objection in limine.
The practical utility to Comair and Nationwide of the declaration sought in these proceedings is that such a declaration is a prerequisite in terms of Section 65(6)(b) of the Act to the institution of an action in the High Court for damages flowing from the anticompetitive conduct ruled by the Tribunal.
What falls to be decided in this case are therefore the point in limine, which was argued at the outset of the hearing but not decided at the time, and unless the point in limine prevails, the merits of Comair’s and Nationwide’s requests for declaratory orders holding that SAA’s incentive schemes with travel agents in the period 1 June 2001 to 31 March 2005 amounted to exclusionary restrictive practices that represented abuses of dominance in contravention of the Act.
The summary set out above of events preceding the hearing of this case is brief and possibly simplistic. For readers who wish to study the relevant history more systematically, Annexure 1 to this decision has been compiled, in the form of a chronology.
The hearing took place over a number of days stretching over a year. The first part of the hearing was held in March 2008 and the latter part in May 2009. Argument was presented on 27 and 28 May 2009. Comair led two factual witnesses Erik Venter, Joint CEO of Comair Limited and Conrad Mortimer, the Commercial Director of Tourvest Holdings (Pty) Ltd. It also led an expert witness Dr Giulio Federico from CRA International. Nationwide led one factual witness Mr Vernon Bricknell, the CEO of Nationwide Airlines (Pty) Ltd, and an expert from Oxera Consulting Limited. SAA led its factual witness Mr Andries Viljoen, the erstwhile President and Chief Executive of SAA, Ms Harris of Rennies Travel and the expert Dr Affuso from RBB Economics.
The hearings were marked by an unusual number of graphs and exhibits, often produced overnight by the parties’ expert witnesses. In the course of the proceedings the Tribunal cautioned the parties that it had accepted those exhibits at that time without making any findings on their probative value and that decisions pertaining to that, if any, would be made on proper reflection of the evidence put before us.
Comair and Nationwide allege that SAA’s 2001 scheme was a continuation, albeit with a few differences, of the earlier scheme. It was designed to induce travel agents not to deal with SAA’s competitors and had an anti-competitive effect on its rivals in that it had foreclosed them from the domestic airline travel market. They argued that despite recent developments in the domestic airline travel market travel agents still constituted the single most important distribution channel for airlines. Such inducement constituted exclusionary conduct and had impeded them from expanding in a segment of the domestic air travel market and amounted to a contravention of section 8(d)(i), alternatively 8(c), of the Competition Act.
Comair argued for a segmentation of the domestic airline market into a time sensitive (TS) and non time sensitive (NTS) market. Nationwide was agnostic on the segmentation of the relevant market but agreed with Comair that because of recent developments in the market the anti-competitive effects of SAA’s exclusionary conduct would be more acutely felt in the TS or price insensitive (PI) segment of the market. This did not mean that the effects were not felt in the other NTS or price sensitive segments of the market. Both argued that SAA was still dominant in the relevant markets.
SAA raised and persisted with a point in limine that the provisions of section 67(2) precluded Comair and Nationwide from proceeding with this prosecution. It argued further that even though it was presumptively dominant in the market for sales of airline tickets through travel agents and the market for scheduled domestic airline travel, it did not enjoy market power in these markets. The market was not segmented along the lines argued by Comair and was, either wide enough to include all airlines and segments or as narrow as a single route. Some of these arguments had been previously raised by SAA and had been dismissed by the Tribunal in the Nationwidecomplaint. SAA argued further that its share of the total airline travel market was declining while those of Comair and Nationwide increasing. This demonstrated that its conduct was not exclusionary and had not impeded its rivals from growing in the market.
In order to assess SAA’s point in limine we are required to determine the relevant period of this enquiry. On 18 October 2000, Nationwide, filed a complaint with the Commission alleging that the respondent had contravened the Competition Act. The Commission investigated this complaint and seven months later filed a complaint referral with the Tribunal. In that complaint referral, what we have termed the Nationwide complaint, the Commission made it clear that it had not referred all the issues that were contained in the original complaint but only those that related to compensation for travel agents the so-called override scheme and the Explorer Scheme. The period of the complaint referral was confined to 1 September 1999 to April 2001. On 9 October 2003, Comair filed a complaint with the Commission alleging that the respondent had contravened sections 8(c) and 8 (d)(i) of the Act by engaging in exclusionary practices. The exclusionary practices alleged included the use of the override scheme and something else referred to as ‘trust payments’ to travel agents in terms of which travel agents receive a lump sum at the end of SAA’ s financial year based on the agents sale of SAA tickets. The Comair complaint related to the period of 1 September 1999 to date. It thus overlapped with the period of the Nationwide complaint a period of some 18 months. Comair’s complaint was referred to the Tribunal in October 2004 and has now been transformed into this application under section 49D(4)(a).
The Nationwide complaint referral was decided by the Tribunal in May 2005 in Competition Commission v SAA (the Nationwide case). The relevant period of that complaint was September 1999 to 31 May 2001. Because SAA’s incentive scheme consisting of the second generation agreements and Explorer programme during the period September 1999 to 31 May 2001, had already been decided by the Tribunal, in these proceedings we are concerned only with SAA’s incentive scheme in the marketplace from a date commencing on 1 June 2001. It was common cause that the third generation agreements and trust payments were in the force until 31 March 2005. The relevant period for this matter is accordingly 1 June 2001 to 31 March 2005.
Because we are concerned with examining SAA’s conduct articulated through a clutch of agreements concluded individually with travel agents possibly with different periods of validity one can expect a small degree of overlap between the agreements under the period of the Nationwide case and those under consideration here. Some agreements may have terminated prior to that end date and others after that date. Some agreements may have been concluded prior to the start date and others after.
For example we were told that SAA had amended the basis of computation post base in the second generation agreements from 1 April 2001 the start of SAA’s financial year and that this was in place until 31 March 2005. SAA had also introduced trust payments. However we could not ascertain whether all these agreements commenced precisely on 1 April 2001 or anytime soon thereafter. By way of illustration, CRA’s Table A2 in Appendix A of their report lists SAA’s domestic override agreements with the main travel agents is footnoted as follows: “2001-02 contracts for these agents were not found in the discovery materials (however contracts...for a number of smaller agents...were discovered)”. Added to this was the difficulty created by the fact that SAA did not have an agreement with some agents in a particular year but did so in subsequent years. A further complication was created by the merger of travel agents and SAA having agreements with only one of these in a particular year. This was compounded by the fact that often agreements were signed long after their actual commencement dates.
While the precise commencement dates of these agreements would be highly relevant to aligning the agreements to a case concerning damages or for computation of payments by SAA to travel agents, these difficulties do not preclude us from evaluating the economic effects of the nature of these agreements from an anti-trust perspective for a period commencing after 31 May 2001. In an anti-trust enquiry the commencement and duration of these agreements would be relevant only for the purposes of defining a particular market context against which these provisions can be assessed. Hence the precise date of commencement of each agreement is not as much a concern to us as is the conduct that flows from those provisions during an identifiable period in time.
In our view it would be sufficient for our purposes – namely to assess the anti-competitive effects of SAA’s conduct articulated through its third generation agreements and trust payments from a start date that is reasonably aligned to the evidence placed before us from a period commencing after 1 June 2001. We do know, from SAA’s own account that it intended to amend its second generation agreements from 1 April 2001. Since these agreements were negotiated individually it would be reasonable to assume that not all of these were concluded on 1 April 2001. Comair contends that agreements with major travel agents were concluded by May 2001. However it appears from the Nationwide case that some aspects of the Explorer programme were still in place during May 2001. By all accounts, it seems that by 1 June 2001, the third generation agreements and trust payments had decidedly become the incentive scheme that SAA had with all major travel agents. These agreements constituted SAA’s incentive scheme with travel agents until 31 March 2005 and are the subject of these proceedings.
Certainly there may be some degree of overlap in the two complaint referrals where some third generation agreements might have been concluded with travel agents prior to the 31 May 2001 date. Likewise there may be overlaps with second generation agreements that may have persisted after the 31 May 2001 date. And it may also be that one or two third generation agreements remained in force for a brief period of time after 31 March 2005. But these overlaps were of a very limited duration and would not have a material impact on our competition analysis for the period 1 June 2001 to 31 March 2005.
Point in limine
In its answering affidavit, SAA contended that the Nationwide decision and the complaint referred to by Comair on 13 October 2004 related substantially to the same conduct and that the issues raised in the Comair referral have already been adjudicated by this Tribunal in the Nationwide decision. The Nationwide complaint and referral accordingly constituted completed proceedings for the purposes of section 67(2) of the Act and the present Comair and Nationwide referrals were incompetent under that section.
In its heads of arguments SAA also contended that the consent order granted by the Tribunal in the Comair referral constituted completed proceedings. This point was never pleaded in its papers. At the hearing, Mr Subel on behalf of SAA indicated that this point was no longer being pursued by SAA. Since that point was no longer pursued, the issue which falls to be determined concerns whether or not the conduct complained of in casu relates to substantially the same conduct as the matter which has been adjudicated by the Tribunal in the Nationwide decision.
SAA argued that in the Nationwide decision the scheme considered by the Tribunal consisted of nothing else than agreements that SAA had concluded with travel agents in which it offered them incentives for the sale of airline tickets. The override agreements under consideration in this case (“third generation agreements”) and trust agreements constituted nothing more than an incentive scheme with travel agents for the sale of airline tickets. Hence this scheme was substantially the same as the previous scheme. For this reason section 67(2) rendered it incompetent for Comair and Nationwide to refer these agreements as a subject of a complaint to this Tribunal. It mattered not whether similar agreements were in place subsequent to the Tribunal’s decision in the Nationwide case. The fact that it had already been prosecuted for such agreements meant that it was entitled to seek the protection against double jeopardy afforded by section 67(2).
SAA also argued that in Nationwide case the Tribunal had not temporally confined its decision to 31 May 2001, but had relied on 18 May 2001 date as a matter of convenience and that the Tribunal’s decision had extended into the period post May 2001 until 2005.
On the one hand SAA argued for an a-temporal interpretation of section 67(2), on the basis that the section only refers to conduct and makes no reference to time. Hence conduct that was substantially the same was protected from prosecution by section 67(2) irrespective of when such conduct occurred. On the other hand it argued that theNationwide decision dealt with a relevant period that went beyond May 2001 into 2005. Hence the “substantially the same conduct” occurred over the entire period from 1999 to 2005. On either basis Comair and Nationwide were precluded from bringing this application.
Section 67(2) states that -
“a complaint may not be referred to the Competition Tribunal against any firm that has been a respondent in completed proceedings before the Tribunal under the same or another section of this Act relating substantially to the same conduct.”
Comair submitted that the test in section 67(2) should not be whether the conduct alleged in the two complaints is “substantially the same” but that what must be shown is that there is substantial overlap between the complaints in respect of the “same conduct”.
The Competition Appeal Court and the Tribunal have both previously provided us with guidance on how to approach section 67(2). In Sappi Fine Paper (Pty) Ltd v The Competition Commission, the Competition Appeal Court stated that: –
“The Legislature enacted the relevant provisions to avoid a firm being “tried” twice for the same or substantially the same conduct. Put differently the aim of the Legislature in introducing section 67(2) was to avoid “double jeopardy”.
In Barnes Fencing Industries (Pty) Limited and Another v Iscor Limited (Mittal SA) and Others,the Tribunal, in an application for intervention stated: –
“In our view, on these facts, the complainants have established an interest not adequately represented in this case and they would be prejudiced if they were not allowed to intervene, as a separate complaint referral since it would be based on the same or similar conduct to the one in casu, would be vulnerable to objection in terms of section 67(2).”
Section 67(2) was clearly enacted to avoid a firm being tried twice for the same or similar conduct. If a respondent has already been prosecuted for certain conduct, it ought not to be prosecuted again, whether or not the earlier prosecution resulted in an adverse finding. A respondent is entitled to seek the protection of this Tribunal against repeated prosecutions for the same conduct. It would seem to us that a respondent who wishes to rely on the protection of section 67(2) would bear the onus for alleging and proving this. Since this was not argued before us we make no findings in this regard.
While the Tribunal and the CAC have both interpreted the section to include the notion of “substantially the same conduct” or “similar conduct” both have indicated that where the particulars of complaint deal with the same or similar conduct in a different time period, this would not make such a complaint vulnerable to an attack under section 67(2).
In Barnes Fencing the Tribunal stated –
“It is worth mentioning what may be meant by particulars of the complaint for the purpose of understanding section 50. If the complainants in the present case had alleged other conduct that they considered discriminatory, and which the Commission had not referred, or that the same conduct contained in the referral had taken place during a different time period to the period alleged in the referral this would constitute an example of particulars not referred. Here the particulars would not be on all fours with the Commission's case, as they differed as to either manner or time and hence a separate referral would be necessary and would not be vulnerable to a successful attack under section67(2)”.
In other words section 67(2) seeks to protect a respondent from double jeopardy related to the same or similar conduct in a specified time period. For example, a respondent may have been prosecuted for abuse of dominance under section 8(c) for conduct in a specified time period. Once the proceedings have been completed, a complaint of the same conduct, occurring in the same time periods, could not be referred to the Tribunal under another section of the Act, example section 8(d)(i) or section 5. Similarly by way of example, if in those completed proceedings the issue of dominance was determined on the basis of the respondent’s market share, a subsequent complaint, for the same conduct occurring in the same time period, based on a notion of collective dominance could not be referred to the Tribunal once proceedings in the former complaint are completed.
Substantiality would thus relate to materiality and would include both manner and time. Both the CAC and this Tribunal have held that conduct occurring in different relevant time periods constitutes a material difference between two complaints.
SAA’s a-temporal approach to section 67(2), if adopted by this Tribunal, would lead to an absurdity as demonstrated the following example. Consider the matter of a cartel member who was prosecuted for a cartel which had lasted for two months. Cartels are considered to be the most egregious offences under the Act. On SAA’s interpretation, this person could now with impunity engage in any number of cartel activities after being prosecuted and found guilty for the first two month long cartel. If the Tribunal were to adopt such an approach, it would never be able to prosecute a respondent for repeated offences. Thus when a party seeks the protection of section 67(2), such protection can only be competent where it relates to substantially the same conduct taking place in a specified or defined period. Substantially the same conduct or even identical conduct occurring in a different time period would constitute new conduct and would not be protected by s67(2).
Furthermore, SAA’s suggestion that the time period of the Nationwide decision extended into 2005 is completely baseless. The Tribunal in that case expressly stated as follows –
“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.”
SAA’s conduct from 1999 to 31 May 2001 has already been evaluated by the Tribunal in the Nationwide case. This complaint is therefore concerned with SAA’s conduct after 31 May 2001, namely the period from 1 June 2001 to 31 March 2005. Even if SAA, for arguments sake, had not introduced a new incentive scheme in 2001 but had continued with the same scheme evaluated in Nationwide, its conduct in the subsequent period of 1 June 2001 to 31 March 2005 would constitute conduct that would not be protected under section 67(2) precisely because it was occurring in a different time period.
The relevant period of the Nationwide decision was October 1999 to 31 May 2001. The relevant period for this matter is 1 June 2001 to 31 March 2005. On this basis alone, we find that SAA’s approach to section 67(2) is without any merit and the point in limine is accordingly dismissed. To the extent that the Comair complaint concerns the second generation override incentive agreements and the Explorer scheme in place until 31 May 2001, SAA is protected from further prosecution by the provisions of section 67(2). However SAA cannot seek the protection of section 67(2) for its conduct occurring after 31 May 2001, even if that conduct was substantially similar in nature to conduct in the previous period.
Moreover, and contrary to SAA’s assertions, the nature of the incentive scheme under consideration in this matter, consisting of the third generation and trust agreementswas never considered by the Tribunal in the Nationwide case. In that decision the Tribunal was only concerned with the override incentive agreements (second generation agreements) and the Explorer scheme for the period October 1999 to 31 May 2001. The Tribunal took heed of the fact that the incentive scheme was possibly still in operation but pointed out that:
“...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 (our emphasis) 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...”16
We turn to consider the merits of this case.
In order to understand the third generation (3G) agreements it is necessary to re-visit the essential elements of the second generation (2G) agreements. In the second generation agreements, the incentives to travel agents were structured as follows. Travel agents were paid a flat basic commission for all sales up to a target that was set for them. The target figure was expressed in rands. If they reached and exceeded the set target they become eligible for two additional types of commission, payable over and above the flat basic commission. The first of these was a commission calculated not only over the amount by which the travel agent exceeded the target but over the total sales achieved above and below the target.
By way of example, let us assume that a particular travel agent had a sales target of R50 million to achieve. The travel agent would earn a flat commission of 7% on this volume, expressed in rand value. However if the travel agent exceeded the target by R5m it would earn an override commission, set at typically 0.5%, calculated on all sales earned namely R55m. Hence the agent at that stage would earn an additional commission of R275 000, over and above its 7% flat commission of R3 850 000. In total the agent would earn an average commission of R4 125 000 translating into an average rate of 7.5%. This is called the override incentive and because it is calculated over the total sales achieved it is referred to as the “back to rand one” principle.
Marginal incentives are an important consideration because of the impact of additional sales on the commission accruing to the agent. The impact of such agreements is that the profitability of travel agents becomes very sensitive to whether they meet the target levels of growth as well as to what extent they can earn higher commission for selling more than the baseline target.
The actual percentage of the override incentive may have differed from agent to agent but the basis of its computation was common across all. In addition to the override commission travel agents also became eligible for a third category of commission, referred to as the incremental commission. If the travel agent exceeded the first target for the override commission, also referred to as base, by a certain target it became eligible for the incremental commission. This commission was not calculated back to rand one but was calculated in relation to the first target (“back to rand base” or “back to base” principle) and was typically much higher in percentage terms than the override commission.
In some agreements the incremental commission was subject to an escalation at specified intervals, rising steeply with higher incremental targets. In the American Express agreement for example, the incremental commission kicked in when it achieved sales in excess of 15% of its base target. The commission itself starts at 14% at that milestone but rises sharply so that sales of 35% in excess of base target could earn a rate of 31%. In the Nationwide complaint the Tribunal found that these override agreements were designed to induce travel agents not to deal with SAA’s competitors and encouraged travel agents to direct ticket sales towards SAA to the detriment of the consumer.
In the course of 2001, SAA changed its override agreements. The changes were announced around the time when the Commission had almost finalised its investigation into the Nationwide complaint and just before it had referred it to the Tribunal.
The general features of the 3G agreements were as follows. Override payments were introduced for the achievement of base revenues. Base (target) revenues were usually set on the previous year’s sales of the particular travel agent and were individually negotiated with each travel agent. Payments to travel agents for achieving base were calculated on a back to rand one basis. In other words, both the basic flat commission of 7% and the override incentive commission that had been offered by SAA in the second generation agreements remained in place.
An important change introduced by SAA in the second generation agreements was that targets and computation of achievement of targets would be done on the basis of flown revenue rather than BSP figures. Flown revenue is a measurement applied across all carriers in the determinant of rebate deals. BSP refers to Billing and Settlement Plan sales which is the gross bookings by IATA-accredited travel agents. The significance of this is that while travel agents could always calculate their BSP figures through reconciliations with other relevant components of gross sales, only airlines would be in possession of the flown revenue figures. Third generation agreements in place for the first two contract years (2001/2002 and 2002/2003) were based on flown revenues and those for subsequent years on number of flown passengers. Flown revenues sold on SAX and SALwere also included in the override agreements for major travel agents for the purposes of computing performance and payments to travel agents.
The basis of computation of flown revenues however was also adjusted from year to year or agreement to agreement during this period by the introduction of a number of exclusions. The agreements for example excluded the acquisition of travel agents in the form of new outlets or new corporate accounts with in-house travel agents from the computation of growth for the purposes of meeting the set targets. Hence if a particular travel agent acquired another travel agent, opened a new outlet or acquired new corporate in-house account, revenue of SAA sales from these sources were not computed as incremental growth but were instead included in the base revenue target and the actual revenues for that agent.
Differentiated override payments depending on the class of tickets sold by the travel agent were introduced during contract year 2002/2003. The classes of tickets were differentiated between premium, sub-premium and discounted. Notwithstanding all of these changes, the basis of rewarding travel agents to achieve the set targets was still calculated on an override basis in that agents were paid commissions over and above the standard commission on a back to rand one principle for achieving targets set by SAA.
On SAA’s version the only significant difference between the second and third generation agreements was that the incremental commission structure after reaching base (post base commissions) was flattened so that the rate remained the same irrespective of the sales in excess of the base. In other words, SAA no longer paid commissions on an override basis for sales achieved by the travel agent post base. Viljoen testified that this was the reason why they had introduced trust agreements – it was in order to compensate travel agents for the loss of revenue they would otherwise had earned under the second generation post base override structure. Despite this evidence SAA’s written contracts with travel agents suggest otherwise and led to Dr Niels’ interpreting the post base provisions as override incentives. We re-visit this issue in our discussion on financial incentives. A general pattern that emerges in these agreements is that whereas SAA had previously rewarded revenue growth handsomely in the period between 1999 - May 2001 through aggressive override payments for reaching and exceeding base, by 2002/3 it sought to reward agents for achieving base, i.e. maintaining revenue targets of the previous year and rewarding agents on a different commission basis post base.
According to Mortimer, during this period targets became more difficult to meet because SAA continued to fine-tune the formula for computing incremental growth, implementing a range of exclusions as discussed above. Travel agents were put under increasing pressure to achieve these targets at the expense of their competitors and that this feature of the agreements in fact tightened over time. For example in SAA’s five year agreement with Seekers the agreement allowed Seekers to install a new ticket printer when acquiring an outlet or in-house corporate accounts and to count the flown revenues captured on that new printer as growth for purposes of the agreement. This possibility was removed under later agreements. Under these later agreements corporate acquisitions were defined according to IATA numbers and did not count for growth, irrespective of whether or not a new ticket printer was installed by the acquiring travel agent.
Over this period there was a constant revision of the computation of base revenues in order to exclude from a particular agent’s base, those SAA sales that would in any event have been made by any other player in the market including SAA itself. The agreements also became more specific over time, rewarding agents only for specified classes of tickets.
Trust agreements were introduced in contract year 2001/2002 and remained in place until the first quarter of contract year 2004/5. TRUST was an acronym adopted by SAA for “True partnership, Respect, Undivided support, Sharing of information and Training of SAA product and knowledge”. Trust payments consisted of lump sum payments made by SAA to travel agents for achieving specific revenue and market share targets and in exchange for the agent’s support of SAA. The payments were additional to the domestic overrides discussed above.
The precise formula for trust payments differed across agents and through time. The formula for the larger agents such as Renfin and Sure Travel initially provided for positive revenue growth during 2001 to 2003 and thereafter for maintenance of the flown revenue levels achieved in previous years. Trust payments for smaller agents seem to contain positive revenue targets until 2003/4. What was common to all of them was the payments were made upon the achievement of a particular target. In addition to such above terms and payments, trust agreements also included allocations of tickets to agents’ promotional tickets and marketing incentives.
A typical trust payment would have been computed on the following basis:
Table 1: Model Trust Payment calculation for Wings
Trust Amount 700,000
%Weight for Rand Amount
Dom Support 20.00% 140,000
Int Support 15.00% 105,000
Dom Rev Growth 20.00% 140,000
Int Rev Growth 15.00% 105,000
Achieving all 4 objectives 20.00% 140,000
Other 10.00% 70.00
TY LY Increase Calc Payment
Dom support 52.28% 45.22% 7.00% 20% for every 1 140,000
Int support 35.26% 33.79% 1.00% 20% for every 1 21,000
Dom Rev Growth 27.04% 20.00% 7.00% 20% for every 1 140,000
Int Rev Growth 26.22% 25.00% 1.00% 20% for every 1 21,000
70,000 Total Trust fund payable 392,000
However it appears from Ms Harris’ evidence and Renfin’s internal correspondence that SAA had developed a practice of not making written commitments in respect of trust payments and had retained a fair amount of discretion in relation to the computation thereof.
The payments made under these trust agreements were not insubstantial and were critical to travel agents because at times they determined the profitability or otherwise of the agent. Ms Harris stated in her evidence that in one year Renfin had received a payment of slightly over R20 million in Rennies Travel alone for achieving some of the targets set out in the agreement. In the case of Renfin the trust payments accounted for roughly a third of the total incentive received by it. In the case of smaller travel agents the trust amounts appear to have been even larger as a proportion of the total payment received.
Ms Harris explained further that travel agents did not always know whether or not they were eligible for payments under these trust agreements. This is because the computation was done on the basis of flown revenue and only SAA could provide travel agents with those figures. Typically, SAA and the travel agents would do a quarterly reconciliation. SAA would provide the agent with flown revenues to date, and the amount they would be eligible for under the formula. The agent would then raise an invoice for presentation to SAA for that amount and payment would be made.
Rationale for the agreements
According to Mr Viljoen, the changes to the override agreements were effected on the basis of legal advice obtained by SAA. It appears that the motivating factor for this change was a finding by the EC in the BA/Virgin case in which similar override incentive agreements were held to be anti-competitive. Undoubtedly the investigation by the Competition Commission at that time played a contributory role in this decision. He also explained that SAA would have preferred to have exclusive dealings with travel agents but they had resisted this. Viljoen explained that trust payments were introduced to “compensate” agents for the loss of income as a result of SAA’s amendments to the override agreements and were directly related to the time and effort travel agents dedicated to SAA products. The more sales they achieved, the more time they spent promoting SAA products, the greater the rewards. Viljoen maintained that this was only one element of the objective of the trust payments and it was meant to achieve a number of other objectives such as marketing of SAA products.Nevertheless he conceded that the override incentive agreement together with the trust agreements had been designed to win the loyalty of travel agents.
According to Ms Harris, trust payments were designed to incentivise agents to achieve market share for SAA in both domestic and international sales. The agent would only be rewarded if they achieved the targets set out.
This explanation taken together with Mr Viljoen’s concession that trust payments were introduced in order to incentivise travel agents for incremental growth and to “compensate” them for the introduction of the flat overrides post base confirms that the rationale for the third generation agreements together with the trust payments was no different than that of the second generation agreements considered inNationwide.
The Relevant market
In the Nationwide decision the Tribunal found that there were two relevant markets namely the market for the purchase of domestic airline ticket sales services from travel agents in South Africa and the market for scheduled domestic airline travel. It found that SAA was dominant in both markets and that it had abused its dominance in the former market in order to exclude its rivals in the latter market. In that case the Tribunal accepted a wide definition of the relevant market for scheduled domestic airline travel on the basis that the effects of SAA’s conduct would be experienced across a range of city-city pairings, passenger classes and flight times.
In that case the Tribunal found that travel agents were an important channel of marketing and distribution of tickets for airlines. It also found that direct sales by airlines over the internet or the counter were not substitute channels of distribution for consumers who wished to examine their choices. The Tribunal found that a significant portion of each of the three airlines tickets were sold through travel agents during the relevant period and that was clear evidence of the centrality of travel agents to consumers.
Since then a number of market developments have occurred in the South African domestic airline market. During 2001 Comair launched the low cost carrier, Kulula. Mr Venter explained that Comair had launched Kulula after observing a similar trend in Europe where growth in airline travel was being facilitated by the advent of low cost carriers (LCC’s) also known as “no frill” carriers.
The LCC model was different from that of full service carriers (FSC’s). LCC’s ran on a very low cost base which they achieved by utilising older aircraft, not offering any free services on board, having one class of restricted airfare and providing only point-to-point flights. A passenger travelling on an LCC would therefore not have the benefit of free in- flight meals and other services, the comforts of premium class or more leg space, fare flexibility, connecting flights through that airline or its alliance partners or access to a lounge. Nor would they have the benefit of any loyalty programmes. Low cost fares were also highly restricted and were usually issued on a use- or- lose basis.
A significant difference between LCC’s and FSC’s in relation to ticket distribution was that LCC’s utilised the cheaper, more direct, routes to market such as the internet in order to ensure a lower cost base. LCCs did not distribute their tickets through the global distribution system (GDS) which were used by travel agents. Flight bookings could only be made on the airline’s website or its call centre and had to be paid for immediately. Accordingly purchasers could not book flights on a provisional basis as they could with travel agents. Nor did they enjoy the expertise and advice offered by travel agents in making their choices.
It appears that when Comair first launched Kulula it included ticket sales in its incentive agreements with some travel agents. It was not clear whether travel agents earned any commission on these sales during this period or whether there was some promotional incentive in place. Ms Harris confirmed that travel agents could not purchase Kulula tickets on the Global Distribution System but offered a service to corporate clients with whom they had concluded in-house deals by purchasing Kulula tickets on the internet utilising the client’s credit card. However Dr Federico suggested that a small proportion of Kulula tickets were distributed through travel agents and by 2006 and 2007 only accounted for 7-8% of Kulula total sales. Nevertheless the overwhelming majority of low cost fares were distributed through channels other than travel agents.
For Nationwide the situation was somewhat different. Its product offering was somewhere in between that of SAA and BA/Comair on the one hand and Kulula on the other. While it operated as a FSC, it utilised a lower cost model than a typical legacy airline such as SAA. Older planes were used, providing point-to-point services, with more restricted fares and a smaller proportion of business class. While it sought to utilise direct channels such as internet and call centres, its tickets were largely distributed by travel agents.
As a result of these market developments, a number of alternative market definitions were debated in these proceedings.
In relation to travel agents Comair argued that this was still a relevant market for purposes of these proceedings. In relation to the market for domestic airline travel Comair argued that subsequent developments in the domestic airline travel justified a segmentation of the market into Time Sensitive (TS) and Non-Time Sensitive (NTS) customers. It argued that the entry of LCC’s and their predominant reliance on direct internet sales and their focus on non-time sensitive passengers called for a more refined market definition and a segmentation into TS and NTS. This distinction was important for a proper understanding of the effects of SAA’s conduct. Comair was in favour of such a distinction as opposed to a distinction between business & leisure travellers or LCCs and FSCs.
Nationwide argued that it was not necessary for the Tribunal to identify a separate market for the purchase of domestic airline tickets from travel agents. The complaint in this case was that SAA’s override incentive schemes induced travel agents to divert sales of domestic airline tickets away from SAA’s rivals. Because the marketing and distribution of airline tickets is inseparable from the airlines’ principal economic function the provision of airline services to consumers, SAA’s conduct can be viewed as a form of input foreclosure. Hence the Tribunal should concern itself with the effects of this vertical agreement between SAA and travel agents on the “ability of competitors to access final consumers”. This was the approach adopted by Oxera, on behalf of Nationwide, in its first report. It concerned itself only with the effects of these agreements on domestic air travel, the primary economic market, rather than for purposes for market definition. But to the extent that the Tribunal found it necessary to define a relevant market, Nationwide argued that it was the market for the purchase of domestic airline ticket sales from travel agents. Nationwide was agnostic on the issue of market segmentation along TS and NTS in the domestic airline travel market on the basis that in its view the effects of SAA’s conduct would largely be felt by those customers who were prepared to pay a higher price for flexibility and comfort and who would utilise the services of travel agents.
SAA contended that because routes other than travel agents were available to Comair and Nationwide, they were not foreclosed from this market and accordingly were not foreclosed in the domestic air travel market. In relation to the domestic airline travel market, SAA argued that market segmentation could not be justified because of the level of product differentiation in the airline industry. It argued for a wide market including FSCs and LCCs. Each class of fare was a discrete product offering with different attributes attached to it. There was a chain of substitution along a ladder with the lowest fare offering the least benefits and the highest the greatest number of benefits. Hence an X fare provided absolutely no benefits such as in flight services, flexibility or comfort but a premium fare would offer the whole gamut of benefits such as comfort, flexibility, free meals and drinks, movies, connecting flights, lounge access, etc. Passengers would sacrifice any combination of benefits depending on how much they were willing to pay. For an X fare passenger price was paramount, moving up along the chain another passenger may purchase an intermediate fare such as a Y fare but sacrifice only a few benefits and so forth. Each class of fare exerted a competitive constraint on the other. Viljoen maintained that price competition between these products was apparent in the marketplace. Both SAA and Nationwide had responded to Kulula’s low fares by launching their own low restricted fares. These low fares exerted a competitive constraint on fares higher up the ladder.
The market for purchase of travel agent services by airlines
Dr Affuso from RBB argued that the relevant market was the market for the purchase of domestic airline tickets and that the Tribunal’s approach in Nationwide was incorrect. The Tribunal ought to consider all goods and services supplied by the dominant firm. Hence in the market for travel agent services it should be concerned with the services provided by travel agents to customers and ought to have examined all alternative suppliers of these services to customers. The basis of her argument was that in abuse of dominance cases the allegedly dominant firm is dominant in the supply of a set of goods or services. She stated that in most cases the supply chain is conceptualised as one flowing from supplier to retailer to customer. Travel agents were supplying airline tickets to consumers as a retailer. Whilst travel agents could be viewed as selling distribution services to airline carriers, the same argument could be made of any retailer who earns a retail margin on the sale of goods and services. These retail margins could be considered as payment for retail services by the customer to the travel agent. Customers could purchase these tickets in a number of ways, travel agents being one of these. Therefore the Tribunal must have regard to all other avenues of distribution such as the internet and direct sales in this market and assess the competitive constraints internet and direct sales place on travel agents.
In our view Dr Affuso’s conception of the supply chain is fundamentally flawed and is not supported by the evidence in this matter. In the first instance airlines do not on-sell tickets to travel agents as one would expect in a wholesale- retail relationship. Travel agents still do not have any discretion with regard to the pricing of the product offered by the airline, the quantity of supply, the terms and conditions on which such product is offered and do not acquire ownership and risk in the product. Hence the travel agent is not entitled to mark up a retail “margin” on these tickets nor is it able to exercise any discretion in relation to discounting the airfare. Moreover, during the relevant period, the customer did not pay for any of the services rendered by the travel agent which could constitute a “retail margin” as postulated by Dr Affuso. All the distribution channels, whether these were travel agents, the internet or direct channels were input costs to the airlines. The effects of SAA’s agreements with travel agents are to be assessed from the perspective of the extent to which SAA’s rivals had been foreclosed through SAA’s commandeering of these input services and not from the perspective of the consumers’ demand for such services.
Indeed the evidence, in this matter, including that of SAA, supports the market definition adopted by the Tribunal in Nationwide. All the witnesses in this matter confirmed that while internet sales and direct sales, either through call centres or corporate agreements had grown during the relevant period, travel agents were still by far the most important avenue for airlines to distribute tickets.
Mr Viljoen described the importance of travel agents to SAA and testified that the services offered to airlines by travel agents would be prohibitively expensive for SAA to offer internally:-
“The expertise of travel agents in establishing and maintaining relationships with customers, results in the offering of an invaluable and efficient service to airlines...In the absence of travel agents airlines would have to offer this service internally... (Our emphasis) Travel agents and their consultants offer specialised services ancillary to the distribution of air travel services. These specialised niche services require skills and expertise not currently held by SAA.”
Mr Mortimer and Ms Harris both confirmed that the principal source of revenue for travel agents were the commissions paid by airlines to travel agents and not from other “retail” activities. Ms Harris also testified that while travel agents provided information and data to consumers they possessed an expertise that was relied upon by airlines and they often assisted airlines in contractual negotiations between them and corporate clients.
All the witnesses testified further that during the relevant period these alternative channels were not suitable substitutes for travel agent services for airlines because the uptake of these channels by consumers was slow for a variety of reasons
In general ticket distribution through travel agents is done through various global distribution systems (GDS) such as Galileo. These are proprietary systems through which airlines distribute their tickets. Travel agents would, at a fee, have access to a GDS system and thereby have access to all airline tickets posted on the GDS. For example a travel agent would, through the GDS, have access to information to all airlines’ domestic tickets over any period of time, their relative pricing and availability. Through these systems the agent would be able to choose the best suited fare for a passenger according to price, availability and scheduling and would also be able to make a provisional booking for that passenger at no cost.
During the relevant period Comair’s sale of tickets excluding Kulula through direct channels was less than 5% of total ticket sales. While internet sales had increased during this period, they still constituted a very small percentage of total airline sales. Direct sales through other channels such as corporate agreements and the call centre had also increased. By late 2004 SAA’s sales through the internet increased above 5% for the first time, much of which was due to X-class fares being taken off the GDS and being distributed only via the internet. Nationwide’s sales through the Internet rose to about 2% in 2005.
Again while the sales through the Internet had increased as a percentage of total sales from the previous period, more than 85% of SAA’s domestic ticket sales were still done through travel agents. Similarly more than 70% of Comair excluding passengers booked on Kulula were sold through travel agents. For Nationwide more than 60 % of its sales, towards the end of the relevant period were still done through travel agents.
Venter described the services offered by travel agents compared to internet sales and stated that these included “managing the total travel budget and consolidating all different elements of travel into reports to the corporate client which is a very important role that the internet cannot do.” He described the value of travel agents to consumers as follows:-
“...the service provided by the travel agents includes “managing the total travel budget and consolidating all the different elements of travel into reports to the corporate client”, which “is very important role and that the internet cannot do”.
He continued further stating:-
“...that there has been a lot of theory about the interchangeability of travel agents with the internet, but the truth of the matter is that there are a lot of services that the internet cannot provide at this stage yet, maybe in the future.”
Mr Viljoen supported Venter’s evidence, stating that:–
“Obviously when Internet came in to some extent changed the game a little because now the Internet is a lot cheaper, but I have to tell you that a lot of people don’t like to use the Internet and that is also quite a mission. So the ease of calling an agent or walking in and doing a booking, now obviously from a pure convenience sometimes is better for the consumer.”
Viljoen reiterated this point stating:-
“We believe we are better served given that it is a vital distribution channel (referring to travel agents) for our business and it is vital, because for me to replicate that distribution channel will come at a huge cost”
Mr Mortimer testified that the tools available on the Internet, for example search engines that look for the cheapest airfare from a variety of sources, which replicated some of the services provided by travel agents, were not available during the relevant period. Some of these tools only came into being as late as 2006.
Mr Bricknell stated that in Nationwide’s experience Internet sales did not represent a complete substitute. He stated:-
“...in our experience back then our internet sales- is all I can go on – were probably half a percent. I must assume from that that the public didn’t use the internet a lot and they were very reliant on the travel agent to extract this information. Now they would phone the travel agent and ask “who has the best fare?” Maybe that has changed a little bit of late because again just based on my experience our internet bookings went up in 2005 to about 2.5%. So I must conclude that people didn’t readily go on to the internet.
That a lot of people phoned the travel agent and asked them their advice and said “who has the best, who flies at that time, which is the cheapest ticket, I need other things, I need reservations, I need car hire, I need travel insurance, I need my itinerary done for me” depending on where he is going. So there is a lot of advantages and I don’t believe a lot of the public used the internet at that stage.”
Even where a passenger obtained an appropriate fare through the internet or other direct sales channels such as call centres, the passenger was required to pay for the fare immediately with very little room for flexibility. Moreover only the cheapest and most restricted fares were exclusively available on the internet or other direct channels during the relevant period. Travel agents could not distribute these very cheap fares because they were not available on the GDS.
In respect of direct agreements with corporates both Ms Harris and Mr Viljoen testified that these were done in two primary ways. On the one hand travel agents themselves would pitch to manage the travel budget of a corporate directly. Such a contract involved the travel agent making all the travel arrangements for the entire company, irrespective of whether or not that corporate had an agreement with an airline or agreements with more than one airline, providing monthly reports and seeking to achieve cost savings for the corporate. Where the airline sought to conclude a direct agreement with a corporate they would utilise the services of travel agents as facilitators and often as partners in the proposal. Often a larger corporate would have both a direct agreement with an airline and an agreement with a travel agent to manage their travel budget. A corporate would typically obtain discounts on fares through these agreements, calculated either on the fare price or on total expenditure by volume or revenue. Fares especially negotiated with a corporate were also included in the incentive agreements. Direct sales, while on the increase, were not a substitute for distribution channel to travel agents.
The evidence of all the witnesses confirmed that channels other than the Internet such as call centres were also sub-optimal alternatives to travel agents. Viljoen testified that for SAA to replicate the network and the services travel agents provided to consumers would be prohibitively expensive. This was the experience of all airlines.
By and large the overwhelming majority of ticket sales for all three airlines were done through travel agents. Quantitatively direct sales (including internet, call centre, counter and corporate agreements) constituted at most between 30% (at the beginning of the relevant period) and 40% (at the end of the relevant period) of total (all airlines) domestic sales. Qualitatively however much of the analysis done by Oxera shows that sales through the internet attracted lower-yield customers compared with travel agents sales. This is confirmed further by the fact that airlines would not have bothered to conclude incentive agreements with travel agents until 2005/2006 if travel agents were not central to the distribution of their tickets.
Conclusion on travel agent market
Apart from Dr Affuso’s theory, all of the witnesses in the matter testified that despite the growth of the internet and sales through other channels, travel agents were by far the most important avenue through which domestic airlines could distribute their tickets. Travel agents provided specialised services to airlines and possessed knowledge and expertise which customers relied upon. If airlines did not purchase these services from travel agents they would have had to replicate these in-house and found this to be prohibitively expensive. All three airlines relied on travel agents for the bulk of their domestic airline ticket sales. The internet and other direct channels were sub-optimal substitutes to travel agents as has been demonstrated by the limitations and low uptake of these services by customers. Hence we define the relevant market as the market for the purchase of travel agent services for the sale of domestic airline tickets.
SAA was the largest purchaser of travel agent services in this market. SAA’s market share including SAX and SAL in the market for travel agent services, calculated by shares of BSP, is as follows:-
Table 2: SAA, SAA+SAX + SAL market shares Apr 01 – Mar 06
On the basis of market share figures above, SAA was irrebuttably the dominant purchaser in the market for airline ticket sales services provided by travel agents in South Africa.
(ii) The market for domestic scheduled airline travel
In the Nationwide decision SAA had argued that each paired (city-to city) domestic route constituted a relevant market and that the Tribunal should examine the effects of the incentive agreements on competition between the airlines on each paired route rather than in the entire country. The Tribunal had rejected that argument on the basis that SAA’s override incentive agreements applied uniformly to all domestic SAA tickets sold. The agreements did not apply only to a specific class of ticket or a specific route. The Tribunal held that the agreements could affect the sale of any of its rivals’ or potential rivals’ tickets on any route at any time or on any class.
In these proceedings SAA persisted with such an argument albeit not arriving at any firm conclusion.
In support of the route-by-route approach to market definition SAA provided us with a table showing market shares by number of flights in 2006:
Table 3: Market shares by number of flights in 2006
Route 1Time Comair Kulula.com Nationwide SAA
Jhb-Cpt 12% 20% 13% 13% 42%
Jhb – Dbn 10% 14% 14% 16% 46%
Jhb – PE 5% 21% 21% 17% 37%
Dbn – Cpt 0% 23% 18% 12% 47%
Jhb-George 9% 6% 18% 21% 47%
The numbers in Table 3 above are said to be drawn from each airline’s website. No explanation was given as to when in 2006 these calculations were done or whether these were accumulated figures over a defined period of that year. The numbers also do not give us an indication of how many flights were actually flown during this unknown period, the number of tickets sold for each flight or the number of passengers on each flight. Furthermore the table includes 1Time a low cost carrier owned by 1Time Holdings which was launched in early 2004, but does not indicate whether South Africa Express (SAX) and South African Airlink (SAL) have been included in these figures, the sales of which had been included in SAA’s incentive agreements. It appears that the table also contains a few inaccuracies as pointed out by Mr Venter. Thus a mere counting of scheduled flights does not give us a helpful indication of the competitive dynamics on each of those routes. Nevertheless, even on this limited information, we see that SAA enjoys a market share ranging from 37%-47% on each of these routes.
In our view while there have been some changes in the market place since the Nationwide decision, such as the launch of Kulula and the subsequent launch of 1Time in 2004, the increased use of the internet or more direct sales channels and changes in the SAA override agreements, these do not justify geographic market segmentation on a route-by-route basis. SAA’s agreements during the relevant period, while undergoing some changes, still applied across all SAA domestic flights in the country. Nor were they limited in application to only those routes on which SAA faced competition from rivals. Hence they also applied to potential competition on routes on which rivals had not yet scheduled flights.
Moreover they included the sales of SAX and SAL. In 1997 SAA formed a strategic alliance with SAL and SAX. By agreement the three airlines had co-ordinated routes so that SAA, SAX and SAL did not compete with each other. While SAA continued to fly the lucrative routes, SAX and SAL operated on the smaller routes. The alliance entitled customers to Voyager benefits, flight schedules and services such as ticketing, checking facilities, mutual branding and flight codes. At the same time the three airlines were branded as one and were indistinguishable in the minds of the consumer from each other. All the witnesses confirmed that it was the sale of SAA, SAX and SAL tickets, on all routes, that were the subject of the incentive agreements. Even if SAA did not fly a particular route that it had, by agreement left to SAX and SAL, it ensured that the sales of tickets on those routes by travel agents were rewarded in the same manner as the sales of its own tickets. Hence the agreements would also apply to those routes on which SAX and SAL but not SAA competed with Comair and Nationwide. Similarly the trust agreements were not limited only to specific routes but related to SAA’s domestic and international market shares.
Hence we conclude that the relevant market at this stage of the enquiry is the market for domestic scheduled airline transportation in South Africa.
At the same time we accept that the golden routes comprising Jhb-CT, JHB- Dbn, JHB – PE and JHB-George constituted more than 70% of the total domestic airline market and that comparative evidence for the different airlines on these routes was more easily available and was at times relied upon by parties as an indicator of the competitive dynamics in the national market.
TS & NTS segmentation
The launch of Kulula in 2001, the launch of 1Time in 2004, SAA’s launch of its X fares and the increased use of the internet did raise a different debate in relation to market definition. Kulula primarily utilised the internet and its call centre as a channel for ticket distribution. It provided point to point services at low cost restricted fares. In- flight services were not free or were unavailable and passengers did not have access to different comfort levels, lounges or loyalty programmes. The airline’s strategy was to fill the planes as much as possible at the lowest possible cost with the result that passengers were packed into older planes with limited leg space. Low prices were the predominant basis upon which the product was marketed. The launch of Kulula revolutionised domestic air travel in South Africa and accounted for most of the growth in the domestic air travel market during the relevant period. Passengers who would have previously relied on road or rail could now afford to fly and more frequently. A similar trend could be seen in other countries where the launch of LCCs propelled the air travel market out of the doldrums following 9/11 and accounted for most of the growth during subsequent years.
According to Viljoen the launch of Kulula created price awareness in the domestic market to such an extent that the other airlines were forced to respond. SAA eventually responded to Kulula by launching its “X” fares. This was a species of fare which was heavily discounted but highly restricted and could only realistically be offered to passengers in the economy class (at the “back of the bus”) who were willing to sacrifice all flexibility in favour of price. At the same time SAA maintained its high value fares, on the same planes, towards the front of the bus. At first, X fares were included in the incentive agreements with travel agents although sales of these were counted “in for growth but not for payment”. Eventually, by August 2004, after the launch of 1Time, SAA removed the X fares from the Global Distribution System and made these available only through the internet. In 2004 the low cost carrier 1Time was launched, which operated a similar model to that of Kulula,
Nationwide also responded to price competition from Kulula. Mr Bricknell confirmed that Kulula created price awareness and competition in the domestic airline travel market and Nationwide responded to this while maintaining its FSC model. By the end of the relevant period Nationwide’s business seems to have evolved closer to that of Kulula. Because of these developments Comair argued that the market for domestic scheduled airline transportation should be segmented into time sensitive (TS) and non-time sensitive (NTS) passengers.
The notion of time sensitive and non-time sensitive passengers is drawn from European merger cases. In the Air France/KLM merger, the European Commission considered the existence of two relatively clear groupings of passengers and held them to be sufficiently different to establish two markets namely TS and NTS. A similar conclusion was reached in the Lufthansa/Swiss case.
A fair amount of time was spent on the characteristics of these passengers. Comair submitted that TS passengers are willing to pay a premium for the timing and other benefits offered by FSCs. These product features include flexibility (the ability to change or cancel a booking at no charge), frequency and scheduling of flights at peak times, lounge facilities and in flight comfort and services. TS passengers are more likely to be business/corporate travellers but this segment also includes passengers travelling for other purposes who value the benefits mentioned above ahead of price. Leisure travellers for example may be willing to pay a premium for travelling in comfort or for the convenience of catching connecting flights at a particular time. TS passengers value the services of travel agents and the advice offered by them. NTS passengers on the other hand can be defined as those passengers who are willing to trade flexibility, comfort and scheduling in return for a lower price. Often these passengers were identified as leisure travellers however there were instances where it was evident that price sensitive business passengers, such as SMMEs, who were willing to trade all benefits for price. Some large corporate also utilised low fares for junior staff in order to promote savings on their travel costs.
SAA’s internal documents show that it too had regard for the varying requirements of different types of passengers. They also refer to the “five times” price differential between fares that had been a feature of the legacy airline pricing model. Comair relies on this to support its argument that the TS and NTS segmentation is strongly supported by the ability of FSCs to exercise significant and durable price discrimination between the fully flexible and restricted fare classes. A further basis put forward in support for segmenting the market between TS and NTS was the channels used by airlines to distribute the different products. Kulula utilised the internet and SAA removed its X fares from the Global Distribution System for the lower fares which were marketed at NTS passengers while the high revenue, flexible and premium fares were distributed through travel agents.
Nationwide, as discussed remained uncommitted to market segmentation in these proceedings.
Not one of the parties in these proceedings could provide the Tribunal with any reliable data which could accurately delineate these two segments. For example no data for SAA’s X fares over time was made available nor was there much information provided on the nature of Nationwide’s belated low cost offering. Comair relied on the travel agents BSP figures of the premium and sub-premium fares as a proxy for shares of the TS market.
An additional factor that militated against market segmentation from a supply side analysis was Comair’s practice of flying overbooked Kulula passengers on Comair planes. While the low cost product was marketed and sold with restrictions, in instances where Kulula was overbooked, passengers were carried by the airline at the back of the Comair plane, giving such passengers a windfall in relation to in-flight services. At the same time large corporate entities who had concluded travel agreements with Comair or with a travel agent, would utilise Kulula for junior staff travelling for training programs that had been planned in advance. Further confusion was created by the suggestion that all business travellers were time sensitive and all leisure travellers were price sensitive. However it was later accepted that not all business travellers were time sensitive and not all leisure travellers were willing to sacrifice comfort or convenience for price.
In our view an understanding of the actual competitive dynamics in the market could be inferred from none other than SAA’s override incentive agreements. Recall that even at the commencement of the relevant period SAA had concluded override incentive agreements with travel agents across all classes of fares. It launched its own low cost carrier in response to Kulula only in October 2006. Even though the lowest cost fare, the X fare, was subsequently removed from these agreements this was done progressively and only finally removed in late 2004. All other fares, even when they were ultimately sold through other channels such as the internet, still remained the subject of the override agreements. Although all three airlines offered a highly restricted low cost fare on the one hand, and a premium fare on the other hand, by and large the vast majority of all three airlines’ business was located in those fares found between the two extremes of the lowest and the most costly fare premium fare, namely the middle-segment fares. This middle segment consisted of various classes along a ladder of restriction and price. From a demand side one would expect a greater degree of substitution from passengers in the lower part of the middle segment than from the upper part and a very limited, if at all, switching from premium passengers who had already indicated their unwillingness to forsake flexibility and comfort for price.
Without making any findings on the applicability or otherwise of Dr Affuso’s chain of substitution in this matter, it appears to us that SAA’s incentive agreements and its conduct during this period demonstrated that it had anticipated competitive pressure on all of its fares after the launch of Kulula. This is confirmed both by Mr Viljoen when he complained about travel agents “off-selling” SAA for Kulula and Ms Harris who testified that corporates would require travel agents to manage travel agents more efficiently using Kulula. Although the size of SAA’s commissions was progressively amended in the relevant period, with higher fares earning higher commissions, all the farescontinued to remain the subject of the override incentives.
This is also why in all probability SAA introduced its incentive schemes – as an instrument through which to immunise itself from competition. In 2006, SAA’s Ultralite business plan confirms this view:-
“While SAA has managed to hold passenger volumes steady, continued expansion by competitors could see SAA lose price sensitive customers to other airlines”.
In the ultralite business plan, the author anticipates a future scenario, namely that 70% of the market could end up with LCCs by 2012, with switching amongst business customers, as compared to leisure customers, being fairly limited. However this is a future scenario, used to justify SAA’s dual strategy of launching a LCC model and maintaining a legacy lite FSC . During the relevant period, SAA strove to immunise all of its fares from price competition through the mechanism of aggressive incentive schemes.
What we see in this time period is an evolving market. We see price competition and growth of the domestic airline market, created by the launch of an LCC model. We see responses to the launch of LCCs from other airlines in the form of limited price competition at first (SAA, NW) and an attempt to immunise their products from competition (SAA). Only in 2006 do we see a complete response by SAA (launch of Mango) and Nationwide (bringing their business model closer to Kulula). In our view the market may have ultimately evolved into price sensitive /LCC/time insensitive or non-time sensitive/FSC segments. However during the relevant period the competitive dynamics in the market did not justify segmentation.
Mr Gotz on behalf of Nationwide argued that it was common cause that the various offerings of the airlines were highly differentiated and that such differentiation is significant for the purposes of evaluating economic effects of the override incentive scheme. Because the offerings of the airlines were differentiated the override incentive scheme may have a differential impact on the products that are offered and on the different competitors. A differential impact does not mean that competition in the market (broadly defined) is not significantly affected. This was the approach taken by Dr Niels from Oxera appearing on behalf of Nationwide. Although it is not clear that the distinction between TS and NTS or between full service and low cost airlines during the relevant period was sufficient to justify separate markets, nevertheless these distinctions should be kept in mind when assessing the strength of the relative competitive constraints in the market.
Recall that market definition is an analytical tool and that the exclusionary conduct we are concerned with are SAA’s override incentive and trust agreements with travel agents and their effect, if any, on Nationwide and Comair in the domestic air travel market. In order to enable us to better understand the effects of SAA’s agreements, it is important for us to appreciate this emerging market segmentation into price sensitive and non-price sensitive passengers but not necessarily to conclude firmly on such segmentation.
We therefore find that there is a single market for scheduled domestic airline travel. Nevertheless, we accept that in this period the various offerings were differentiated and that SAA’s conduct, if exclusionary, would predominantly have an effect on its rivals on that part of the domestic air travel market which was distributed by travel agents and which would include the qualitatively higher margin fares. For convenience we refer to this as the travel agent segment (TAS). This segment would consist of the less price sensitive and more time and comfort sensitive passengers and would exclude Kulula and 1Time and all Nationwide and SAA fares which were exclusively distributed through the internet or other direct channels. Once again we do not have data enabling us to accurately delineate the boundaries of these segments. However what is available is the size of the market that was distributed through travel agents. During the relevant period the total size of domestic air travel sales through travel agents was estimated in BSP, at R3,598bn in 2000/01 and R3,430bn in 2004/5 representing approximately 70% of the domestic airline travel market. Certainly it was a declining market but the decline was relatively slow, contracting to R3,382bn in 2006/7.
Dominance of SAA
Having concluded that there is a single market for scheduled domestic airline travel, we turn to consider whether SAA was dominant in this market. Once again SAA pursued an argument in these proceedings, that despite its high market shares, SAA had no market power. It also persisted with the argument that the market shares of SAX and SAL should be excluded from the computation of its shares.
In Nationwide the Tribunal held:-
“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 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(b) where the presumption of market power is rebuttable. SAA is not just dominant but overwhelmingly so.”
There is no need for us to re-iterate the approach to section 7 previously adopted by this Tribunal. But for purposes of completion we point out that the provisions of section 7 are abundantly clear. Section 7(a) provides that a firm is dominant when its market share is 45% or more. Section 7(b) creates a rebuttable presumption of dominance in the event that the firm’s market share is between 35% and 45%. In that case the firm must show that it does not have market power. Section 7(c) creates a presumption of dominance if a firm has less than 35% but enjoys market power. An inquiry into market power is only necessary when a firm’s market shares are less than 45%.
Mr Viljoen explained with a measure of pride, that SAA had a strong alliance with SAX and SAL since 1997. The alliance was created to co-ordinate routes between the three and to prevent duplication or competition amongst the partners of the alliance. SAA provided SAX and SAL with marketing and ticketing service. Passengers that flew on SAX and SAL were entitled to all SAA benefits including use of checking facilities and Voyager benefits. In the eyes of the passenger, SAA, SAX and SAL were one economic entity. As discussed above the override incentive agreement included commissions for sales of SAX and SAL tickets, confirming that SAA itself saw its alliance as one single economic entity vis-a-vis its rivals. Through its incentive agreements with travel agents, it sought not only to reward growth of SAA’s share of the market, but the alliance’s share of domestic airline travel market. Hence we find that the SAX and SAL shares of the domestic scheduled airline transportation market must be included in the computation of the SAA market share.
The market share calculation for SAA including SAX and SAL for the period between April 2001 to March 2005 reflect the following:-
Table 4: SAA, SAX and SAL’s market shares Apr 01- Mar 05 (%)
1Time Comair Kulula Nationwide SAA,SAX
Apr01-Mar02 0 18 4 7 71
Apr02-Mar03 0 15 8 8 69
Apr03-Mar04 0 15 11 8 66
Apr04-Mar05 6 15 12 10 58
Hence we conclude that SAA is presumptively dominant, as provided in section 7, by virtue of its market shares in the wider market. SAA is also presumptively dominant in the market for travel agent services. This dominance has not only been demonstrated by its high market shares but also by its ability to impose terms and conditions of purchase with travel agents which we discuss later.
We have established that SAA is presumptively dominant in both the market for the purchase of travel agent services for the sale of domestic airline tickets and the scheduled domestic airline travel market. We now turn to consider whether it had abused its dominance.
In the Nationwide decision the Tribunal held that SAA through its incentive schemes contained in its override agreements and Explorer scheme induced travel agents not to deal with SAA’s rivals in the domestic scheduled air transportation market and hence constituted an exclusionary act under section 8(d)(i). It held further that the exclusionary act had a significant anti-competitive effect on SAA’s rivals in that it foreclosed the market to rivals. In that decision the Tribunal found that while it was highly likely that this foreclosure had had an adverse effect on consumers quantifying this harm was difficult.
The approach taken to section 8(c) and 8(d)(i) in the Nationwide decision has been endorsed by subsequent decisions of this Tribunal. In that decision the Tribunal summarised its approach as follows:-
“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.
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.
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.
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.
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.”
The Tribunal was critical of following a form-based approach in section 8(d), reasoning that the words “exclusionary acts” served as a signal to respondents that the defined acts in the sub-sections of 8(d) were commonly associated with egregious behaviour rather than creating an presumption of anti-competitive effects simply because the conduct under investigation fell into one of the defined acts.
In that enquiry the Tribunal conducted a two stage enquiry, asking whether travel agents were financially incentivised by SAA to move customers away from rivals and towards SAA, and that travel agents had the ability to do so. We deal with the latter issue first.
Ability to divert
SAA’s defence in these proceedings was that even if agents did have that ability such a strategy was unsustainable because customers would discover these unethical practices and this would lead to reputational damage for travel agents. It argued that recent developments in the industry such as the rise of internet sales increase in corporate agreements between airlines and large companies and the increasing number of travel agents being appointed as travel managers by large corporates to achieve savings in their travel budgets acted as constraints on travel agents’ ability to divert sales. It was suggested that the increased awareness of airfare prices made it almost impossible for travel agents to influence customer’s preferences to their detriment.
While market conditions may have changed to some extent we find that the ability of travel agents to influence customer’s preferences to a significant degree had not been affected by these changes. We have already dealt with the slow growth of internet sales and the evidence by travel agents and airlines alike that South Africans, at least during the relevant period and those not travelling on the extremely restricted fares, preferred to use travel agents and relied on their expertise and advice. Moreover sophisticated search engines on the internet had not yet developed to enable customers to conduct searches for the cheapest or the most convenient of fares. Customers could not hold bookings on the internet so comparisons between airlines and schedules were almost impossible to achieve. In contrast travel agents still utilised the GDS and were able to provide travellers with a variety of options and advice on airfares. The asymmetry of information between travel agent and customer still prevailed and persisted for the duration of the relevant period.
Mr William Puk of Sure Travel, in an email addressed to his managers, after SAA had made its announcement of moving to a zero commission structure, advised them of the future strategy as follows:-
“...going forward though the signs are not good. It has become very clear that we cannot rely on saa for a decent override agreement in future and our basic commission is about to disappear altogether. Furthermore I see no signs that their utter contempt and disregard for travel agents is being reconsidered. Therefore, I am formally advising you that our group strategy is to move our discretionary business away from saa onto more agent friendly carriers, hence the new deals with Virgin & Nationwide. Our international priorities must now lie firmly with British Airways, Virgin, Lufthansa, Nationwide, Cathay Pacific etc and domestically with BA/Comair & Nationwide as a first priority. An effort and directive to this effect must therefore be communicated by you to all your staff. It makes sense from a business point of view, 0% commission from saa and generally expensive GDS fares to sell to consumers, versus standard guaranteed commission from other airlines “provided we move the business to them) and generally better value fares for the consumer. We need to show saa in the months of Feb/Mar/ & Apri that travel agents are still vital to their business and that we can and will, direct the business away from them.”
Ms Harris, Dr Affuso and Mr Viljoen despite contending that this strategy was not sustainable, all conceded that travel agents did indeed have the ability to influence customers’ preferences. Mr Mortimer testified on travel agent’s ability to influence customer’s preferences as did Mr Venter.
Dr Federico in his testimony, stated that the empirical evidence on Comair’s share of BSP across travel agents supports the proposition not only of directional selling, but also that such directional selling was significant in terms of market share movements for carriers. Dr Federico further demonstrated that during the two years (FY 2001/02; 2003/03) when Amex, which is part of Tourvest, did not have a contract with SAA, Comair’s share at Amex increased by 9%. Dr Federico further argued that by contrast, Comair’s share over the same period for agents not supportive of Comair declined by 5%. We attach Dr Federico’s exhibit 16, Slide 8 as Annexure 2 and return to discuss it later.
Ms Harris confirmed that while the Rennies Group would never do such a thing, some travel agents, such as Sure Travel, engaged in such practices. However, during that period, and in an attempt to persuade SAA not to remove the front end commission, Ms Harris herself had written to SAA stating that her company “would be forced to move our support to your competitors”, confirming that travel agents were able to shift business away from SAA.
Viljoen suggested that Tourvest was a maverick and was manipulating ticket sales in order to earn its commissions. All other travel agents did not engage in these questionable practices and this was the reason for the dispute between SAA and Tourvest. In the first instance this version was not contained in any of the pleadings or witness statements. Nor was it canvassed with Mr Mortimer or Mr Federico in cross-examination. Viljoen also attempted to reduce travel agents claims to mere threats (puffery). This would beg the question: If travel agents did not have the ability to divert sales why did SAA conclude all the override and trust agreements with travel agents at all? Mr Viljoen attempted to ward off this question and his concession by saying that SAA was “uncertain” about the ability of travel agents to directionally sell and concluded these agreements in order to manage that risk. Once again that explanation was never foreshadowed in the witness statements and remained totally unsupported by any internal strategic or risk management documents we would have expected to see.
During the relevant period SAA spent approximately R300m per annum in incentives through its override and trust agreements on travel agents. It had concluded agreements with travel agents that amounted to 70-90% of that market. A travel agent such as Rennies earned incentives in tens of millions of rands per annum from SAA agreements during the relevant period.
Mr Viljoen himself, when explaining the role that travel agents played in the market and SAA’s decision to launch its X fares, testified to travel agents’ ability to “sell off SAA” in favour of Kulula for their corporate customers. In light of his own evidence regarding the importance of travel agents in the South African air travel market and SAA’s dependency on them, its desire to have exclusive arrangements with them and the vast sums of money paid to travel agents for achieving these targets, we find Viljoen’s attempts to deny travel agent’s ability to divert sales unpersuasive.
All of this shows that travel agents ability to influence customer’s preferences was much greater than that suggested by Viljoen in his evidence. This great ability is confirmed by none other than Mr Ngqula, the then CEO of SAA, in its 2006 annual report when he states that:-
“At first the trade directed business to our competition before the other airlines followed suit cutting commissions some 6 months later.”
In conclusion, all of the evidence, including that of SAA’s own witnesses and documents, strongly supports a finding that during the relevant period travel agents did indeed have the ability to influence customer’s preferences to a large extent and that the growth of the internet and other direct sales channels had not eroded this ability to any significant extent. There is some evidence that travel agents such as Sure Travel actually engaged in these practices.
Having found that agents had the ability to divert customers, we now turn to consider whether the financial incentives offered to travel agents during this period induced them not to deal with SAA’s rivals. In order to understand the impact of the 3G agreements we need to re-cap the impact of the previous agreements.
In Nationwide the Tribunal found that SAA’s override incentive agreements and the Explorer scheme constituted exclusionary practices in terms of section 8(d)(i). As discussed above during that period the 2G override agreements between SAA were designed to reward agents with a flat standard 7% commission for all SAA’s sales up to a target. Once that target was exceeded, two further types of commission were paid. The first of these was the override commission calculated on a back to rand one principle, calculated over all SAA sales below that target. The second was the incremental payment made only on the amount of growth in excess of the target, calculated on the basis of the back to rand base (the target was referred to as base in this calculation) principle.
The Tribunal held that the override commission incentives under those agreements were particularly strong because the additional commission was granted not only on the additional sales achieved (i.e. on a marginal basis) but on all the tickets sold in excess of that target or threshold. The incremental commission was equally aggressive because it was calculated not only on marginal sales but on all sales in excess of that threshold. The impact of those agreements was such that the profitability of travel agents became very sensitive to whether they met the target levels or not. Travel agents were not concerned about the average rate of commission they earned when selling SAA tickets but rather the impact of the additional sale on the total commission accruing to that agent. In other words the agent would be concerned about whether or not that additional sale brought it closer to the agreed thresholds.
The Tribunal highlighted the strength of those incentives in a modelling exercise contained in Appendix 1 of its reasons demonstrating that travel agents faced very high commission rates in excess of 30% when targets were exceeded. The models showed that the agent could maximize total commission revenue by increasing SAA’s market share at the expense of SAA’s rivals. Even though SAA’s average commission paid to travel agents remained relatively low, its marginal commission rates were very high. A smaller rival, attempting to match the same cash value of the marginal commission payment offered by SAA, would have to pay much higher average commission rates.
Viljoen explained that in the 2001 incentive scheme agents were still paid the standard 7% commission on each ticket sold. They would still earn an override commission on a back to rand one basis when they reached a particular target (set by SAA) calculated on the value of all tickets sold up to target (base). This computation was the same as that in the 2G agreements. However in relation to ticket salespost base, the agents were no longer paid a commission calculated on an override basis but were paid a flat commission. This resulted in less revenue for the agents on post base sales. Trust agreements were introduced to “compensate” agents for the losses they would incur as a result of this amendment.
Given this explanation, how were the 3G agreements together with the trust payments, different from the 2G agreements? In other words, Viljoen’s explanation that the agreements were amended and trust payments were introduced in order to compensate travel agents for losses incurred as a result of the amendments confirms that the 3G agreements were no different, in effect, to the 2G agreements. They were designed to continue providing travel agents with the same level of reward previously received under the 2G agreements.
This fact taken together with travel agent’s ability to divert sales away from SAA’s rivals would lead us to conclude that the 3G agreements constituted an exclusionary act inducing travel agents to deal with SAA at the expense of its rivals in contravention of section 8(d)(i). This would be the short answer to our enquiry. The long answer and the evidence submitted to this Tribunal in these proceedings bears this out.
It is common cause that the structure of the 3G override agreements up to the base threshold remained largely unchanged from the 2G agreements. The agreements also included SAX and SAL. In other words agents were still rewarded a standard 7% commission on all SAA (including SAX and SAL) sales. However once they reached the agreed target, which was usually the agent’s previous year’s sales, they would be paid an additional override commission, calculated on a back to rand one principle on all sales below that target.
The wording of some clauses in the 3G agreements did however create another debate as to the incentives post-base or post that target. SAA maintained that the commissions post base were not paid on an override basis. Dr Niels, from Oxera, testifying on behalf of Nationwide, maintained that the commissions post base were still paid on an override basis. He explained that in a typical agreement such as the one with Sure Travel, the base was defined as the annual “domestic flown revenue derived by SAA, SAX and SAL from sales effected by the agent for the financial year preceding the agreement.” The base could be calculated monthly, quarterly or annually in SAA’s sole discretion. SAA agreed to pay a “cash override incentive” to the agent calculated in accordance with the table set out in Part 1 of Annexure A to the agreement. In his view the structure of the agreements resulted in the marginal incentive rates for SAA’s larger travel agents such as Sure Travel being very high, exceeding 100% in many cases when the base itself is met. The annexure referred to by Dr Niels in his testimony led him to conclude that in addition to the very high marginal incentives for travel agents to reach base, there was a range of thresholds, beyond base, at which the marginal incentives are much higher than the 7% basic commission rate and which ensured that the travel agents continued to be incentivised to meet additional sales targets on behalf of SAA.
We have attached Dr Niels interpretation as Annexure 3 to these reasons.
He explained further how he came to understand the commissions post-base, and stated the following:-
“If you look at page 52 this is indeed one of the annexes that every agreement typically has. And you can see that there are, in this case, 5 step changes in the override arrangement. At base you get an override payment, at base plus 5, plus 10, plus 15, plus 20. And that works its way through these calculations. So you do also see the step changes when you reach base, when you reach 5% when you reach 10%. Every percentage point increase that you achieve between these step changes, the marginal rate that you get is just, again the standard commission of 7%. But once you reach the next target level, if you like, you get again the override payment. So that’s for the marginal payment at those points is high.”
Dr Neils was referring to the table reproduced below where column 1 represents the milestones to be reached by the agent and column 2 represents the commission in relation to that milestone:-
Table 5: Determining Domestic Cash Override Incentive
Total Domestic SAA Flown
Revenue for the Financial
Applicable % Cash Override Incentive
Discounted K-class Premium
2% 2.3% 2.5%
2% 2.3% 2.5%
2% 2.3% 2.5%
2% 2.3% 2.5%
2% 2.3% 2.5%
2% 2.3% 2.5%
Oxera’s interpretation of column 2 agreements taken together with the trust payments would suggest that the incentives were moreaggressive and therefore represented a greater financial incentive than that submitted by Mr Viljoen.
SAA however vehemently opposed this interpretation claiming that all commissions paid on sales that exceeded the target (base) were paid at a flat rate and not as an override rand base. In support of that interpretation Dr Affuso plotted the incentives in the 2G agreements against those in the 3G agreements showing a flatter step up pattern post base for the latter. Viljoen insisted that the Trust payments were introduced in order to “compensate agents” for the removal of the incremental override payments i.e. back to rand base commissions.
Ms Harris supported SAA’s version, despite the fact that the agreement with Rennies contained similar provisions as those in the Sure Travel agreement. When questioned as to the purpose of the milestones post base reflected in column1 she responded as follows:-
“CHAIRPERSON: What was the significance of having these thresholds of base plus 5, base plus 10?
MS HARRIS: I have no idea, because there was no application that said you would get 2% plus another 2%, in other words, it would be 4% it was 2% and it was 2% irrespective of any growth. So with effect and I think this was the year in which the ruling from the previous case was then brought to bear by South African Airways where no targets applied and I guess what they were seeking to highlight here is that it didn’t matter what you grew you would get your 2%. It didn’t matter what you would grow buying your K class, whatever was in the K class category from that previous year would be rewarded at 2.3%.
“CHAIRPERSON: So they are meaningless.
MS HARRIS: Totally.
CHAIRPERSON: The milestones are completely meaningless?
MS HARRIS: That is why I say it is rather bizarre...”
None of the SAA witnesses could shed light on why SAA’s agreements still contained the milestones post base. The glaring and most obvious question that presents itself is this. If SAA intended not to calculate commissions post base on a back to rand base principle i.e. on an override basis, why did it not, when it was engaged in a process of amending its agreements not simply include a statement to that effect? Instead it retained these provisions and range of other vague clauses, an explanation for which could not be provided.
We see that Ms Harris herself, despite being a director of Rennies and the chief negotiator of these agreements, was not completely knowledgeable about the manner in which the overrides were calculated. Furthermore Rennies like all other agents were in the dark about achieving their targets because all the information relating to flown revenue and targets was in the hands of SAA. They relied on SAA to inform them of their progress and commission earned every quarterly. Mr Mortimer confirmed that the survival of their business depended on having these override agreements in place. Ms Harris explained that having an override agreement with SAA was imperative for Rennies’ financial survival. Even if the average or marginal incentives or the percentage commissions were lower than those offered by SAA’s rivals, the volume of SAA sales that were the subject of the override incentive represented a huge cash value for Rennies which they could not afford to risk. She testified that the financial value represented by an SAA override agreement for Rennies was such that during negotiations with SAA at one point in time she was advised to accept the terms of a very vague trust agreement simply to secure the override agreement.
Let us assume for arguments sake, that despite the wording of the agreements, SAA in fact only calculated commissions post base as flat payments and not as overrides and that the reason why the actual clauses were not amended to reflect the reality was because of some benign reason. Whether or not the incentives post base were computed on an override basis did not alter the fact that the incentives to achieve base or targets were still paid on an override basis on a back to rand one principle.
On SAA’s own version the 3G agreements together with the trust payments constituted as great an inducement as the 2G agreements. As explained by Mr Viljoen, travel agents were compensated by trust payments for loss of revenue caused by the amendments. SAA also introduced minor changes to the agreements every year, adjusting what was included and excluded for determination of base and thereby making it increasingly difficult to achieve the targets. But essentially the incentive agreements continued to reward travel agents on a back to rand one principle for achieving base, topped up with trust payments made on the achievement of market share targets. This is demonstrated clearly in Oxera’s table discussed above and attached hereto as annexure 3 which computes the marginal incentive earned by Sure Travel to achieve base over the relevant period as follows:
Base 57.0 157.0 189.5 158.6 53.5
These represent significant financial incentives which rivals could not match.
SAA’s ability to conclude agreements containing vague provisions with travel agents as sophisticated as Ms Harris, both parties having access to substantial legal resources, nevertheless demonstrates the extent of SAA’s market power in relation to the purchase of travel agent services. The fact that sophisticated travel agents signed agreements with meaningless clauses demonstrated both SAA’s market power vis-a-vis these agents and that the agreements still represented significant financial incentives to such an extent that they were willing to accept vague and meaningless clauses in their agreements as long as they had concluded an override agreement with SAA.
What then of the trust payments? On Viljoen’s version these were compensation for the loss of the override incentives post base. As we have found above the third generation agreements on their own constituted an inducement for travel agents to not to book passengers on rival airlines. The cash value of these incentives was critical to travel agents’ survival. The trust payments also constituted a financial incentive. Trust agreements themselves were a type of override agreement. They were designed to incentivise travel agents to achieve certain targets and were paid only after those targets were achieved. However partial achievements of those targets were also rewarded. Amounts paid were not insignificant and often determined the profitability or otherwise of a travel agent. The degree of discretion enjoyed by SAA in those agreements further induced travel agent’s to move customer preferences away from rivals towards SAA.
As stated by the Tribunal in the Nationwide decision, it is not the existence of the override incentive agreements and the trust agreements that is of concern here but their nature. In our view the amendments introduced by SAA during this period created a greater not a reduced anxiety on the part of travel agents to please SAA. In the first instance travel agents were being rewarded to maintain or slow down decline of SAA sales of previous years on an override basis. This meant that even where travel agents were offered a higher marginal incentive by rival airlines, they were still induced to achieve SAA volumes (base) in order to achieve the override commission. As explained by Harris, the back to rand one formulation in the override incentive agreement was designed to incentivise the agent to reach the target and to be rewarded handsomely. However the agent was not always certain whether the target had been achieved because it was calculated on the basis of flown revenue. Only SAA was in a position to establish when the agent would have achieved the target. Because agents were uncertain about whether or not they had achieved that target they would focus all of their efforts on sales of SAA tickets until they had received confirmation from SAA. While the incentive agreements rewarded the agents for maintaining and achieving the sales level of the previous year (base) over the total sales from rand one, trust payments rewarded agents for increasing market share in both the domestic and international markets and for supporting SAA.
In our view these amendments further influenced the behaviour of travel agents to direct sales towards SAA. This influence and hold over travel agents that SAA exercised was palpably apparent in Ms Harris’ demeanour and explanation when she was asked why Rennies would sign an agreement with SAA which contained vague and meaningless terms.
In Nationwide this Tribunal found that override agreements structured in such a manner have provided financial incentives to travel agents to direct customer’s preferences. The evidence in this matter is no less persuasive of such a finding. We find that the financial incentives contained in the override agreements are in breach of section 8(d)(i) as travel agents were induced to direct customer’s preferences towards SAA and away from its rivals.
Accordingly we find that the override agreements and the trust payments, separately and collectively, were in contravention of section 8(d)(i). Having found this we do not need to consider whether its conduct was in breach of section 8(c).Anti-competitive effects
Having found that SAA’s override agreements and trust agreements constitute an exclusionary act in terms of section 8(d)(i), we turn to consider whether such exclusionary conduct resulted in anti-competitive effects. In Nationwide the Tribunal set out its approach to section 8(d). The Tribunal stated that anti-competitive effects for purposes of section 8 can be proved in two ways namely:-
“(i) evidence of actual harm to consumer welfare or
(ii) if the exclusionary act is substantial or significant in terms of its foreclosing the market to rivals”.
Hence an anti-competitive effect could manifest in two ways. 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. The latter criterion is partly factual and partly based on reasonable inferences drawn from proven facts.
Moreover, it is not necessary to show that the exclusionary act completely foreclosed rivals from entering or accessing a market or segment of a market, it is sufficient to show that the exclusionary act “prevents or impedes a firm from expanding in the market”. InNationwide the Tribunal stressed that section 8(d)(i) did not require the showing of actual harm. It held that a finding of abuse could be arrived at ” 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”.
In that case the Tribunal limited its enquiry to the question of foreclosure because it was of the view that there was no direct evidence that consumers were paying more for their domestic airline tickets or had made inappropriate or wrong choices. However the Tribunal did remark that this did not mean that no such effects had occurred. It concluded that SAA’s incentive scheme with travel agents satisfied this latter criterion on the grounds that the effect of the anti-competitive 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.
In this case the evidence suggests a similar conclusion.
Counsel for SAA made much of the lack of evidence of harm to consumers to justify a finding that SAA’s conduct had no anti-competitive effect. But it appears that evidence of this nature is difficult to find in the context of the airline travel business. In the first instance the asymmetry between travel agents and consumers would severely constrain a consumer’s ability to detect whether he or she was being offered the most appropriate flight, both in terms of price and schedule, by the agent. While that asymmetry may have been mitigated to some extent in recent times by the development of sophisticated internet tools, increased use of online services and other direct sale channels in later years, the evidence in this case confirmed that during the period under consideration the asymmetry still favoured travel agents. Hence a passenger would very seldom know what type of fare was available on which airline in any given moment.
In the second place the fare management system utilised by airlines either in order to improve yields or in response to competition from other airlines resulted in variable ticket pricing along the fare ladder. SAA for example employed a dedicated team to monitor prices of fares – either for purposes of maximising yield per flight or in response to changes in competitor’s prices. Price variations would often be implemented on an hourly or per flight basis, accompanied by variations in the restrictions applicable to that class of fare. Such changes would be captured on the GDS and would be immediately available to travel agents. Comair employed a similar system. Because of this and because of the limitations of internet tools during this period, a consumer would not be able to meaningfully compare prices on the internet with that offered by travel agents or vice versa.
The lack of this evidence does not mean that there was no actual harm to consumers. It is reasonable to infer that SAA’s incentive agreements and travel agent’s ability to influence consumer’s choices would have led to some consumer harm in the form of higher prices or reduced choice. It is also reasonable to expect that such harm may be less substantial in the time sensitive or price insensitive segment of the market than in the price sensitive market because consumers in this segment – who largely purchase their tickets from travel agents - would be more willing to pay a higher price in exchange for comfort and flexibility.
For purposes of our enquiry it is sufficient for us to show that SAA’s incentive scheme had the potential or did in fact foreclose its rivals in the domestic airline market. As we stated earlier there is ample authority to be found in abuse cases where a relevant market is defined by looking to see where the effects of the exclusionary act are to be found. While we have declined to firmly conclude on segmentation of the domestic airline travel market, logic dictates that the anti-competitive effects of SAA’s agreements would impact largely on that part of the domestic airline sales which rely on travel agents for distribution, namely the TAS segment. It matters not whether those travellers were business or leisure, time sensitive or price insensitive. All of these travellers relied on travel agents, rather than the internet or call centres, for their flight arrangements and were therefore exposed to the incentives of travel agents.
The only fares that would not fall into this market would be those sold exclusively through the internet and other direct channels. By the end of the relevant period, the size of domestic air travel market sold through travel agents constituted approximately 60%-70% of airline ticket sales, and amounted to approximately R3,3bn per annum.
Relative performance in terms of sales and yields
Comair submitted evidence on the relationship between yields, flown revenue and flown passengers to demonstrate that SAA’s incentive scheme had enabled it to carry more high yielding passengers and had an anti-competitive effect on its rivals.
Comair submitted that it was an equally efficient rival to SAA and had similar yield management systems, a high reputation through its relationship with BA, international alliances, feeder passengers, business class and lounge and in-flight services, However its relative performance in terms of overall sales and yields was much lower than that of SAA. The inference to be drawn from this was that SAA’s higher yields were obtained as a result of its incentive agreements with travel agents.
Yields are a measure of average revenues per passenger in the airline business and are calculated by dividing the total flown revenue by the number of passengers. The higher the yield the higher the revenue earned by the passenger. Total yields would be calculated for total revenues in a particular year. However a calculation of yield could be done for each flight or a particular route for any period of time.
A comparison of the relative performance of Comair and SAA between 1999/00 and 2004/5 is shown in the Table 7 below on the basis of overall flown revenues, overall flown passengers and yields on Comair’s domestic routes:-
Table 6: Comparison of SAA and BA/Comair overall performance during the period 1999/00 and 2004/05
2004/05 and 1999/2000
Flown revenues (Rm)
Flown revenue (%)
Flown passengers (‘000)
Flown passengers (%)
Table 6 shows that SAA significantly outperformed Comair throughout the total abuse period. SAA’s total flown revenues grew by almost 40% over this period representing a three-fold difference in absolute revenue growth. SAA also outperformed Comair in terms of growth of passengers and yields. This means that it flew more passengers at significantly higher average prices than Comair during this period. The inference to be drawn from SAA’s superior revenue and yield performance over Comair is that SAA had been able to capture more of the high-yield passengers than Comair as a result of directional selling pursuant to SAA’s incentive agreements with travel agents.
Comair argued that these figures ought to be computed across the overall abuse period because SAA’s override incentive agreements were in place throughout this period. The reason why SAA had continued with its third generation agreements was to “lock in” the gains it had made in the earlier period. It was therefore necessary to examine the evidence across the entire period and not simply analysing data within such period. In general we agree with this proposition.
SAA contended that yields are very sensitive to any number of factors such as frequency, network size, capacity, sophisticated yield management systems and distribution of premium seats versus economy on a flight. It argued that SAA had recently implemented an improved yield management system which would account for its better performance. It argued that the launch of Kulula, which had impacted on SAA, also had had a cannibalising effect on Comair which had led to a decline in its performance; and lastly that SAA had been more affected by LCC competition than Comair. SAA had not put up any supporting data or documents for these arguments.
SAA’s own documents confirm the fact that it earned higher revenues because it carried proportionately higher yielding passengers than its rivals. According to SAA, LCC entry had had a significant impact on the SAA domestic market. These low cost operators had by January 2006, grown to take more than 30% of the market in the last three years. Since 2003 the domestic air travel market had grown by almost 50% while SAA passenger numbers had grown by less than 5%. SAA’s market share had slumped from 60% to 45% by January 2006. However while its market share had dropped, its revenue share of the market was higher than its passenger share because it carried more high yielding passengers.
A fair amount of time was spent on debating the relationship between flown revenue, flown passengers and yields. Notwithstanding all of this, in our view the relationship between yield, revenue and passenger numbers is not an easily explained one because all three are highly sensitive to the factors highlighted above and on issues such as seasonality and distribution effects. Nationwide for example always performed better in the peak holiday seasons. Both Comair and SAA offered more business class seats on their flights than Nationwide. Comair itself experienced an increase in revenue and passenger performance during 2004 when it increased capacity on its routes. These trends tend to make us wary of placing too much reliance on the relationship between yields, revenues and passengers as an indication of anti-competitive effects – whether as harm to consumers or of foreclosure - without considering it in the context of all the other evidence put up in these proceedings.
We have already held above that any foreclosing effect of the 3G agreements together with the trust agreements were likely to have resulted in substantial foreclosure for Nationwide and Comair in the TAS segment of the scheduled domestic airline travel market.
A significant difference between this case and Nationwide is that we have evidence of increasing market shares for Comair and Nationwide in the wider market definition which includes Kulula, and decreasing market shares for SAA. However these market shares do not necessarily reflect the underlying dynamics occurring in a market. When considering exclusionary conduct in particular, market shares on their own, are not necessarily a reliable indicator of where the effects of an abuse occur. We know for instance that the domestic airline travel market during this period was a growing market. Hence SAA’s declining market share during this period as depicted in Dr Affuso’s slides cannot simply be interpreted to mean that SAA was losing market share to its rivals. Nor can the increasing shares of its rivals suggest a lack of anti-competitive effect. In such a case an excessive reliance on market shares could in fact be misleading.
A closer look at the wider market shows growth was largely driven by LCCs and was to be found in the lower fare parts of the market. While Nationwide and Comair, through Kulula, were actively participating in this segment of the market, SAA by its own account was not participating in that growth. However, during this period the segment of the domestic airline travel market sold through travel agents remained relatively stagnant. Hence the growth of LCC’s and low cost fares accounted for a large part of the growth in the market and accordingly the increase in market shares of those airlines participating in that segment of the market.
We must bear in mind that the complaint from Comair and Nationwide was not that they had not experienced growth at all but that they had been excluded from a particular segment of the market. Both alleged that SAA’s agreements with travel agents impeded their growth in that segment of the market. As we have said previously, foreclosure of rivals does not require a showing that rivals are completely foreclosed from entering or accessing a market or segment of a market, it is sufficient to show that they were prevented or impeded from expanding in the market or in a segment of the market which was still distributed through travel agents (TAS). All the evidence of the witnesses in this case thus far suggests that SAA’s rivals were prevented or impeded from expanding in the TAS segment of the market by SAA’s incentive agreements with travel agents.
Counterfactual period evidence
SAA contended that even if Comair and Nationwide had been foreclosed from the travel agents market this did not result in any significant foreclosure in the domestic airline travel market as evidenced by both experiencing increasing market shares in the counterfactual period.
During her testimony Dr Affuso presented the Tribunal with a graph showing that the market shares of both Comair and Nationwide remained substantially the same before and after the abuse period. On this basis she concluded that the agreements could not have had a significant anti-competitive effect. We have attached her exhibit as Annexure 4 to these reasons.
This is the first time that the Tribunal and the applicants had been presented with this evidence. Nowhere in her witness statements and reports, did Dr Affuso put forward this evidence or the diagrams she relied upon in her testimony. Dr Affuso’s exhibit aimed to show that post the relevant period, namely during 2005/2006, there was no change in competitive performance or market shares in respect of Comair and Nationwide after SAA changed its override agreements. The period 2005/2006, the period when SAA’s override agreements are no longer in existence, is referred to as the counterfactual period. She testified that if Nationwide and Comair’s allegations were true, then one would expect them to grow and increase their market shares after the change in SAA’s agreement structures post 2005.
In Nationwide, the Tribunal in dealing with the issue of counterfactual, stated the following:-
“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...”.
The mere fact that their market shares remain the same does not necessarily support a finding that Comair and Nationwide were not foreclosed from expanding in the segment of the market covered by the SAA agreements and distributed by travel agents. Needless to say Dr Affuso’s last minute production of this evidence resulted in a flurry of hastily prepared data sheets by both Nationwide and Comair in rebuttal and presented to the Tribunal in the closing moments of the case. This Tribunal had in the cause of these proceedings cautioned all parties about the high number of data exhibits presented to us on the spur of the moment which allowed limited time in to digest and explore such data meaningfully with witnesses. Accordingly we view this evidence and the lateness of its production in an unfavourable light and afford it little probative value.
In any event a closer examination of Dr Affuso’s diagram shows that the counterfactual evidence relied upon by her was of limited assistance to this Tribunal and possibly misleading. In the first instance Dr Affuso’s comparison of the relevant periods is inaccurate. A significant issue to highlight is that the conduct complained of in this matter was a continuation of the course of action SAA had adopted in 1999. As explained by Mr Viljoen SAA had embarked on a strategy with travel agents in order to gain market share in the airline travel market. SAA did not wish to reward travel agents for what he termed CPI or GDP growth but required them to actively promote and grow SAA in the market. SAA sought to do this through the introduction of override incentive agreements and the Explorer scheme.
In 2001 it amended these agreements on advice but still maintained the objectives of the agreements. The trust agreements were introduced to compensate travel agents for any loss of revenue occasioned by the amendments. In his testimony Mr Viljoen re-iterated the rationale of the 3G agreements and trust payments. Hence SAA’s objectives and the strategy it pursued, namely that of override incentive schemes with travel agents with commissions calculated on a back to rand one basis for reaching targets, albeit with minor changes, remained unchanged for the period 1999- 31 May 2005.
SAA only decided to adopt a new strategy with the launch of its own LCC in 2006. The period between the Nationwide decision and the relevant period in this case was not interrupted by an absence of SAA override incentive agreements. Hence any discussion of counterfactual periods must necessarily treat the period in which the abuses took place as contiguous. By this we mean the world in which the alleged abuse takes place stretches from 1999 to May 2005 and is not limited only to the relevant period identified in this case.
Furthermore in order to construct a meaningful comparison such an exercise would need to involve controlling for other drivers of performance. For example, the domestic airline travel market was in a growth phase, such growth driven mainly by the introduction of LCCs which intensified competition in the low cost segments. It also saw the introduction of new players such as 1Time. It would also need to control for any ongoing effects of the 2nd generation agreements, namely the “lock in” effects.
As it appears from her cross examination, Dr Affuso’s calculations do not seem to have taken any of these factors into account. Although her exhibit refers to the period utilised by her as 2001-2002, a closer examination of her data and analysis revealed that her analysis begins in January 2002, excluding even the latter half of 2001, which is the early part of the relevant period in this case, and ends by capturing growth until the end of December 2005. Her analysis also failed to capture any growth experienced by SAA and Nationwide in 2001 and Nationwide’s significant growth after May 2005. For these reasons we find her evidence on the counterfactual period particularly unreliable.
Nature of agreements
In our view the effect of the agreements in this case are no different to those considered by the Tribunal in Nationwide. Viljoen himself confirmed that the 2001 scheme sought to maintain the level of incentives travel agents previously enjoyed. The nature of the 3G override agreements together with the trust payments were substantially the same as the second generation agreements. This is supported by the BSP figures presented by both Comair and Nationwide. At a theoretical level, the anti-competitive effects of these agreements have largely been demonstrated by Oxera, in both this case and the Nationwide case. The evidence of all the witnesses, together with that of CRA and Oxera demonstrated that despite SAA’s amendment of the override agreements, the override payments to achieve target (base) still provided travel agents with the same incentives as those contained in the second generation agreements. Hence the foreclosing effects of the third generation agreements together with the trust agreements were likely to be substantial.
We have already shown that during the relevant period travel agents still constituted the single most important route to market in South Africa. All the witnesses, including Mr Viljoen testified that in South Africa, passengers, especially corporate travellers, preferred to utilise travel agents. While the internet and direct channels had increased, only very low fares or discounted corporate fares were usually distributed through these channels. Qualitatively the high margin shares were still distributed through travel agents.
The evidence also shows that travel agents engaged in directional selling in order to maximise their profitability. As evidence of foreclosure effects one would expect to see a decline in Comair’s share of travel agent sales over the period, despite the fact of its override agreements with agents. Dr Affuso presented the Tribunal with data which according to her showed that the respective proportions of total flown revenues sold through travel agents were similar for SAA and Comair. She concluded that this suggested that the cause of any decline cannot have been a set of agreements affecting only one distribution channel but was due to other factors. Comair however was able to demonstrate that the evidence relied upon by Dr Affuso did not support her conclusion and that the same data relied upon by Dr Affuso demonstrated that Comair’s share of total flown revenues sold through travel agents actually dropped by 16% whereas SAA’s shares dropped by only 6%. This would be consistent with Comair being foreclosed from the market.
Dr Federico produced an exhibit in which Comair’s share of BSP sales with the five largest travel agent groups through time was compared against Comair’s share at Tourvest. This is attached as Annexure 2 referred above. The bottom graph shows Comair’s share at travel agents who primarily supported SAA until FY 2005. The top graph represents Comair’s share at Tourvest, who had concluded an incentive agreement with Comair and not SAA. Dr Federico explained that it was not the precise numbers but the gap between Comair’s share of Tourvest (who had not concluded an agreement with SAA) and Comair’s share of travel agents who had concluded agreements with SAA that was relevant. Also of interest is not that there are movements along each of the graphs representing Tourvest but that thedifference between the two graphs during the periods 2002/03 – 04/5 remains relatively stable ranging between 13 and 14%. He submitted that Comair’s share of travel agents, calculated in BSP, who did not have override incentive agreements with SAA, was significantly higher than with travel agents who did have override incentive agreements. In Dr Federico’s view this gap demonstrated the loyalty inducing effects of SAA’s override agreements and trust payments.
Dr Affuso produced her own version of Dr Federico’s graph and argued that if travel agents’ ability to divert sales resulted in foreclosure effects we would have seen an incline in Comair’s share of the travel agents represented in the bottom graph. She did not explain why this ought to be the case. We have attached her exhibit as Annexure 5.
We have already expressed our view on the difficulties inherent in creating hypothetical counterfactuals. As we stated above, the basis of Dr Affuso’s bald submission that we ought to observe an incline in Comair’s share of travel agents in the later counterfactual period (2006/2007) was never fully explained, and on reflection understandably so. We have no insights as to the prevailing market conditions during the later counterfactual period. All we know for a fact is that during this period all of the airlines had dispensed with incentive agreements. It would be impossible to construct a counterfactual that enabled a meaningful comparison between these two periods.
In our view it is preferable to evaluate the economic evidence, as presented above, in the light of other factual evidence presented to us. Dr Federico’s evidence of the foreclosure effects of SAA’s agreements is supported by two significant facts. In the first instance, during this period, while both Comair and Nationwide were seeking to increase their distribution through channels other than travel agents, both were still heavily reliant on travel agents, as demonstrated by their sales and efforts through travel agents. Even though Comair had launched Kulula, it had maintained it as a separate brand alongside its legacy airline utilising a low cost model in that business. As far as its legacy brand was concerned it still persisted in seeking new business. To this end we see that it increased its capital investment in the price insensitive segments by introducing increased capacity. Bricknell in his evidence expounded on the type of influence travel agents have in the industry. He further stated that Nationwide had gone through various strategies to continue its efforts to become a more effective competitor and grow its market share, and that despite all its effort, its market share did not grow as a result of travel agents’ influence in response to SAA’s incentive agreements. Both were also actively striving to conclude incentive agreements with travel agents.
In the second instance travel agents themselves confirmed that the effects of the foreclosure were likely to be substantial. Sure Travel moved all of its discretionary business away from SAA in favour of Nationwide and Comair after SAA decided to do away with its override incentives. Nationwide’s fortunes at Sure Travel changed drastically as a result of this shift climbing significantly from 10% to 14.5% of BSP share at Sure Travel in the last quarter of 2004. Nationwide’s fortunes at Tourvest on the other hand remained unchanged because Tourvest was not willing to support Nationwide until they had reached their targets with SAA.
During this period SAA had concluded override agreements with most of the largest travel agency groupings as well as with many smaller agencies. While it may not have had agreements with each of the agencies in every year of the relevant period, its agreements with the larger travel agents remained intact throughout this period which represented 70% and 90% of the airline sales distributed through travel agents. The larger travel agents or the agreements represent a significant size of the market. From this it is reasonable to infer that the SAA’s agreements had the potential to significantly foreclose the airline travel market from Comair and Nationwide.
Thus we can conclude that foreclosure of its rivals by SAA in the domestic airline travel marketwas likely to be substantial and that this impact would have been greater on that segment of the market which was distributed through travel agents and which consisted of the higher price fares.
Given the differentiated nature of the products offered by these airlines, we can expect that such effect would have impacted on each of them differently. Comair argued that because it was SAA’s closest rival the impact of such foreclosure would be greater for Comair than Nationwide.
We know that Nationwide experienced greater difficulties in being able to gain a foothold in this market. Bricknell testified that in some cases Nationwide could not even get a foot in the door and travel agents flatly refused to conclude agreements with Nationwide on the basis that they would not be able to support more than one “preferred” partner, that being SAA. Does this mean that Nationwide was not foreclosed as a result of the nature of SAA’s agreement with travel agents but was foreclosed because of the existence of such agreements? Not in the least.
Bear in mind that we are concerned here with SAA’s conduct and the nature of SAA’s incentive scheme. It matters not whether Nationwide or for that matter Comair had an agreement with a particular travel agent or not. All fares distributed through travel agents were available on the GDS. Hence a travel agent, irrespective of whether it had an agreement with a particular airline, would have simultaneous access to price and availability of tickets of all the airlines distributing through the GDS. In order to achieve the targets it had agreed with SAA a travel agent could simply promote SAA products on the GDS. It was the nature of the SAA override agreements offering large financial incentives that induced travel agents to favour SAA above its rivals’ products, irrespective of whether or not they had concluded incentive agreements with rival airlines. The fact that Comair may have been a closer rival to SAA does not mean that Nationwide was not foreclosed. On the contrary it could lead to a conclusion that SAA’s agreements had a greater impact on Nationwide than on Comair precisely because it was a more distant rival.
Hence the fact that Nationwide was unable to conclude incentive agreements with some travel agents, does not alter our conclusion that both Nationwide and Comair were foreclosed by SAA’s conduct. It is not necessary for us to make any findings on the relative impact of foreclosure on Nationwide and Comair. It is sufficient for us to find that SAA’s conduct had a significant anti-competitive effect on both of them in that it impeded their growth in that segment of the domestic airline travel market distributed through travel agents.
SAA relied on Nationwide’s poor safety record as a reason for its poor performance in the TAS segment. Nationwide had received a fair degree of bad publicity which had been exacerbated by end of 2007, and Mr Bricknell’s protestation to the contrary notwithstanding, appeared to have a poorer safety record than its competitors. Surprisingly, despite the negative perceptions of its safety record and financial upheavals faced by Nationwide, it achieved good growth. Its flown revenue market share increased from 6% to 9% from 2001 to 2005, representing growth of 50% over a four year period. Nationwide’s entire fleet was grounded sometime in 2008 after it had experienced what Mr Bricknell euphemistically referred to as “engine separation” which ultimately led to its liquidation. However the fact that Nationwide may have had a poor safety record does not mean that SAA’s agreements did not have a foreclosing effect on it.
Rivals could not match
That the foreclosure was likely to have been substantial is further supported by the fact that rivals could not match the incentives paid by SAA. In its supplementary report, Oxera considered what might be considered ‘best practice’ in analysing the effects of retroactive rebates. The rebate test suggested by EU Guidance Paper looks at whether the effective price is below average avoidable cost (ACC) and that sufficient economic data relating to cost and sales prices must be available. Essentially the Commission will seek to examine economic data relating to cost and sales prices, and in particular whether the dominant undertaking is engaging in below-cost pricing. In our view the EU test for predatory rebates is not applicable to the case at hand, and the EU guidance is of limited relevance because it deals with rebates that are offered directly to final consumers or retailers who have the ability to set prices. SAA’s payments to travel agents are clearly not rebates which directly or indirectly benefit consumers by stimulating greater price competition between agents. Agents have no ability to determine the ultimate price to consumers. Commissions paid to agents are an input cost for airlines.
Nevertheless both Nationwide and Comair were able to demonstrate that the lower marginal incentives offered by SAA post base during this period did not alter the fact that rivals could not match the override incentives paid by SAA to agents to achieve base. Mr Bricknell testified that Nationwide, assuming it had a market share of 10%, would have to multiply whatever it offered travel agents as a flat commission by 10 in order to equate, in rand terms SAA’s turnover. This was prohibitively expensive for a small airline such as Nationwide.
SAA argued that the foreclosure in the domestic airline market could not be significant because the ability of travel agents to shift business away from SAA, given its natural dominance and frequent flyer programme, was only in the region of 3%. Hence a reduction of 3% market share would be insignificant foreclosure. Dr Federico demonstrated that even if for argument’s sake one would assume that the divertible size of the market (also referred to as the discretionary business) was only in the region of say 5%. On his calculations a shift across 80% of the domestic airline market represented by travel agents would cost roughly R160m and a 2.5% shift would cost R80m. For an airline as small as Comair this would be of the order of 8% of total revenue. A loss of this magnitude would have a large impact on the profitability of an airline. He testified that the airline business was a high volume low margin business and for a small rival small shifts in market share could seriously affect its profitability. SAA itself recognises that relatively small shifts in market share can result in substantial losses in revenue. In its Ultralite Business Plan it projects that a 10% loss of market share would result in R700m loss in revenue and reduce profit by R230m.
The answer to the question asked earlier - as to why SAA persisted with these incentive agreements and threw large sums of money at travel agents - lies within SAA’s own conduct. Recall that SAA had introduced its aggressive incentive scheme sometime in 1999. The scheme in existence from 1999 to April/May 2001 consisted of the second generation incentive agreements and the Explorer programme.
In April 2001 SAA implemented an amended incentive scheme, consisting of the third generation agreements and trust payments. Viljoen conceded that SAA had introduced these amendments on legal advice and trust payments were introduced to compensate travel agents for the loss of income in consequence of these amendments. It continued with this scheme, with minor adaptations, until 31 March 2005. The third generation incentive agreements and trust payments were meant to achieve the same objectives of the previous agreements, namely to grow market share for SAA. This was the explanation given to us by SAA’s own witnesses.
Dr Niels suggested that SAA through these incentive schemes sought to immunise itself from the anticipated price competition from the launch of Kulula. Instead of responding to competition on the merits, SAA sought to offer aggressive incentives to travel agents in an effort to maintain and extend its dominance or market power in the airline travel market. Comair put forward a similar theory arguing that SAA’s second generation agreements and Explorer Scheme aggressively foreclosed rivals and the subsequent amendments were utilised to “lock in” the gains made in the earlier period.
In our view SAA’s persistence with its incentive agreements and the fact that its amendments sought to reward agents for maintainingbase or reducing the decline in SAA’s sales support both Dr Niels’ interpretation and Comair’s theories. A further factor that lends support to this interpretation is the case of the missing documents. In these proceedings, not a single internal strategic SAA document for the relevant period was placed before us. Board minutes for the years 2002- 2005 were apparently nowhere to be found.
Ms Zondo, SAA’s chief legal counsel, testified that she had not been directly involved in this matter and that for some inexplicable reason all the board minutes, presentations and other strategic documents for the relevant period could not be found. She undertook to continue the search. To date those documents have not been located. Given that SAA had sought to achieve these alleged savings at costs running into hundreds of millions of rands, we would have expected to see some internal documents detailing the decision to implement this strategy and ongoing review of its success or failure.
The absence of these critical documents raised more questions than answers. Did the executive obtain Board approval for its strategies when faced with critical market events such as the launch of Kulula? If so why was there absolutely no record of that placed before us? Are we to infer that SAA executives conducted business without any regard to corporate governance? Was the culture in that organisation such that it could expend billions of taxpayers’ rands to finance losses incurred without any approval policies and procedures or proper record maintenance? Mr Viljoen, quite opportunistically, in our view, attempted to cast blame on Mr Coleman, the erstwhile CEO of SAA, alleging that he had actually shredded key strategic documents before he left office.
We find it highly implausible that a multi-billion rand entity, entrusted with taxpayer’s money, bearer of the national flag, required to publish its results annually, required to report to a board consisting of highly reputable individuals and accountable to government, did not have such documents at hand or was unable to locate them. While we do not doubt Ms Zondo’s credibility, we can but only draw an adverse inference from this parlous state of affairs. Given the hundreds of millions of rands spent on travel agents in override incentives and trust payments we have no doubt that if there was any document within the business of SAA which tended to support Viljoen’s version that the agreements with travel agents had no or limited success such a document would have been placed before us.
Likewise, no evidence, apart from Mr Viljoen’s unsubstantiated claims, was placed before us that the third generation override incentive schemes and trust payments achieved any real efficiencies in ticket distribution for SAA or that such efficiencies outweighed the foreclosing effects of the scheme. Similar claims of efficiencies had also been made in the Nationwide complaint and were dismissed by the Tribunal. Indeed the fact that SAA terminated these agreements in March 2005, and decided to implement a zero commission arrangement with travel agents suggests to us that Viljoen’s claim of efficiencies in ticket distribution had no basis in reality at all.
In conclusion we find that SAA’s incentive scheme consisting of its third generation override agreements and trust payments with travel agents in effect from 1 June 2001 until 31 March 2005 constituted a contravention of section 8(d)(i) of the Act.
In 1999 - 31 May 2001, SAA’s incentive scheme, consisting of second generation agreements and the Explorer programme was held to be in contravention of section 8(d)(ii) of the Act. In April/May 2001 SAA amended that incentive scheme, and introduced its third generation override incentive agreements and trust payments. This scheme prevailed in the market place, albeit with minor amendments, until 31 March 2005.
Amendments to the scheme were done under legal advice and around the time of the ruling by the EC in the Virgin/BA case. On its own version SAA explained that it had introduced trust payments to “compensate” travel agents for losses they would incur as a result of the amendment. The stated rationale for this incentive scheme, consisting of third generation override agreements and trust payments, was no different to that of the earlier scheme.
We have examined this scheme in light of market developments during that period and have concluded that this scheme was in contravention of section 8(d)(i) of the Act and had resulted in ongoing foreclosure effects in the domestic airline travel market. SAA sought to “lock in” the gains it had made in the earlier period with this scheme.
During the relevant period we find that the launch of the low cost model Kulula created price awareness in the market and led to the growth of the domestic airline travel market. This development also promoted the use of cost effective distribution channels such as the internet. We see the emergence of nascent market segmentation between price sensitive /non time sensitive passengers and price insensitive/time sensitive passengers during this period. Despite these developments, travel agents remained the single most important route to market and distributed some 70% of total domestic airline tickets representing approximately R3.3bn. Internet and direct sales represented only 30% of the total domestic air travel market.
Through this incentive scheme, SAA sought to immunise its fares distributed through travel agents against competition and to extend its market power in that segment of the market. Travel agents had the ability to divert sales away from rival products and engaged in such practices in order to receive the handsome rewards for achieving the volume or revenue targets set by SAA. This inducement foreclosed SAA’s rivals from the domestic airline travel market, the impact of such foreclosure was likely to be greater in that segment of the air travel market distributed by travel agents. Rivals could not match the financial incentive, in rand value, offered by SAA. SAA had concluded agreements with approximately 70-90% of the airline sales distributed through travel agents which suggested that the foreclosure of rivals in the domestic airline travel market was likely to be substantial.
Instead of engaging in competition on the merits, SAA sought to extend its dominance in that segment of the domestic airline travel market distributed through travel agents which qualitatively represented higher margins with aggressive override incentives. While the foreclosing effects of its conduct were greater in this segment of the market, competition in the overall domestic airline travel market was reduced by SAA’s incentive scheme.
Given that we have found that SAA’s incentive scheme consisting of the third generation agreements and trust payments contravened section 8(d)(i) of the Act and resulted in or had the potential of foreclosing its rivals from the segment of the scheduled domestic airline travel market there is no need for us to conclude whether the scheme resulted in harm to consumer welfare. However the fact that SAA’s revenue share of the market was higher than its passenger share because it carried more high yielding passengers tends to suggest that consumers were harmed by paying higher prices or making poorer choices. Furthermore, no credible evidence of any efficiency achieved through this scheme was placed before us.
In its application Comair sought a declaratory order, as it was entitled to do in terms of section 49D(4)(a) read with s58(1)(a)(v) or (vi), that SAA had contravened section 8(d)(i), alternatively 8(c), by making trust payments to and concluding override agreements with travel agents. Comair also sought a declaration from the Tribunal, in terms of section 58(1)(a)(vi), that all the agreements, arrangements and/or understandings comprising such conduct from May 2001 onwards, are void. Since SAA had already terminated these agreements in March 2005, the latter relief sought was moot.
Nationwide sought a similar declaratory order. As far as its prayer for the imposition of an administrative penalty is concerned we have already stated that this was not properly before us and there is no need for us to consider this relief.
As far as the Comair application, in terms of section 49D, is concerned, the powers of this Tribunal in respect of the relief that can be granted are clearly delineated. Section 49D(4) provides that:-
“A consent order does not preclude a complainant from applying for –
A declaration in terms of section 58(1)(a)(v) or (vi); or
An award of civil damages in terms of section 65 unless a consent order includes an award of damages to the complainant.”
Section 58 deals with the orders of this Tribunal. Subsection (1)(a)(v) thereof provides that the Tribunal may “declare conduct of a firm to be a prohibited practice in terms of this Act for the purposes of section 65” and subsection (1)(a)(vi) provides that the Tribunal may “declare the whole or any part of an agreement to be void”.
A plain reading of the above sections shows that it is not competent for this Tribunal to impose an administrative penalty in respect of an application made in terms of section 49D(4) of the Act. The only relief that this Tribunal can grant in an application brought under section 49D(4) is limited to a declaratory order for purposes of section 65 and/or declaring agreements or parts thereof void. Moreover this procedure is only available to a complainant who has not been awarded damages in a consent order and who may wish to pursue its rights under section 65 of the Act. Comair, who was the complainant and who was not awarded any damages in the consent order has approached this Tribunal for precisely such relief.
Accordingly we grant the following order;
We declare the following conduct of SAA to be prohibited practices in contravention of section 8(d)(i) of the Act -
The override incentive agreements between SAA and various travel agents from 1 June 2001 to 31 March 2005; and
The trust agreements/payments between SAA and various travel agents from 1 June 2001 to 31 March 2005
An order of costs, including the costs of two counsel, in favour of Nationwide and Comair.
________ 17 February 2010
Y Carrim Date
Concurring: L Reyburn and M Holden
Researcher : Londiwe Senona
For Nationwide : Adv Gotz instructed by Bowman Gilfillan and previously by Bell Dewar and
For Comair : Adv Unterhalter SC with Adv Wilson instructed by Webber Wentzel
For SAA : Adv Subel SC with Adv Bhana instructed by Cliffe Dekker Hofmeyr Inc.