---

COMPETITION TRIBUNAL 

REPUBLIC OF SOUTH AFRICA

Case No: 08/LM/Feb02

In the large merger between: 

Distillers Corporation (SA) Limited Primary Acquiring Firm

And

Stellenbosch Farmers Winery Group Ltd Primary Target Firm

______________________________________________________________

Reasons for Tribunal Decision - Non-Confidential
______________________________________________________________


Finding

1. We have found that a substantial lessening of competition is likely in the proprietary spirits market, one of several markets implicated in this transaction. For the reasons outlined below we find that a consideration of the claimed efficiency gains is not pertinent. We have found that there are no consequences for the public interest that influence our finding.

2. A further hearing will be convened in order to determine an appropriate remedy in respect of that market in which we have found the likelihood of a substantial lessening of competition.

The Parties

Primary acquiring firm¹

3. Distillers Corporation (SA) Limited (“Distillers”) was a listed investment holding company, involved, through its subsidiaries, in the production and wholesale distribution of branded spirits, wine and ready to drink / flavoured alcoholic beverages (‘RTDs’ or ‘FABs’). Distillers produced, marketed, sold and distributed various brandy brands (including Oude Meester, Richelieu, Viceroy, Klipdrift), whisky (Harrier), vodka (Count Pushkin), cane (Seven Seas), premium wines (including Fleur du Cap, Le Bonheur, Neethlingshof and Grunberger), sparkling wines (J.C. le Roux) and liqueurs (Amarula Cream). The FABs manufactured and distributed by Distillers included Bacardi Breezer, Bernini and Castello. Distillers also acted as the South African agent and distributor of international brands such as Gordon’s gin, Martini, Bacardi rum, and Glenfiddich whisky.

Primary target firm

4. SFW was a producer and wholesaler of wine, spirits and alcoholic fruit beverages within South Africa. As a leading wine producer, it boasted names such as Nederburg, Zonnebloem, Graca, Chateau Libertas and Plaisir de Merle. Its spirit brands included Mellow-Wood brandy, Old Buck gin, Mainstay cane spirit and Romanoff vodka. It had the distribution rights in SA for Martell brandy² and Bols brandy³. It was the market leader in the FABs market with brand names such as Hunter’s Dry, Hunters Gold, Crown, Savannah, Esprit, Montello and Manhattan.

Shareholding structure

5. SFW and Distillers were both controlled by the same trio of shareholders. The two companies had an identical shareholding structure:

· Rembrandt-KWV Investments (“RemKWV”) held 60% of the shares of both parties. RemKWV is a joint holding company of Rembrandt and KWV, in which each holds a 50% interest. KWV’s interest in RemKWV is held through a listed subsidiary, KWV Investments Limited in which KWV owns approximately 54%;

· SAB held 30% of both companies through its wholly owned subsidiary Other Beverages Industries (Pty) Ltd (“OBI”);

· The general public held the remaining 10% of both companies.

Other significant participants in the production and distribution of alcoholic beverages

6. The merged entity’s most significant competitors in the production and distribution of spirits are GUDV and E. Snell & Co. GUDV is the South African subsidiary of multinational spirit producer Diageo, which was established out of the merger between Guinness and Grand Metropolitan. GUDV has the largest market share in whisky (including the J&B, Johnny Walker and Bell’s brands) and vodka (Smirnoff). It has smaller stakes in brandy, gin, and FABs. GUDV thus competes primarily with Distell in the middle and upper segments of the spirits markets. As elaborated below these are commonly referred to as the ‘proprietary’ or ‘prop’ and ‘premium’ spirits.

7. E. Snell & Co is a smaller South African company, which produces mainly, although not entirely, ‘value-for-money’ spirits – the low-price end of the market - including brandy (Wellingtons and Bols), whisky (Two Keys and Firstwatch), vodka (Absolut), cane (Cape to Rio), Gin (Strettons Deluxe Gin) and an alcoholic fruit beverage or ‘FAB’ (Snapper).

8. Douglas Green Bellingham (DGB), a long established South African company, is mainly a wine merchant, but its portfolio does encompass some well-know spirits brands in whisky (Balantine Finest) and brandy (Connoisseur). Brown and Forman, a major international liquor company, also distributes some of its important proprietary and premium spirits, notably Jack Daniels whiskey. The South African licensee of Brown and Forman brands is the Really Great Brand Company, which also performs distribution and related sales functions for E. Snell & Co. Other competitors include Seagrams (whose brands were subsequently acquired by Pernod Ricard), African Wines & Spirits and a large number of wine producers. The UK-based Bulmer, which has substantial international interests in cider, entered the South African market in 1999, when it acquired certain cider brands from Gilbeys. Bulmer exited the local market in 2002.

9. South African Breweries has a near-monopoly in beer where it enjoys a market share of approximately 95%. It has also recently begun producing FABs.

The Merger

The transaction

10. On September 20, 2000 Distillers and SFW entered into an agreement in terms of which Distillers would acquire, subject to the approval of the shareholders, the assets and liabilities of SFW, including the shares held by SFW in the issued share capital of Western Province Cellars Limited, SFW Holdings Limited, Bofor Properties (Pty) Ltd, and Devon Road Property (Pty) Ltd, and all the trade names and trademarks of SFW. The assets sold by SFW to Distillers included SFW’s shares in its operating companies and all its trademarks, but excluded certain specified assets. An addendum to the sale agreement was executed on 9 October 2000 (A171). Pursuant to the transactions, the merged entity was renamed Distell Group Limited (“Distell”).

11. Two common shareholders, Rembrandt-KWV Investments (currently known as Remgro-KWV Investments Limited) and South African Breweries, held 90% of the voting equity in both acquiring and target firm. The remaining 10% of each firm was held by the general public.

12. Post-merger the (simplified) share holding structure is as follows:

Remgro KWV Investments South African Breweries Public shareholders

60% 30% 10%

Distell Group Limited

South African Distillers and Wines

SFW Holdings Limited Others

History of the transaction

13. On 8 June 2000, the legal representatives of the merging parties approached the Commission and asked it to clarify whether the proposed transaction to merge the businesses of SFW and Distillers constituted a notifiable transaction. The parties’ essentially held that because of the common identity of the parties’ shareholders the transaction constituted an internal restructuring and not a merger as defined in the Act. On 7 August 2000, the Commission concluded in a letter addressed to the parties’ legal representatives, that the proposed transaction would not constitute a merger as defined in section 12 of the Act and accordingly was not notifiable in terms of section 13. Based upon this opinion, the parties proceeded to issue cautionary announcements advising of the proposed merger.

14. In terms of an agreement dated 20 September 2000 (amended on 9 October 2000) the merging parties effected a transaction whereby Distillers acquired all the principal assets and liabilities of SFW. The purchase consideration in respect of the SFW assets, in the amount of R515 157 950,31, was settled through the issue by Distillers to SFW of 55 580 000 Distillers ordinary shares in the share capital of Distillers. These Consideration Shares were distributed by way of a dividend in specie and reduction in share capital to the SFW shareholders.

15. Seagrams, a large multinational producer of various alcoholic beverages, subsequently launched an application in the Cape High Court on 10 November 2000 in which it asked the court to find that the transaction between SFW and Distillers constituted a merger in terms of the Act. The applicant sought an interdict restraining the respondents from implementing the merger, alternatively an order referring the matter to the Competition Tribunal. In his judgment, Jali J ruled that section 65(3) made it clear that the High Court did not have jurisdiction to hear the matter, insofar as it related to competition matters within the exclusive jurisdiction of the competition authorities.⁴

16. Bulmer SA (Proprietary) Limited (“Bulmers”) (the local subsidiary of another large multinational producer and distributor of alcoholic beverages) and Seagram Africa (Proprietary) Limited (“Seagrams”), both competitors of Distell, subsequently brought an application to the Competition Tribunal in terms of section 62(1) of the Competition Act 89 of 1998. The basis of the application was that the respondents failed to notify a transaction that the applicants contended was a merger as defined in terms of section 12(1) of the Act.

17. The Tribunal found that the transaction constituted a merger as defined in terms of section 12 of the Competition Act and ordered the parties to notify the merger to the Competition Commission. This judgement was upheld by the Competition Appeal Court on 27 November 2001.⁵

18. The merger was subsequently notified on 12 December 2001. The Competition Commission recommended in June 2002 that the merger be approved subject to certain conditions. Essentially, the conditions recommended by the Commission relate to the sale of a number of brandy and sparkling wine brands.

Rationale for the transaction

19. The parties claim that the merger will generate increased efficiencies that will enhance international competitiveness and shareholder value. In particular, they argue that, absent the merger, neither company could afford the intensive marketing strategies nor effectively manage the supply and distribution of alcoholic products overseas. The merged entity, on the other hand, would, through combining marketing budgets and by cost savings, achieved through economies of scale and reduced duplication, have a significantly greater chance of penetrating international markets. The cost savings would be realized by a rationalization of support services, manufacturing and distribution facilities and by reductions in working capital and fixed assets.

The Hearing

20. After the merger was referred to the Competition Tribunal in June 2002 a prehearing conference was held July 7^th 2002, and a hearing was duly convened, spanning five days in total:

• 15 August 2002

• 16 August 2002

• 22 August 2002

• 9 October 2002

• 15 November 2002

21. A total of nine witnesses were heard. These were called by the Competition Commission, the merging parties and the Competition Tribunal.

22. The following witnesses were called by the Competition Commission:

- Mr. Alistair Norman Lewis, AC Nielsen South Africa

23. The following witnesses were called by the merging parties

    • Mr. John Forsyth, partner McKinsey and Company

    • Mr. Johannes Jacobus Scannel, MD of Distell

    • Mr. Valerio Doriano Toros, Distell

    • Mr. Jacobus Hendrik Visser, Distell

24. By the Competition Tribunal:

    • Mr. Michael Clifford Veysie, MD Bulmers

    • Mr. Tim Hutchinson, CE Douglas Green Bellingham

    • Mr. David Hooper, MD E. Snell & Co

    • Mr. Colin Robinson, MD Robinson Liquors

Competition Evaluation

Background

25. The alcoholic beverages sector represents to competition folklore in South Africa, what, we imagine, the oil industry represented to those concerned with competitive markets in the USA at the turn of the last century. Not only do we have what is, to all intents and purposes, a single domestic beer producer, but we have a longstanding history of state intervention in the production of wine and spirits, intervention manifestly designed to support narrow private interests rather than the public interest, that is possibly unparalleled in its breadth and intensity. The 1982 Competition Board Report on the liquor industry notes:

“As early as 1918 a written agreement, a so-called ‘gentlemen’s agreement’ was entered into by the KWV and the wine merchants under which the KWV would refrain from competing with the merchants it supplied. Specifically, the KWV, as a quid pro quo for the co-operation of private entrepreneurs, undertook not to compete with the existing interested parties in the trade in or distillation or manufacture of wines and spirits in Africa south of the equator.”

26. And, further: “In 1924 Parliament incorporated this principle in Act 5 of 1924, so that the KWV is not allowed to sell any wine or spirits to any person not being a bona fide distiller, wholesaler, association of distillers or wholesaler or co-operative society”.⁶

27. However, the apotheosis of anti-competitive conduct in this sector is surely the agreement which secured South African Breweries’ beer monopoly and the Rembrandt Group’s pre-eminent position in the spirits, particularly the brandy, market. We refer, of course, to the notorious market sharing arrangement between the beer producer and its counterpart in the wine and spirits sectors that saw the former agreeing to limit its involvement in wine and spirits in exchange for an undertaking from the Rembrandt group to stay out of the beer market. To add insult to injury, KWV was allowed to take up a significant share of the new spirits and wine company.⁷

28. The market sharing arrangement was effected by the restructuring in 1979 of the South African liquor industry. This involved an arrangement between SAB, SFW, OMG (effectively Distillers’ predecessor) and KWV, culminating in the formation of a new entity, Cape Wine and Distillers Limited (“CWD”).⁸

29. CWD was listed on the Johannesburg Stock Exchange, its shares being allocated (and held) in the following manner:

- The Rembrandt group: 30%

- SAB: 30%

- KWV: 30%

- General public: 10%

30. This restructuring was designed to facilitate a split in the liquor industry in terms of which:

    • SAB purchased the Rembrandt Group’s beer interests (Intercontinental Breweries, or ICB, the large brewery with which SAB had been in a price war that year);

    • SAB agreed to pool its wine and spirits interests (including SFW and Henry Tayler & Ries) via the CWD and to limit its involvement in wine and spirits to its 30% investment holding in CWD;

    • The Rembrandt Group sold its wine and spirits operations (the Distillers Corporation Ltd and the Oude Meester Group and thereby the retailers Western Province Cellars and Liquortown);

    • The Rembrandt Group undertook to have no interests in the beer market;

    • Rembrandt and SAB undertook to divest their retail liquor interests; and

    • SFW, Distillers and OMG become wholly-owned subsidiaries of CWD.

31. This restructuring effectively meant that SAB sacrificed its wine and spirits interest to CWD in return for a beer monopoly and a stake in CWD, whilst spirit and wine production was concentrated in the CWD, which acquired the two largest producer-wholesaler bodies, namely SFW and Oude Meester Group (OMG).

32. Shortly thereafter, the Rembrandt Group and KWV pooled their shares in CWD (60%) in a jointly owned holding company, Rembrandt-KWV Investments Limited.

33. In 1982 the Competition Board (Competition Board Report no.10) recommended that the vertical integration in the liquor industry be prohibited and that the merger which had taken place in 1979, giving rise to the formation of CWD, be reversed:

”The competition that previously existed between SAB and ICB and between the two largest producer-wholesalers of wine and spirits SFW and Oude Meester, has inevitably been either terminated or restricted by the restructuring”

And further: ”The Board is convinced that the circumstances described … do not justify in the public interest the KWV’s acquisition of an interest in CWD”.

34. This recommendation was rejected by the government of the day.

35. However, in 1988, the then Minister of Economic Affairs supported a separation of the two main components of the CWD, namely SFW and OMG, reportedly motivated by a desire to enhance competition. The separation was effected by a separate listing on the Johannesburg Stock Exchange of SFW and a new entity, named the Distillers Corporation SA Limited, comprising the interests of OMG. This event constituted a partial reversal of the 1979 restructuring that had created a concentration of wine and spirit interests within a single corporate structure.

36. It is undoubtedly the breathtaking audacity of these manifestly anti-competitive agreements and their endorsement by the political powers of the time, that accounts for the persistence of anti-competitive structures in the alcoholic beverages sector and for the intensity of the disquiet articulated by consumers, distributors, the current government and, in particular, other, inevitably smaller, producers at the state of affairs in this industry. However, while the structure of the industry that has emerged as a result of these agreements undoubtedly demands an unusual degree of vigilance from the competition authorities, we cannot use the provisions of the Competition Act to turn the clock back, to redeem, ex post facto, the sins of the past. We are, regrettably, obliged to take the structure of the industry as we find it and, in merger proceedings at least, to limit our interventions to those transactions that result in a substantial lessening of competition.

The Relevant Markets

The Geographic Market

37. It is common cause between Distell and the Competition Commission that the relevant geographic market is national. The Tribunal agrees that the relevant geographic market is indeed the South African market and this issue will not be considered further.

The Product Market

38. In common with other contested merger proceedings, the main area of contention between the parties and the Competition Commission centres on the identification of the relevant product market. In short, whereas the merging parties contend for a broad product market that encapsulates all alcoholic beverages, the Commission prefers a narrower definition that places each traditional category of alcoholic beverage – brandy, whisky, wine, etc – in separate relevant markets.

39. The Commission defines a variety of relevant product markets based on traditional liquor categories, including spirit type, different types of wine (table wine, sparkling wine and fortified wine) and a market for flavoured alcoholic beverages or FABs.⁹ The Commission finds product overlaps in the following markets:

Table 1: Market shares and HHI changes per liquor category

Liquor category	Post-merger Market share (% sales value)	Post-merger HHI	HHI increase due to merger
Whisky	11.8	2487	65.6
Brandy	71.7	5366	1941.1
Vodka	16.6	4565	130.3
Cane	39.4	4309	424.7
Gin	73.2	5147	1542.9
Table wine	59.5	N/a	N/a
Sparkling wine	74.4	Approx 5580	Approx 2710
Fortified wine	80.8	N/a	N/a
FABs	61.8	4793	861.4

Source: AC Nielsen data and Competition Commission recommendations

40. While, as will be elaborated below, the Commission found that the impact on competition of the horizontal overlap in most of these markets is ameliorated by other factors – for example the unusual dynamism of the FABs market or the merged entity’s relatively small market share in whisky – the Commission’s narrow, category-based market definition was the basis for its finding of a substantial lessening of competition in the brandy, sparkling wine and gin markets and, accordingly, for its recommendation that the merged entity be compelled to divest itself of a number of brands in these markets.¹⁰

41. In the brandy market the Commission recommended that Distell be compelled to dispose of 16% of its market share whilst terminating the manufacture and distribution of all KWV brands. In the sparkling wine market the Commission recommended that Distell dispose of brands with a cumulative market share of between 20-30% in volume.

42. The merging parties on the other hand define the relevant market to include all alcoholic beverages, ranging from beer to spirits, including wine and FABs. The parties find that Distell’s post-merger market share in the national alcoholic beverage market, based on litres absolute alcohol, is 19.7%.¹¹

43. Accordingly, the parties have argued that there is no substantial lessening of competition in the relevant market and suggest that, even if there was, the efficiencies generated by the merger would offset any detrimental effects of the merger.

44. Much hinges then on the identification of the relevant product market. Unfortunately, however, the Commission has produced scant evidence in support of its view of the relevant product market. For the most part, it elected to support its case through a critique of evidence produced by the parties and through examining witnesses called by the Tribunal. It is not surprising then that counsel for the merging parties should have raised, at the commencement of the hearings, the question of onus, arguing that it is not for the parties to prove that competition will not be substantially lessened by the merger, but rather for the Commission to establish the likelihood of a substantial lessening of competition.

45. However, the question of onus is not as clear-cut as the parties would have us believe. The Tribunal is the decision maker in respect of all large mergers. It is, in other words, required to establish not whether some participant has discharged an onus, but, rather, whether or not there has been a substantial lessening of competition. This it will do on the basis of the evidence and argument submitted to it, including evidence garnered through the Tribunal’s exercise of its inquisitorial powers. It will, indeed, become apparent that our reading of the evidence and argument that we have heard has led us to a view of the relevant market distinct from that of both the Commission and the parties.

46. That having been said then, the Commission has argued for a particular conclusion, namely, that the transaction is likely to substantially lessen competition, and the remedies proposed by it consequent upon this finding embody potentially important consequences. In doing this the Commission has relied almost entirely on a critique of the parties’ own arguments and the evidence of the parties’ witnesses. However, a critique, no matter how trenchant, of the parties’ argument and of the evidence led by them may establish that the parties are wrong; but it cannot, on its own, establish that the counter-argument is correct. In short, for an adverse finding the Tribunal must be satisfied that the evidence and argument that has been presented, whether from documents discovered or oral evidence led by the Commission, the parties and the Tribunal, affirms that the transaction is likely to lead to a substantial lessening of competition. Evidence presented by the Commission has made but a small contribution to meeting this standard.

47. To return then to the definition of the relevant market, the Commission’s finding of a substantial lessening of competition in the brandy and sparkling wine categories rests heavily on its insistence that there is a range of narrow relevant markets defined by traditional liquor categories including spirit types (e.g. brandy, whisky, vodka, etc), different categories of wine (e.g. table wine, fortified wine and sparkling wine) and a separate market for FABs.¹² In short, the Commission avers that faced by an increase in the price of brandy, consumers will not switch to another spirit category or, even less will they switch to one of the other categories of alcoholic beverages such as wine, FABs or beer. It insists, in other words, that inter-category competition will not constrain an exercise in market power on the part of a producer whose market share in one or more of the separate categories increases as a result of this transaction.

48. The Commission’s contentions with respect to the relevant market rest on two pillars. The first is international jurisprudence which, the Commission pointed out, mostly supports the narrow, category based relevant markets contended for by the Commission. Secondly, the Commission insists that evidence presented to the Tribunal establishes the weakness of inter-category competition in the South African market. As we have already intimated, the overwhelming bulk of the evidence was presented by the parties themselves, although important evidence was also presented by witnesses called by the Tribunal. Hence, the Commission relies overwhelmingly on a critique, on its particular interpretation, of this evidence. The Commission called a single witness, Mr. Alistair Lewis, an employee of AC Nielsen, the well-known market research firm.

International Jurisprudence

49. The Commission avers that competition authorities and courts elsewhere have, when confronted with the task of establishing a relevant market in the alcoholic beverages market, found for narrow, category-based markets. The Commission particularly relies on US and European decisions in two prominent mergers, namely, the Guinness Plc and Grand Metropolitan Plc (1997) and Pernod Ricard/Diageo/Seagram Company Ltd (2001). These mergers were evaluated by the European Commission¹³ and, in the case of Guinness/Grand Metropolitan, also by the US Federal Trade Commission. The Guinness/Grand Metropolitan merger was also investigated by the Australian ACCC.

50. In the Guinness / Grand Metropolitan transaction the European Commission based its market definition on spirit type. This conclusion rested, inter alia, on the finding that spirit drinkers display brand loyalty within the category of choice and also on the observation of well-entrenched ‘occasion-based’ consumption patterns which renders substitution on the basis of small price variations unlikely. The European Commission highlights the importance to competition in the spirits market of branding and its application to individual spirit types as a further justification for product markets not wider than that for each main spirit type, i.e. whisky (further segmented to differentiate Scotch whisky), brandy (further segmented to differentiate Cognac/Armagnac), rum, gin, vodka, tequila and flavoured spirits.

51. Note that the European Commission conceded the possibility of market segmentation based on price and quality differentiation observing that ‘a consumer who habitually drank a premium brand would not regard a cheaper one as providing an adequate substitute in terms of taste, image and so forth’.¹⁴ However, this observation did not, in the view of the European Commission, alter the finding that placed individual spirit categories at the centre of the relevant market definition but rather constituted the basis for a further segmentation, this time within the separate spirit categories.

52. In its assessment of the Guinness/Grand Metropolitan merger, the Federal Trade Commission focused on whisky and gin, and particularly on the premium segments within those categories. Within the whisky category it distinguished between origins (i.e. Scotch Bourbon and Irish) as well as different quality and price categories. The definition of premium gin included a reference to its origin (i.e. England) and a retail price level, comparing prices of specific brands.¹⁵

53. The ACCC found limited demand substitution between various spirit categories, with price increments in a particular category tending to result in brand shifting rather than a reduction in sales in that category, supporting a relevant market definition based on the spirit categories.

54. In the Pernod Ricard / Diageo / Seagram Company Ltd transaction the European Commission cites the Guinness/ Grand Metropolitan judgement in its relevant market definition and re-emphasises the importance of branding. Although the European Commission notes the possibility of defining the market according to different price/quality combinations, i.e. premium, secondary brands, private labels, low price etc, the market definition in this judgement is at one with the finding in Guinness / Grand Metropolitan. The European Commission concludes that as ‘there is a continuous price spectrum ranging from the most expensive to the cheapest’ and as ‘rebates can change the position of a given brand in the spectrum’ the price/quality distinction was only found applicable to the exclusion of Cognac/Armagnac from the product market that otherwise included all brandies.¹⁶

55. The Competition Commission insists that we would need particularly powerful contrary evidence to row against the tide of international opinion that has found narrow category-based markets. We are mindful of this. Indeed Section 1(3) of the Act explicitly empowers those interpreting or applying the Act to have recourse to international jurisprudence, a wise provision given the immaturity of our jurisdiction relative to those who have worked with competition law and economics for many years. However, whilst foreign jurisprudence may be, indeed it certainly has been, of great assistance in refining our understanding of legal questions and economic theory and in guiding our factual enquiries, it cannot detract from the strong factual basis that must ultimately underpin all efforts to determine a relevant market. Counsel for the merging parties cites an extract from our decision in the large merger between Bromor Foods (Pty) Ltd. and National Brand Limited:¹⁷

“Defining a relevant market for consumer products is notoriously difficult. Delineating a relevant market for beverage products is especially difficult because one is faced with not only the subjective proclivities of consumers but also the marketing stratagems of firms as they attempt to differentiate their products in response to competitive threats.

Beverage antitrust cases have long been the subject of bitter contestation over relevant market definition. On the one hand merging parties contend they are merely minor players fighting for their "share of the throat", in a market where the fizzy drink competes with anything that can be imbibed from fruit juices to tea. On the other hand competition regulators argue that the fizzy drink is the relevant product market.

Ultimately each case must be determined on its own facts and foreign judgments can do no more than give us guidelines to method for they cannot serve as a way for us to come to a conclusion on facts. The behavior of a teenage consumer of carbonated beverages in Texas is no more use to us as evidence than the behavior of the French consumer of carbonated mineral water.”

56. It should be noted that the European Court of First Instance has itself explicitly insisted on the overriding significance of a current factual enquiry when determining relevant markets even to the extent of diminishing the weight of a prior finding of the European commission when making a subsequent relevant market determination or, what is the same thing, a finding of dominance:

“Second, a finding of a dominant position by the Commission, even if likely in practice to influence the policy and future commercial strategy, does not have binding legal effects as referred to in the IBM judgment. Such a finding is the outcome of an analysis of the structure of the market and of competition prevailing at the time the Commission adopts each decision. The conduct which the undertaking held to be in a dominant position subsequently comes to adopt in order to prevent a possible infringement of Article 86 of the Treaty is thus shaped by the parameters which reflect the conditions of competition on the market at a given time.¹⁸”

57. In short, while we will not simply ignore the US, European and Australian findings, the weight assigned them is reduced if the evidence indicates that the general features of the South African market differ significantly from those that characterize these other national markets. And, if evidence is adduced that directly conflicts with the inferences drawn from the general features of the market – in other words, if, for example, we are shown persuasive evidence of inter-category substitution in the South African market then this must, at least in respect of those categories that have been shown to be substitutable, surely trump a decision based on the general characteristics of the market.

58. That then is the sequence of the argument: we first consider the international jurisprudence which finds overwhelmingly for relevant markets defined by liquor category; we then ask whether the evidence demonstrates that the characteristics of the South African market approximate those of the markets that generated these findings; and we finally ask whether there is evidence of consumer behaviour in South Africa that is at odds with the inferences drawn from the general characteristics of the market. In this case, the evidence in point would relate to the issue of inter-category substitution.

59. The Competition Commission argument effectively holds that consumer behaviour in the USA, Europe and Australia is sufficiently similar to that of South African consumers of alcoholic beverages to justify the simple importation of conclusions regarding the relevant market from these jurisdictions. It would have us accept that if brand loyalty and occasion-based drinking – the two features upon which their relevant market findings are based – are prevalent in the USA, European and Australian markets, then they will be present in the local market as well.

60. However little evidence has been presented in support of this far-reaching and essentially factual assertion. And yet there are clear prima facie grounds for questioning its validity. South Africa’s particular income distribution and the absolute levels of poverty with which a large proportion of the liquor-consuming population contend is, per definition, an extremely powerful determinant of consumption patterns and behaviour, particularly where ‘discretionary’ consumption is concerned – little evidence is needed to assert the massive disparities between South Africa and the highly developed countries from whom the Commission would have us draw essentially factual conclusions regarding the relevant market.

61. Moreover, the truly unique features of South Africa’s liquor history - recall that until relatively recently the vast majority of South Africa’s population was, by law, not permitted to enter outlets at which spirits were sold – throw the Commission’s questionable proposition into sharp relief. Indeed the parties argue that the fact that the bulk of liquor purchased in South Africa is still sold and consumed in the semi-legal shebeen environment is evidence of the unusual character of the South African market. The rapid and massive shift in consumption away from sorghum beer to other alcoholic beverages is also a unique feature of the South African market. The single witness called by the Commission, Mr. Alistair Lewis of AC Nielsen, identified another distinguishing feature of alcoholic beverage consumption in South Africa, namely, the:

“huge wine market, which is not so predominant in other parts of the world. In other words, I’m not talking necessarily of quality wine. I’m talking about the bottom end of the market, which traditionally started here back in the sixties or even before that.”

62. Lewis nevertheless insisted that, but for this distinction, which strikes us as rather significant, the consumption patterns in the South African market match those found elsewhere.

63. The merging parties have, for their part, presented evidence of the distinctive features of South African liquor consumption. Hence they submit – and this was not contested by the Commission - that South African spirits consumers, in contrast with their developed country counterparts, rarely drink spirits neat, but rather use it to add alcoholic content to a mixer, so that the key attribute of the spirits is, in the minds of South African consumers at least, its alcohol content rather than its particular taste. This would portend a greater possibility of substitution, at least between spirit categories. The parties also insist that liquor consumption in South Africa, as opposed to other societies in which similar research has been conducted, is less ‘occasion based’, less structured by the time of day at, or occasion on, which it is consumed, an assertion borne out by the research conducted as part of the ‘brandy study’.¹⁹

64. The Commission purports to find support for its claim that “the broader South African alcoholic beverage market is not substantially different from the markets in the United States of America, Australia and all the member states of the European Community …” in research results which indicate that a significant proportion of the respondents consumes more than one kind of drink at a particular occasion. This, the Commission argues, is evidence of occasion-based consumption among South African liquor consumers.

65. However, this response confirms at best that more than one brand is consumed by a significant proportion of consumers on any one occasion. This does not show that the occasion determines the switch – indeed, in the absence of further evidence, it may be reasonably inferred from this that South African liquor consumers, in apparent contrast with their counterparts in the other jurisdictions cited, display little brand or category loyalty. In short, this evidence may well support the notion of a wide relevant market, in which consumers drink more than one type of drink at any one occasion.

66. Moreover, Mr. John Forsyth, the parties’ expert witness, testified that market surveys that attempted to relate specific occasions to particular liquor products using a sample of South African consumers, called into question the notion of occasion-based drinking in South Africa:

(Paragraph omitted, contains confidential information)

67. And, Forsyth concludes:

(Paragraph omitted, contains confidential information)

68. In summary, we find unpersuasive the Commission’s reliance on foreign jurisprudence in determining the relevant market. Certainly, the evidence provides little justification for the uncritical application of the European and US findings to South Africa. On the contrary, the available South African evidence suggests that there are important unique features of South African liquor consumption that will have an important bearing on the definition of the relevant market.

Inter-category Competition: the South African evidence

69. We have been presented with a confusing welter of evidence, some of it empirical and much of it anecdotal drawn from a combination of market survey and direct experience of the market. We are also faced with appeals – largely emanating from the Commission - to ‘common sense’, to the personal knowledge or, at least, personal opinion, that many have of this mass consumption market. The proponents of ‘common sense’ effectively ask us to accept that consumption of a particular category of alcoholic beverages is a matter of deep personal taste that will not be compromised by a mere increase in price. While an increase in the price of tea may plausibly give rise to a switch to coffee, a loyal brandy – or whisky or wine or beer – consumer will, in a manner of speaking, simply swallow the price increase, his expressed commitment to his alcoholic beverage of choice would inhibit him switching to another category or even decreasing significantly his overall consumption of his chosen alcoholic beverage.

70. While we cannot simply ignore these conventional wisdoms – the particular categories do, after all, have different tastes or, in the language of marketing specialists, ‘intrinsics’ – we cannot base a definition of the relevant market on these insights alone. It would introduce an intolerable degree of subjectivity and uncertainty into competition analyses, most particularly where consumer goods markets are at issue.

71. The evidence on substitutability presented by Mr. Lewis and on which the Commission sought to rely for its version of the relevant market related, firstly, to the form in which Nielsen collected data, and, secondly, to evidence of long term stability in the market shares of the various traditional categories within the broader alcoholic beverages market.

72. Lewis testified that Nielsen’s clients in the liquor industry do not generally and traditionally request information on liquor sales as a share of total spirit sales or of all alcoholic beverages. The services rendered to AC Nielsen clients in South Africa include a bimonthly presentation that incorporates an overview of economic trends in the SA market and a brief overview of the total liquor market. For the most part, however, Nielsen is required to report on the individual spirit categories. This, the Commission averred, constituted evidence that, in their daily practice, the actual participants in the liquor industry themselves worked with category-specific relevant markets. ‘Why’ it is implicitly asked ‘would they ask for information delineated by traditional liquor category if they genuinely believed that their products in these categories competed with products in all other categories, that is, with products in the alcoholic beverages market?’

73. Even if we assumed that the Nielsen experience confirmed that firms were predominantly interested in sales data in traditional liquor categories, it is by no means clear that we should be drawing the inference sought by the Commission. It is wholly possible to imagine motorcar manufacturers asking for data to be collected that would help identify the most popular colour vehicle that they produced but this would surely not justify a conclusion that placed red and blue vehicles in separate relevant markets!

74. Nor is it clear that Nielsen’s clients do, in fact, always require that data are collected in traditional liquor categories or, when tracked in traditional categories, that they use the data in this form. Firstly, Lewis acknowledges that Nielsen is no longer asked to collect separate data on gin, vodka and cane spirits but rather to aggregate these in a collective ‘white spirits’ category. Particularly interesting is Lewis’ acknowledgement that this has shifted over time, an admission that supports the parties’ notion of a dynamic market characterized by significant shifts as new consumers and new products enter the market and tastes change. Secondly, Lewis admits that a major Nielsen client, South African Breweries, requires data of the overall alcoholic beverages market and that it uses this in order to track the performance of its products, beer, and, recently, FABs, vis-à-vis spirits and wine. Again, this is identified as a relatively recent mode of collecting and presenting data.

75. The parties predictably deny that the form in which Nielsen data is collected has any bearing on the identification of the relevant market. They effectively suggest that this simply reflects a convenient mode of organizing masses of market data.

76. Moreover, the parties insist that their broad view of the relevant market is shared by many of their competitors. This is however only partly true. Hence SAB and GUDV, the large multinational producers of a broad range of alcoholic beverages, both consider their various products to be competing in the broad alcoholic beverages market. Snell, a locally based participant in the alcoholic beverages market, avers that it is ‘an acknowledged fact’ that ‘beer, wine, RTDs, ciders and spirits fiercely compete for liquor consumers’. On the other hand, Seagram, also a large multinational producer of alcoholic beverages, and DGB, a local producer and distributor, appear to support a narrow market definition. On the basis of its experience in a range of national markets and on a telephone survey of local FAB consumers, Bulmer, yet another multinational producer and recent entrant into the South African market, contends that FABs are a separate market, although it is not clear that this conclusion is supported by the results of the survey that they commissioned.²⁰ The merging parties also find support for their broad market definition in the views of industry analysts.

77. The parties claim that the experience of the ‘brandy study’ also confirms their view of the relevant market. This study was commissioned by Distell and undertaken by McKinsey, the large US based management consultancy.

78. Although initially conceived as a study of brandy consumption and hence dubbed the ‘brandy study’, Mr. John Forsyth, a McKinsey executive who testified on Distell’s behalf at the hearings, averred that it immediately became apparent to the McKinsey research team that the patterns of, and prospects for, brandy consumption could not be understood without locating it in a wider study of the consumption of alcoholic beverages generally. Forsyth testified that his company characteristically approached requests to analyse the positioning of particular products in the market by asking consumers a range of pertinent questions concerning the product under scrutiny. If, in the course of their survey of consumers, the respondents persistently mentioned other products, this signalled the necessity for widening the scope of the study to include these other products.

79. Hence, the ‘brandy study’ effectively became a study of brandy’s place within the broader alcoholic beverages market. In essence the study found:

(Paragraph omitted, contains confidential information)

80. In fact it found that a significant proportion of the population (accounting for a significant part of consumption) is prepared to change drinking patterns in response to a change in price or other product attributes.

81. While the finding cited in the preceding paragraph is, on the face of it, pertinent, we are provided with no further basis for this conclusion, nor are we told what liquor categories are referred to. Moreover, changes inspired by ‘fashion’ are likely to be of a longer-term nature – more akin to a shift of, rather than along, a demand curve – than are those inspired by price changes.

82. This goes to some of the reservations that we have about the brandy study for the purposes of conducting a competition analysis. It is a study manifestly designed to inform Distell’s marketing strategy and although it does, in the process, unearth insights and information that are of some indicative interest in a competition analysis, it is not a competition analysis. That it asks ‘how can I, predominantly a producer of brandy and other spirits, persuade beer drinkers to come over to my product’ or ‘what steps do I need to take to prevent the breweries wooing away my customers’ is not, on its own, the identification of a relevant market that includes both brandy (and other spirits) and beer (or, for that matter, wine).

83. It merely acknowledges that, new consumers aside, the most likely source of additional custom for any alcoholic beverage is to be found in the ranks of existing consumers of alcoholic beverages, rather than in the ranks of, say, church congregants. It simply says that those who are regularly to found in a bar or shebeen – rather than at Sunday School - constitute the most fertile ground on which to market an alcoholic beverage. Many of the patrons of a shebeen or bottle store – the vast majority, given current South African consumption patterns of alcoholic beverages – are beer drinkers. Hence, that beer drinkers should be a major target of a spirits’ producer is not surprising. It is, indeed, as little surprising as the converse – we would expect beer producers to be as much concerned with its ability to make inroads into the ranks of brandy consumers and this likely explains SAB’s interest in gathering data of alcoholic beverages consumption from AC Nielsen.

84. At most it may reflect that in the long term battle to change tastes, to develop both ‘intrinsics’ and ‘extrinsics’ that would boost the long term prospects of a particular type of spirit – that would move the demand curve for that particular type of spirit outwards – a spirits producer would target, inter alia, beer consumers, that great rump of consumers who already consume alcoholic beverages. It does not suggest that the current competitive strategy, particularly the pricing strategy, of a spirits producer is materially influenced by the competitive strategy of the beer brewers.

85. We have already dealt with the question of ‘occasion-based’ drinking, the overriding basis for the European decisions that confined relevant markets to traditional categories. The absence of evidence of occasion-based drinking in South Africa clearly opens the way to hypothesise that consumers would switch between categories in response to price movements and this hypothesis appears to be confirmed by the brandy study and a number of authoritative studies and observers.

86. The brandy study finds that ‘..consumers consume products from a number of categories and freely switch between categories’

(Rest of paragraph omitted, contains confidential information)

87. The Brandy Study also identifies a blurring of the division between traditional liquor categories.

(Rest of paragraph omitted, contains confidential information)

88. It is also noteworthy that the consumer category whose participants are most wedded to the occasion-based drinking patterns identified in the European cases represent a very small proportion of the adult South African population and that large proportion of the South African population (representing a significant part of liquor consumption) is prepared to change drinking patterns in response to a change in price or fashion.

89. However, when the brandy study isolates consumer responses that may assist us in arriving at precise conclusions regarding substitutability, it appears that the actual blurring of traditional liquor categories is between spirit categories:

(Paragraph omitted, contains confidential information)

90. The assessment by another group of market researchers commissioned by the parties, Ingwe Communications, of the abovementioned research reinforces the focus on inter-spirit category substitution:

(Paragraph omitted, contains confidential information)

91. Moreover, Distell appears to have acted on these findings:

(Paragraph omitted, contains confidential information)

92. What little independent research has been presented to us appears to support the conclusions of the parties’ research insofar as it demonstrates substitution between spirit categories. Hence independent research on the behaviour of South African consumers reported in the 2001 Alcoholic Beverage Review concludes that ‘consumer behaviour in the face of economic hardship continues to move from a premium brand in one category to a cheaper category rather than a cheaper brand’. We were informed of research on product innovation that concluded that: ‘the growth of RTD’s/FAB’s has resulted in a blurring of the traditional product categories’.

93. The Commission however presents evidence that purports to question substitution between spirit categories. It argues that certain of the data presented by Nielsen, in particular those data indicating the apparent long-term stability of the distribution of market shares between spirit categories, support a category-based definition of the relevant market. The Commission effectively contends that if competition was occurring between categories, then one would expect to see movement over time in the shares of the alcoholic beverages market commanded by the various categories – the stability in these shares indicate that competition occurs within the respective categories.

94. The Commission presents a 10-year trend in the market shares (based on litres sold) in narrow spirit categories:

Table 2a Trends in spirit category market shares

	1993 (%)	2002 (%)
Brandy	44.1	40.2
Whisky	19.7	22.9
Cane	9.7	5.0
Gin	5.9	5.4
Vodka	14.2	14.5
Rum	2.5	4.6
Liqueurs	3.9	7.4

Source: AC Nielsen

95. However, the Commission’s reliance upon long-term trends in the liquor industry as evidence of low inter-category substitution has several drawbacks. First, substitution in response to a price increase does not have to be long-term in nature in order to qualify as evidence of substitutability for the purposes of a competition analysis – indeed it is generally accepted that long-term trends are rather indicative of changes in tastes, new product innovation, etc., that is, changes that cause a shift of, rather than a movement along, the demand curve. Secondly, using the long-term liquor sales data runs the considerable risk of masking underlying and, from the perspective of a competition evaluation, critical shorter-term trends.

96. In other words, it is wholly conceivable that an increase in the price of a particular liquor category, A, may have caused a significant sales decline in the short term as consumers switched to category B. However, assume that, for whatever reason, the producer of Category B could not hold its prices at the relatively low level and, a year later was obliged to increase its prices to the level of Category A thus losing sales to that category. If the producers of the respective categories then, for say the next five years, maintained this parity between the prices of the two categories, the long term trend would indicate relative stability in the overall shares of the two product categories and, in the Commission’s view, an absence of inter-category competition, while an analysis of short term trends and events may indicate a degree of substitutability between the categories that placed them in the same relevant market. In our view year-on-year changes in market shares of the various products in a declining market are important to indicate substitution is occurring in that market, even if the market shares return to historical levels every decade or so.

97. Indeed, although more rigorous econometric analysis would be necessary in order to establish a structural break in the data set, it is nevertheless reasonably clear that there are at least two conflicting trends discernable in the period under review. Hence, if one takes 1993 and 2002 as a basis for comparison it appears that brandy lost only 8.8% and gin 8.5% while vodka increased its share by 2.1%. However, a significantly different picture emerges if we distinguish between 1993-97 on the one hand and 1998-2002:

Table 2b Trends in spirit categories market shares

	1993-1997 (%)	1998-2002 (%)
Brandy	+ 6.6	- 11.3
Gin	- 18.6	+ 10.2
Vodka	- 15.5	+ 10.7

Source: Calculations based on AC Nielsen data

98. Note that the decline (increase) in the brandy share is accompanied by increases (declines) in vodka and gin. For the remaining spirit categories, the trend is indeed one of either long-term decline or growth:

Table 2c Trends in spirit categories market shares

	1993 – 2002 (%)
Cane	- 48.5
Rum	+ 84.0
Liqueurs	+ 89.7

Source: Calculations based on AC Nielsen data

99. Clearly for the market share data to provide any meaningful insight into the extent of substitutability between categories, they would have to read alongside price fluctuations over the same period – hence, as we will elaborate below, it is extremely pertinent that the break in the trend coincides with the Oude Meester price hike in 1997. The Commission has argued – and this too is examined below – that the break is caused by the behaviour of a statistical outlier, namely Kwazulu-Natal. However, eliminating this important piece from the overall data set requires detailed justification. Certainly, a simple assertion based on the long-term trends revealed by the Nielsen data reveals little about the competitive interplay between the traditional liquor categories.

100. Some of the witnesses have confirmed the simultaneous existence of stable long-term market shares and short-term fluctuations. Mr. Hooper, the witness from Snell, identified this phenomenon:

“I think if brandy were to go up in price significantly and there have been other instances where Distell might point to that having happened. In Natal, for example, fairly recently brandy went up fairly substantially in relation to Smirnoff. And Smirnoff picked up a lot of market share in Natal. You do get these shifts and you do get them in pockets, but if one looks over a medium period of time, I think you’d find that those shifts are there, but they are relatively minor. And if you take a brandy category, for example, which is as large as brandy is in the South African market that the brandy price goes up out of kilter with other spirit categories or any other alcohol category, it may suffer, but it would suffer to a relative degree. I don’t think you’re going to suddenly find brandy going from its fifty percent (50%) market share of the spirits market, for example, rocketing down to forty percent (40%) or below forty percent (40%). I don’t think you’re going to get that sort of shift occurring in a short to medium term. In the long term it is a possibility, but I don’t see it as a short or medium term reaction.”²¹

101. Other witnesses have pointed to the significance of shifts in market shares, however small, in a declining market. Forsyth, the parties’ expert witness, testified as follows:

“If the market on the other hand has been growing, then you could say well some of that additional consumption in alcohol beverages could come from other drinks that they may have, but because it’s shrinking or stable then there’s probably a high degree of substitution going on.”

102. Clearly, then the Nielsen data on market share do not materially assist in determining the extent of inter-category substitutability and do not make a significant contribution to the task of identifying the relevant market.

103. The parties, for their part, insist that the evidence of substitutability between the various traditional categories supports their case for defining an alcoholic beverages market. They rely on a range of sources in support of their claim that competition occurs across the traditional liquor categories and, accordingly, that a broad definition of the relevant market that includes all alcoholic beverages should be accepted.

104. The parties’ marketing research provides some interesting data indicative of the vulnerability of spirits to substitution, illustrating that within the spirits category certain spirits enjoy considerably less category loyalty than others and that in those same categories a significant proportion of consumers is highly ‘at risk’ to other categories.

105. We reiterate that, in our view, the responses gleaned in the brandy study provide a strong indication of substitutability within spirit categories but it does not help in identifying substitutability between spirits, beer and wine.

106. However, the parties have also attempted more traditional statistical analyses of actual and likely responses to increases in the price of key spirit brands. Firstly, they have attempted to calculate own-price elasticities and cross-price elasticities based on actual consumer responses to actual price movements. Secondly, they have undertaken a ‘brand price trade off’ study, based on data gathered through the brandy study. This latter study, in contrast with the more conventional elasticity studies, is not a study of actual behaviour but is rather in the nature of a laboratory experiment based on survey responses to hypothetical increases in the price of different spirit categories (as a whole) as well as particular spirit brands.

Elasticities

107. The parties have attempted to calculate own-price elasticities and cross-price elasticities. Own-price elasticities are generally used to indicate market power: if the own-price elasticity is low then prices can be increased as the resultant decline in volume would be small, leading to an increase in revenue; low own-price elasticities are therefore associated with market power. The parties have provided own price elasticities for Distell’s products, most of which are greater than (-)1 and therefore price-elastic.

108. However, the Commission correctly points out that the own-price elasticities calculated here are unreliable, largely because the prices of other brands in all liquor types were increased by varying amounts during the period under review thus violating the ceteris paribus condition and making it impossible to isolate the impact of the own price increase. The simple ‘textbook’ calculation of elasticities employed here is in fact only possible in theory, as the requirement to hold ‘all other variables constant’ can simply never be fulfilled in reality. Changes in income, prices of other products, product promotions and a potentially endless range of other variables can affect the demand for a certain product, particularly of ‘luxury’ items, even in the short term.

109. This is not to say that elasticity calculations are not possible, but rather, that they should take the functional form of an econometric demand model, that uses multiple regression, to estimate the demand for a certain product. Such a model would assign values to different influences and, combined with the appropriate econometric tests, correct for bias problems.

110. A calculation of own price elasticities based on mere sales data that would hold up to scientific scrutiny is therefore highly unlikely. The approach taken here, namely comparing two annual sales aggregates and two sets of retail prices, is simply too crude to result in any meaningful findings, thus explaining the confusing sizes and signs of the elasticities found.²²

111. The Commission also correctly points out that the cross-price elasticities submitted by the parties are counterintuitive as only 2 out of 8 calculations even have the correct sign for substitution - the result must be positive, indicating that a higher price of one product leads to higher demand of its substitutes. A negative is indicative of complementary products. The cross-price elasticities suffer from the same weaknesses as the own-price elasticities, namely that changes in relative prices, income and other relevant variables are not taken into account.

112. In summary, the elasticities submitted by the parties are of little or no assistance in determining the existence or extent of inter-category substitution and should be disregarded. The inadequacies of the elasticity calculations were in fact admitted by the parties.

The Brand Price Trade Off study

113. The Brand Price Trade-Off study (based on the data collected for the Brandy Study) analyses substitution between specific brands, spanning several liquor categories. Consumers are presented with different price combinations for several liquor brands and asked to indicate their purchase preference. The liquor brands tested include several brandy brands and several brands from other liquor categories including vodka, whisky and beer.

According to Forsyth, the parties’ expert witness, this methodology has been used by marketing departments for several decades to measure how consumers react to price changes. Forsyth explains why this is the preferred methodology in market survey:

‘The way it works is the following. It’s based on a presumption that if you just go to someone and say what are you willing to pay; are you willing to pay more and what have you, you tend not to get very sensible answers. What you need to do is to put something more into it, more, as much as you can in a research setting, realistic situations where they are choosing between brands at different prices, kind of like they would be in a retail outlet, trying to choose a product to buy. And then observe those different choices and they will infer from that how important price is and how important brand is in their different choice.’

114. Forsyth explains further that the BPTO study ‘…would be very important for me as I put together my marketing strategy, because in my marketing strategy what I want to do is I want to take certain brands and target them at key segments and then look at those segments to see where I’m at risk and to understand that as I put together my marketing plan and my brand plans, that I have to worry about the potential switching to somebody’s other market, to these other products.’

115. The Brandy Study found that the majority (..%) (confidential) of (confidential) drinkers consume more than (confidential) of alcoholic beverage within a ..-day (confidential) period, compared to ..% (confidential) of (confidential) drinkers and that a high proportion of (confidential) consumers are at risk to other categories. The BPTO analysed the substitution between (confidential) and other alcoholic beverages.

116. The main findings were that (confidential) consumption is price elastic as a 10% price increase (in all (confidential) brands under scrutiny) leads to a (confidential) market share loss of ..% (confidential). This (confidential) market share loss corresponds to the following gains in the market shares of the other products:

Table 3 Market share increases per product due to 10% price increase in a particular liquor category (confidential)

Confidential	Confidential	Confidential	Confidential	Confidential
Confidential %	Confidential %	Confidential %	Confidential %	Confidential %

Source: Calculations based on BPTO

117. These figures suggest inter-category substitution, at least between brandy brands, on the one hand, and beer, whisky and vodka brands, on the other.

118. In addition, when the price of (confidential) is increased by 10%, as much as ..% (confidential) of its market share is lost, which does not translate in equivalent market share gains for the other (confidential) brands from which the respondents could choose, thus clearly indicating inter-category substitution.²³

Table 4 Market share increases due to 10% price increase in a particular spirit brand (confidential)

Conf.	Conf.	Conf.	Conf.	Conf.	Conf.	Conf.	Conf.
Conf. %	Conf. %	Conf. %	Conf. %	Conf. %	Conf. %	Conf. %	Conf. %

Source: Calculations based on BPTO

119. From the perspective of a competition analysis, the main weakness of the BPTO study is the fact that the brands included are purposively selected, i.e. they are a selection of brands presumed to compete against each other, and if an important competitor is excluded, the results will suffer in accuracy.²⁴ In this case the selection of the brands and price points were done for the brandy study itself and based on consumer focus group information that indicated that competition was taking place between these brands. The data sample included a statistically significant number of respondents and was performed for the purposes of Distell’s marketing overhaul and not for the purpose of any competition investigation.

120. Note also that consumer responses gathered in this type of laboratory experiment are likely to be biased towards overstating the extent of substitution. That is, consumer responses tested in this way are likely to be stronger than their responses to a real life price hike. In real life consumers will not necessarily remember the prices they paid prior to a price increase, nor be able or willing to compare the prices of the near-substitutes this closely.

121. Other aspects of the methodology employed in this study are also open to criticism. The model does not allow for the possibility of consumers reducing alcohol consumption due to a price increase (either reducing their consumption of a particular brand or by switching out of alcoholic beverages completely) as consumers are expected either to continue buying their brand or to buy an alternative alcoholic beverage, and this stylisation of consumer behaviour will further amplify the expected response.

122. However, these shortcomings notwithstanding, the BPTO data is impressive and the findings, albeit open to qualification, informative from a competition perspective. We are, at least, satisfied that they are made on the basis of consumer research. After all is said and done, consumers indicating their preferences based on their reserve prices constitutes the basis of how demand curves are constructed. Moreover, whilst maintaining the ceteris paribus condition, the methodology mimics consumer behaviour in a SSNIP test, thereby predicting substitutability. The approach itself is interesting in that it provides a manageable alternative for researching substitutability, one less involved than econometric modelling, and which, when prudently conducted, may be less prone to bias and other data related defects.

123. On balance then the BPTO appears to be a useful, albeit far from flawless, measure of substitution. However, while this particular study does demonstrate that, in the face of a generalised increase in the price of a particular spirit type, consumers will substitute other liquor categories, these are, by and large, spirit categories. And, while it indicates that in the face of an increase in the price of a particular spirit brand, consumers will substitute other brands, again these are mostly spirit brands. This may be a consequence of the range of possible substitutes selected – but while it constitutes further evidence of substitution between spirit categories, it does not persuade us that beer and spirits are substitutable categories of alcoholic beverage.

The Kwazulu-Natal Experience

124. In most substitution analyses the research possibilities would end here, with a hotchpotch of evidence of varying reliability, which, on a balance of probability, suggests that South African consumers consider various sprit categories as substitutes, but without a real case study, without, that is, the opportunity to analyse a sufficiently significant event that would allow for a more confident conclusion on substitution. Fortunately in this case there is a real life event that can be analysed and on which we place greatest reliance and which represents a major challenge to the Commission’s already weak findings. We refer to the dramatic shifts that occurred in the liquor market of Kwazulu-Natal province in 1997/8. As the tables presented by AC Nielsen show a decline in the market share of brandy of some 8.6% was accompanied by a rise in the market share of vodka in the order of 34.0%.

125. The shifts between the two categories have been largely attributed to movements in two brands, namely Oude Meester brandy and Smirnoff vodka. When, in 1997, the price of Oude Meester increased to above R30, large decreases in market share ensued, and these volumes were lost mainly to Smirnoff vodka. Both the Commission and the parties attributed this evidence of substitution to Oude Meester breaking a psychological price barrier by pricing itself at the thirty Rand mark.

126. On the face of it this provides powerful support to those who argue that consumers will substitute between categories in response to a price increase. But why does this response appear to be confined to Kwazulu-Natal and does that fact allow us to follow the Commission and simply treat Kwazulu-Natal as a statistical outlier, an anomaly? In support of its claim to have this evidence treated as an anomaly, the Commission in fact presented data that excluded Kwazulu-Natal, thereby demonstrating that absent the data from this province the evidence of substitution is not significant.

127. The witness for the Commission, Mr. Lewis, alludes to the specific attributes of the Kwazulu-Natal liquor market, indicating that Oude Meester was consumed mainly by black consumers:

Manoim: “But you have no explanation as to why Natal should be different in that respect.”

Lewis: “Well I hope there are not too many people from Natal, but historically Natal has always done its own thing. And what’s interesting, and again I’m not positive, because as I say it’s not my field of expertise, but I’m sure one witness will be able to say something later - that the Oude Meester brand was predominantly consumed by the black market, not by the Asian, although a fair amount was obviously, but it was predominantly a black brand. So we have seen brands within that sector of the market change overnight. Now you can classically look at Lion. You can look at what’s happening to Castle at the moment. You can go through a list and when markets turn they turn quickly in that field.”

128. Lewis has in fact hit the nail on the head by indicating that the Oude Meester drinkers in Kwazulu-Natal belong predominantly to a particular consumer demographic. It appears that province has a particularly high concentration of one particular consumer segment (as identified by the Brandy Study) who switched en masse as the result of the Oude Meester price moving above R30. This segment of consumers is characterised as:

(Paragraph omitted, contains confidential information)

129. Kwazulu-Natal has a significantly greater concentration of this consumer segment than any of the other provinces. It follows that if the rest of the consumers in this category are spread equally over the other 8 provinces, then one would expect the aggregate figures minus Kwazulu-Natal to indicate, on the face of it, a considerably lower degree of substitution. In fact the responses of consumers in any of the other provinces may mirror that of their Kwazulu Natal counterparts but the extent and direction of the substitution is only clearly revealed when Kwazulu Natal, the locale of the largest concentration of this segment, is included in the aggregate data. The inclusion of Kwazulu Natal does not distort the data – on the contrary the exclusion of the provincial data would vastly understate the data by excluding the bulk of the affected consumer segment.

130. Note that by excluding the Kwazulu-Natal numbers, the brandy-vodka substitution becomes less obvious, but the pattern of rising shares in gin and vodka when brandy is on the decline remains clear, particularly in the period 1998-2002, the period following the Oude Meester price increases:

Table 5 Trends in market shares of spirit categories 1993 – 2002 (brandy, gin, vodka)

	1993 – 1997 (%)	1998 – 2002 (%)
Brandy	+ 1.1	- 9.6
Gin	- 20.6	+ 17.6
Vodka	+ 13.4	+ 8.0

Source: Calculations based on AC Nielsen data

131. For the remaining categories, the trend is again one of either long-term decline or growth:

Table 6 Trends in market shares of spirit categories 1993 – 2002 (cane, rum, liqueurs)

	1993 – 2002 (%)
Cane	- 41.4
Rum	+ 78.6
Liqueurs	+ 86.4

Source: Calculations based on AC Nielsen data

132. The Commission argues that the KZN data shows that the price increases post-1997 occurred when the pricing was on the elastic part of a generally price inelastic demand curve and that it merely demonstrates that market power was exercised ‘a bit too far’. However, Oude Meester is one of many brandies in the proprietary (medium-priced) market segment, so that the switch to an alternative liquor category (rather than another brandy brand within the broad price segment) is evidence of inter-category substitution.

133. Obviously there are some important caveats. The type of statistical ‘eyeballing’ engaged in here does not prove causality. Causality between the rise in the price in Oude Meester and the rise in Smirnoff sales can only be established scientifically when a reliable demand model of the sort described earlier is constructed. However, Oude Meester’s ‘small, but significant, non-transitory’ price increase coupled with the concentration in Kwazulu-Natal of the consumer segment at which Oude Meester was directed, persuades us to treat this experience as the single most plausible piece of evidence for inter-category substitution on the record.

134. We should, in concluding this discussion of the Commission’s and parties’ views of the relevant market, comment in greater depth on the Commission’s insistence that where inter-category substitution occurred – as in the Oude Meester/Smirnoff example – it reflects a case of monopolistic pricing taken a step too far, a case of the consequences of pricing on the price-elastic portion of a generally inelastic demand curve, a phenomenon referred to as the ‘Cellophane Fallacy’ and first explored in the much cited case of United States v. E.I. du Pont de Nemours and Co.²⁵ In that case, it was argued that cellophane was not a separate relevant market since it competed directly and closely with flexible packaging materials such as aluminium foil, wax paper and polyethylene. It was not recognised that a high own-price elasticity may mean that a firm is already exercising monopoly power and that, as the sole supplier of cellophane, Du Pont was likely to have set prices at monopolistic levels above which alternative products became substitutes.

135. In the matter before us the Commission’s contention of ‘cellophane fallacy pricing’ is based on its insistence that there is little difference in the cost of production between a ‘value for money’ brandy and a premium and proprietary brandy. On the basis of this argument it concludes that all brands in the higher ‘proprietary’ and ‘premium’ are priced at supra-competitive levels. The Commission concludes that the prices should be adjusted to ‘competitive levels’ before determining whether they are substitutable with other competitively priced products. This argument serves to underpin the Commission’s narrow relevant market definition.

136. Admittedly, the ‘cellophane fallacy effect’ should be borne in mind when using demand elasticities to determine market power or the extent of the relevant market, as any profit-maximising firm with a degree of market power will set prices at a level where demand for its product is elastic (otherwise it would raise prices further). Using elasticities that are based on elastic demand pricing are therefore misleading, as substitution at monopoly prices is much more likely than at competitive levels. As Bishop & Walker remark:

“The mere fact that at the monopoly price, a monopolised product faces demand substitutes does not mean that the firm producing the product has no market power.” It is therefore generally argued that in non-merger inquiries observed own-price elasticities may understate the degree of market power.”²⁶

137. For merger enquiries however the elasticity of demand refers to the elasticity at the prevailing price level as this assessment is concerned with future market power. The Commission is therefore not justified in suggesting that the prices should be adjusted to competitive levels before price elasticities are used.

138. The Commission’s argument that the observed decreases in, for instance, Oude Meester brandy sales are due to price increases that “pushed it towards the elastic part of Oude Meester’s normally inelastic demand curve”, is based on the production cost of Klipdrift brandy and a reference to the brandy report which indicated that:

(Paragraph omitted, contains confidential information)

139. According to the Commission, the Tribunal further “heard evidence that there is little difference in the cost of production between a ‘value for price’ brandy and a premium or proprietary brandy”. The Competition Commission submits that brands in the proprietary and premium categories are priced well above competitive levels, and that therefore their prices should be adjusted to competitive levels, before determining whether they are substitutable with other competitively priced products. In the absence of this adjustment, interpretations of pricing behaviour will fall prey to the cellophane fallacy.

140. There are two problems with this line of reasoning. Firstly, there is no evidence presented by the Commission that establishes that it is price inelasticity (reflected in a movement along the demand curve) rather than vertical differentiation (reflected in a shift of the demand curve) that allows for higher prices - the evidence from the parties merely indicates that substitution takes place. The Commission attempts to argue that any pricing above marginal cost provides evidence that demand is inelastic. In fact pricing at marginal cost only occurs when all the conditions of perfect competition are met – it is a theoretical artefact. Pricing above marginal cost is evidence of an imperfection in the market – notably, in this instance, product heterogeneity or differentiation - that enables the exercise of market power, but it fails to provide any evidence on the elasticity of demand in market analysis, and, hence, of cellophane fallacy-type pricing behaviour.

141. Secondly, the ‘admission’ by the parties that there is little difference in the production cost of the different types of brandy does not include marketing expenditure or image-driven strategic pricing. Note also that the parties have presented evidence that demonstrates that the costs of producing a premium or ‘prop’ (that is high and medium priced) brandy is higher than that of producing a value (that is, low priced) brandy. We have, in short, been presented with no evidence on marginal cost – we have simply been presented with an assertion to the effect that the differential in price between two brands of brandy establishes that the higher priced brandy is priced at a supra-competitive price.

142. Moreover, it should be borne in mind that the cellophane fallacy was identified in a situation where the supplier was the sole manufacturer of a particular product so that an exercise of market power would lead consumers to consider alternative products with different characteristics. The situation is very different in the South African liquor market. Oude Meester is one of many brandies in the proprietary (medium-priced) market segment to which consumers could have switched in response to a price increase of Oude Meester. Therefore, the switch to an alternative liquor category altogether, more plausibly indicates genuine substitution between spirit categories rather than a market power-induced switch to products that are not really substitutes. This point was noted by the parties:

Mr. Rogers: “Indeed the existence in a market of significant branding and the availability of brands across a wide range of prices is incompatible with the application of the so-called cellophane fallacy.”

143. The Commission itself acknowledges that overly narrow market definitions may penalise vertical product differentiation:

Adv. Pretorius: “In the United States monopolisation complaints are not recognised in respect of single product categories. The reason for this is that it seems to create markets which are too narrow and therefore punish the success of a company makes him dominant or monopoly, who has successfully branded its products … who has differentiated his products from other products through successful branding. The FTC has a tendency to define narrow product margins often resulting in the punishment of companies that have successfully differentiated their products from very similar products.”

144. However, the Commission nevertheless opposes a segmentation of the market that reflects the wide price differentiation within each category or group of categories because, it insists, there are no differences in the intrinsics of the products involved:

Adv. Pretorius: “We submit that these types of distinctions protect consumers who could unnecessarily because what prevents them from informing themselves better about the content of the brandy or the content of what is in the superstore. And places an undue…. places an undue premium on social status such as in these instances that people often drink Bells because they would like their friends to see that they drink Bells. Although the content remains the same. On the basis of the hypothetical monopolist to distinguish between these categories in our view is protecting a lazy consumer and a consumer preoccupied with his social status in the community. If we accept … if we accept that there is no quality difference or the quality difference is limited and in any event not to the extent that it is double in price.”

The Relevant Market: a third perspective

145. As already indicated, we do not believe that the Commission has adequately supported its argument for a number of narrow, category-based relevant markets. The extensive evidence and argument presented by the parties casts severe doubt on the Commission’s case and has contributed to our rejection of the latter’s version of the relevant market. This is certainly true where spirits are concerned – on the evidence before us we are able to conclude that there is sufficient substitutability between the various spirits categories to include them in the same relevant market, although, as elaborated below we believe that this substitutability occurs within broad price bands. As we shall indicate, the Commission’s arguments for placing the various spirit categories in separate relevant markets is plausible only with respect to the highest price band.

146. However, although we are persuaded that there is considerable substitutability between the various spirit categories, we nevertheless reject the parties’ argument for a single alcoholic beverages market, particularly insofar as it presupposes the inclusion of beer in the same market as spirits. The parties have attempted to rely on the brandy study and their calculation of the ‘Brand Price Trade Off’ to establish the substitutability of beer and spirits but, as we have already pointed out, this evidence is not wholly persuasive.

147. While the parties will no doubt insist that it is not for them to establish the substitutability of beer and spirits, nor is it acceptable to make a wholly novel and counter-intuitive claim to the Tribunal without establishing a factual basis for the claim. Spirits and beer are, to be sure, both beverages and both contain alcohol but their ‘intrinsics’ are otherwise distinct, their production processes are equally distinct, and we are aware of no single instance in which the claim that they belong to the same relevant market has been made, let alone upheld. Fish paste and beluga caviar are both commonly spread on crackers and both have some relationship to fish, but this does not make a claim to place them in one market at all plausible.

148. We note too that breweries and distilleries are still, by and large, housed within wholly separate corporate entities. Indeed when South African Breweries and the various distilleries entered into their market sharing arrangement they did so on the basis of a division between beer, on the one part, and, on the other, spirits and wine. In the recent past the brewers and the distillers have entered into competition in the very diverse FABs market. Indeed we have indicated that we believe that the evidence presented in this transaction suggests that the competitive interface between spirits, FABs, wine and beer is considerably more complex than that contended for by the Commission. However, we do not think that it supports the parties’ argument for a single relevant market comprising all of these various categories. We find, instead, that there are, implicated in this transaction:

• Three relevant spirits market - a ‘premium’ spirits market, a ‘proprietary’ or ‘prop’ spirits market and a ‘value’ spirits market. The basis for this three-way division is the vast price differential between the various spirits brands. This is further elaborated immediately below.

• We find secondly that there is a wine market. Indeed, consistent with the basis on which we have divided the spirits markets, there are undoubtedly several wine markets whose boundaries reflect the vast price differential between premium wine, at one end of the spectrum, and low value, relatively inexpensive wine, at the other end of the spectrum. However, we have been provided with insufficient evidence to identify with any confidence the location of the boundaries between brand or label price bands.

• We are not convinced that there is sufficient evidence to arrive at a definitive conclusion regarding FABs. This is clearly a very broad ranging category with certain brands and sub-categories, for example the various ciders, possibly in the same market as beer, while others are more likely to compete with spirits, while yet others may well be best situated in one or other of the wine markets. While it is precisely the varied nature of the products collectively referred to as FABs, that allows them to act as a bridge between spirits, wine and beer, it is also this quality that resists attempts to pigeonhole FABs in a single relevant market.

149. Spirits are, in our view, the potential area for concern in this transaction. As already elaborated, we reject the Commission’s argument in favour of placing the traditional spirit categories in separate relevant markets. We believe that there is sufficient evidence of inter-category substitution to support a finding that places different categories of spirits in the same market. However, we do not believe that there is a single spirits market. Instead, we find three relevant spirits markets.

150. However, in contrast with the Commission’s view, our spirit markets are not segmented by the category of spirits, by, as it were, the contents of the bottle, but rather by the vast price differential within each of the various categories. Hence, dare we say it, just as ‘common sense’ would reject an argument for placing a multi-million Rand Rolls Royce in the same market as the humble Opel Corsa, so too does it rebel at the notion that Remy Martin competes with Wellington (or Cape to Rio) or Chivas Regal with First Watch (or Russian Bear). However, as we shall elaborate below, we do think that the parties’ have, through their evidence and argument on substitutability, made out a sufficient case for a relevant market that includes, for example, the likes of a brandy like Wellington, a whisky like First Watch, a cane spirit like Cape to Rio and a vodka like Russian Bear because they belong in the same broad price brand, just as we accept a second relevant market that includes, for example, Smirnoff vodka, Gordon’s gin, Klipdrift brandy and Bells whisky.

151. In other words, we are persuaded that there is substitution between traditional spirit categories but we are equally persuaded that this substitution will take place between spirits in the same broad price band. We do not claim to be able to identify precisely the boundaries of each of those bands but, just as in previous mergers in other consumer goods industries, we were satisfied to accept the LSM categories in common usage in those sectors, so here are we satisfied to accept the broad categories delineated as value, proprietary and premium that are commonly used in the trade. Hence faced by an increase in the price of Wellington a consumer may well seek the comfort of another value brandy brand, but so too may the consumer in question turn to First Watch whisky. However, the affected consumer is unlikely to substitute a proprietary or premium brand of any category. By the same token an increase in the price of Smirnoff vodka may well force a hitherto loyal customer into the hands of Gordon’s gin or Bells whisky but it is unlikely to have him reaching down into the value brands or ascending into the premium sector.

152. Let us be clear that these are spirits markets. That is, we do not believe that inter-category substitution is unlimited. That is, while we are persuaded that there is sufficient evidence of substitution between the spirit categories in the same broad price band to justify several price-segmented spirits markets, we do not believe that there is sufficient evidence to include FABs, wine or beer in any of these markets.

153. An immediate qualification is required here: we are persuaded that there is sufficient evidence that changing tastes, new products and rapidly evolving demographics are injecting new dynamic features into the broad alcoholic beverages market. For example, we are persuaded that FABs not only constitute an important new category of alcoholic beverage, one that has clearly made inroads at the expense of other liquor categories, but that they also constitute an important bridge between consumers of spirits and beer, spirits and wine, and wine and beer. Or, by way of a second example, we believe it reasonable to hypothesise that inexpensive wine is part of the same relevant market as the lowest price spirit band where, several witnesses have pointed, there is particularly strong evidence of inter-category substitutability. But we do not have sufficient evidence to broaden these relevant markets in this way – hence we will restrict our relevant spirits market to the traditional spirit categories although we will factor the interplay between FABs and spirits into our assessment of competition in the middle segment spirits market – the ‘proprietary market’ – and we will consider the influence of inexpensive wine in our assessment of the bottom segment – the ‘value segment’ – of the spirits markets.

154. Our argument for a relevant market that is based on inter-category substitution and income or living standard based segmentation draws on more than common sense.

155. We have already dealt with the evidence on substitution between the traditional spirit categories at some length and have concluded that there is sufficient evidence to factor this phenomenon into the definition of the relevant market.

156. As for the market segmentation based on price differentials, consider how deeply embedded in the thinking of industry participants is the three-way distinction between ‘value’ brands, ‘proprietary’ brands and ‘premium’ brands. Most of the witnesses who participated at the hearing articulated their marketing strategies and that of their competitors in terms of this approach.

157. While these consumer segments are, by no means, absolutely coterminous with income level (nor, we should note, is the LSM concept conventionally used in market research), there is clearly a close relationship between branding strategy and price simply because income and the related ability to participate in the market is a critical, the most critical, element in the construction of the consumer segment at which a particular brand is targeted. Certainly, a fundamental marketing strategy is to drive consumers up the brand ladder, to persuade them to desert the lower reaches of the market in favour of higher value brands. But there are limits imposed on this strategy – as the saying goes ‘the poor will always be with us’, the rich will always be willing to pay a premium for consuming the ‘extrinsic’ qualities that identify them as a group distinguishable from the rest of society, and entrepreneurs will constantly strive to segment their markets.

158. The occasions on which the distinction between the value, proprietary and premium categories informed the various analyses presented to us by all the witnesses are too numerous to list. Seagrams and DGB presented us with a detailed map locating the various spirit brands in the three segments of value, proprietary and premium. Snell’s competitive advantage (and, we are told, its Achilles heel) manifests the existence and significance of this market segmentation most starkly. Snell is the dominant producer and distributor of ‘value’ brands, those brands serving the bottom end of the market. It has developed highly successful brandy, whisky and white spirits brands. But it is struggling to transform one or more of its brands from the low margin value segment to the higher margin proprietary segment. In recent years it has been anchored to the value segment by Distell’s efforts to penetrate that segment.

159. Clearly, while the low value segment of the market does not, on its own, particularly attract Distell, by entering Snell’s turf, Distell increases the already considerable risks that attach to Snell’s attempts to re-brand an established successful low value brand – Snell is simultaneously taking on the likes of Distell and GUDV in the proprietary sector, while, through migrating a successful value brand, it eases the ability of these formidable competitors to penetrate the value segment. Indeed, this predicament, which we re-examine below, appears to be the basis of Snell’s objection to the transaction.

160. What emerges then is a plausible third view of the relevant market, one distinct from the narrow category-based market definition contended for by the Commission, but also clearly distinguishable from the overly broad definition advanced by the parties. It is a view of the relevant market that takes account of the particular features of liquor consumption in South Africa which, for present purposes, is most starkly manifest in a blurring of the distinction between various spirit types.

161. Moreover, our view of the relevant market recognizes the central significance of branding in the marketing of alcoholic beverages. Inter alia, the evidence of substitution between Oude Meester brandy and Smirnoff vodka, indicates that increasingly spirit manufacturers compete on image and branding, on the ‘extrinsics’ of the product, rather than on spirit category, the ‘intrinsics’. This means that, when faced with a significant price increase, a consumer that prefers a branded brandy is likely to switch to another branded spirit, possibly a white spirit or even a FAB, of equal ‘extrinsic’ appeal rather than to a less expensive, but less ‘extrinsically’ satisfying, product in the same traditional spirit category.

162. Contrary to the conclusions reached in the other jurisdictions that have examined liquor mergers, we view the rise in the importance of branding – something acknowledged by all participants in these hearings - as contributing to a decrease in the significance of the differentiation between traditional categories. Furthermore, the successful penetration of FABs, has introduced a dynamic element into the market that has contributed to this blurring of categories.

163. We find then that there are three relevant spirits markets implicated in this transaction each bounded by the broad price categories of the products that populate these markets. We will use the language of the trade to delineate these markets and so dub them the value, proprietary and premium markets.

164. We are aware that where the blurring of product categories is concerned our conclusions are likely to apply more strongly to the value and proprietary markets than to the premium market. Most market participants acknowledge occasion-based drinking and category loyalty in respect of the premium market. At the opposite end of the spectrum, the most extreme example of category blurring is apparent in the value market segment. Clearly this segment attracts the most price-conscious consumers who would more readily substitute between different spirit types and, further research may well reveal, may even substitute between spirits and, because of this country’s unusually large market for inexpensive wine, wine.

165. This caveat is supported by the evidence of Mr. Tim Hutchinson, the witness from DGB, who, while insisting on the salience of traditional categories in the premium segment, conceded that switching between categories is more pronounced in the value segment:

Mr. Hutchinson: “Well I think if you did a lot of research on a good loyal brandy consumer, he’s a brandy drinker. You know, you are a gin drinker. There can be arguments in the market like Natal that the consumer there in the emerging market swaps over between gin and vodka. There could be an argument that says that as whisky gets more expensive at the bottom-end of the market, you might have people who, on just pure economic grounds, are gravitating towards brandy, just from a retail selling price point.²⁷”

Is there likely to be a substantial lessening of competition?

166. The Commission has determined that wine and FABs produced by the merging parties belong to separate relevant markets. Distell has a significant share of each of these markets.

167. Despite Distell’s significant share of the FABs market, the Commission has not found a substantial lessening of competition in that market. It accepts the parties’ argument that FABs constitute a particularly dynamic market characterized by fickle consumer behaviour and rapid new entry and innovation. The Commission avers that these features of the FABs market will undermine any attempt on the part of the merged entity to exercise market power. Bulmer, the multinational FABs producer, has offered evidence purporting to demonstrate that FABs’ consumers are willing to absorb significant price increases before switching to alternative beverages.

168. We note also that the FABs category is extremely diverse – certain FABs are effectively mixed spirits, others are cider products, while a third category is akin to fortified wine. It is by no means clear that these are part of a single market. Further evidence may well show that some should be placed in one of the spirits markets, that cider should be placed in the beer market and that the remainder should be in the wine market. Moreover, the evidence regarding substitutability between FABs and other alcoholic beverages is conflicting. The Lexecon study performed for Bulmer suggests that FABs consumers are very brand-loyal, but there is little to suggest that they are equally category-loyal, and that, when faced with a price increase in a particular FAB brand, consumers will switch predominantly to other FABs. Moreover, the parties marketing research, as well as several witnesses and other industry sources, suggest that substitution between FABs and other liquor categories is likely.

169. For all these reasons we accept the Commission’s finding that the transaction will not lead to a substantial lessening of competition in the FABs market.

170. The Commission has divided the wine market into separate categories, notably table wine, fortified wine and sparkling wine. We note that, as with the various spirits categories, there is a significant divergence in the prices of different wine brands or labels and this will certainly dictate a further segmentation of the wine market into a number of markets defined by a price band. However, as already noted, we have been provided with insufficient overall data on the wine market to identify the boundaries of these segments.

171. Despite high post-merger shares in the table wine market and the possible anti-competitive effects of the merger, the Commission has found that the merger ‘could be justified by the ability of the parties to become internationally more competitive’. While we are sceptical of the Commission’s conclusions on efficiencies – there appears to be no reason why inter-firm cooperation on international wine marketing requires a merger - we have not been provided with sufficient evidence to make a finding on the competitive consequences of the merger in this market. We note however persuasive evidence on the low entry barriers in this market and the existence of competing cooperatives and numerous wine estates and will accordingly not investigate this matter further.

172. In the fortified wine market, the Commission argues that since there is ‘no growth taking place in this market’, it is highly unlikely that the parties would be able to obtain market power through this transaction. Again we have not been provided with sufficient evidence to sustain a clear finding in this market – we are, indeed, yet to be persuaded that fortified wine constitutes a separate relevant market. We note also the relatively insignificant size of this market compared to the total of alcoholic beverages and will accordingly not investigate this matter further.

173. The Commission has identified a separate relevant market for sparkling wine, arguing that its consumption is reserved for particular occasions for which other wine does not substitute. The parties, on the other hand, have argued that the link between the consumption of sparkling wines, on the one hand, and, on the other hand, particular occasions is breaking down. We note too that the merger combines SFW’s 31,5% pre-merger share with Distillers 43,7% share, of which the lion’s share (40,6% of the total) is accounted for by a single brand, JC le Roux, with the rest of the market populated by a myriad of very small brands. Short of an outright prohibition, this makes it very difficult to construct an effective remedy. However, again we note the inadequate evidentiary basis for definitively determining the existence of a separate relevant market in sparkling wine. We note also that sparkling wine represents a minimal part of the transaction under examination and that entry barriers are relatively low as effectively any current wine producer is a potential sparkling wine producer and entry barriers into wine production per se are low. Accordingly, we will not examine this issue further.

174. Our evaluation of the transaction’s impact on competition is thus restricted to the three spirits markets. There are the ‘value spirits market’, the ‘proprietary (or ‘prop’) spirits market’ and the ‘premium spirits market’.

175. We have calculated market shares to the best of our ability although we have been hampered by significant data imperfections.

Table 7 Merging parties market shares in different price segments

Shares per category	SFW	Distillers	Cumulative
Value	21.4%	8.5%	29.9%
Proprietary	7.7%	37.6%	45.3%
Premium	16.4%	26.1%	42.5%

Source: Calculations based on AC Nielsen data

176. In summary, the above table shows that the merged entity commands a share of approximately 29,9% of the value market, 45,3% of the proprietary market and 42,5% of the premium market. However, the impact of the merger differs between the three relevant markets. Hence the table above illustrates that the pre-merger SFW enjoyed a significantly greater presence in the value segment (21,4%) than did its erstwhile rival, Distillers (8,5%). This is also marked in the ‘prop’ segment where Distillers (37,6%) – largely because of its commanding presence in brandy – overshadowed SFW (7,7%). In the premium segment, on the other hand, the pre-merger entities were, with SFW at 16,4% and Distillers at 26,1%, of similar size and so the accretion of market share is greater than in respect of the other two segments, although here too Distell owes the lion’s share of its position to Distillers’ contribution.

177. One possible conclusion that might be drawn from these data is that because the pre-merger value and proprietary markets were dominated by SFW and Distillers respectively, competition has not been substantially lessened by the merger. We are, however, inclined to treat this argument cautiously – although the accretion, and hence the ‘lessening of competition’, may appear modest when a firm with very large market share merges with a firm enjoying a smaller share, this may, by the same token, reflect the removal of one of the few remaining sources of competition.

178. This is of particular concern in a market structured along the lines of the spirits markets that we are currently scrutinising. There is not, after all, vibrant competition beyond the walls of Distell. In the value segment there is Snell with a very large market share. In the proprietary segment, GUDV looms very large indeed, Out of the total spirits market (i.e. value, prop, and premium combined) GUDV is estimated to have a market share of 23 - 29%.²⁸ Most of GUDV’s products are in the prop and premium markets, including powerful whisky brands (including J&B, Johnny Walker and Bell’s) and vodka brands (Smirnoff) as well as smaller stakes in brandy, gin, and FABs. This is graphically illustrated in Annexure ‘A’.

179. In the context of this highly concentrated market structure even a relatively small pre-merger market share may represent one of the last hopes for robust competition.

180. The following two tables show the composition of the spirit portfolio of each of the parties prior to the merger. The first of these tables illustrates the particularly powerful position of the pre-merger Distillers in proprietary and premium brandy brands. This contrasts with SFW’s position as revealed in the succeeding table. SFW clearly did not, in any of the other spirit categories, enjoy an equivalent position to that held by its merger partner in brandy. It is, instead, more broadly spread across the various liquor categories with noteworthy positions in value white spirits and whisky as well as premium brandy.

Table 8 Distillers’ market share per liquor category

Distillers shares per category	Brandy	White spirits	Whisky	Rum*	Total spirits
Value	16.2%	5.5%	3.1%	0.0%	8.5%
Proprietary	77.4%	27.9%	2.4%	0.0%	37.6%
Premium	46.1%	0.6%	3.5%	24.8%	26.1%

Source: Calculations based on AC Nielsen data

* Based on assumptions

Table 9 SFW’s market share per liquor category

SFW shares per category	Brandy	White spirits	Whisky	Rum*	Total Spirits
Value	20.7%	28.4%	28.3%	0.0%	21.4%
Proprietary	11.0%	14.6%	0.0%	0.0%	7.7%
Premium	24.2%	0.0%	18.8%	0.0%	16.4%

Source: Calculations based on AC Nielsen data

* Based on assumptions

181. The following table shows the market shares in various spirit type / price level categories of the merged entity.

Table 10 Distell’s market share per liquor category

Distell shares per category	Brandy	White spirits	Whisky	Rum*	Total spirits
Value	36.9%	33.9%	31.4%	0.0%	29.9%
Proprietary	88.4%	42.6%	2.4%	0.0%	45.3%
Premium	70.2%	0.6%	22.3%	24.8%	42.5%

Source: Calculations based on AC Nielsen data

* Based on assumptions

182. In summary then, in the proprietary and premium spirit markets Distell owes its position largely to Distillers’ dominance of the brandy category. In the value segment Distell’s strength derives largely from SFW’s contribution. The upshot is that while Distell’s market shares in each of the relevant markets under scrutiny is, on the face of it, cause for concern, this is somewhat ameliorated by an examination of the relatively small additions to market share achieved in consequence of the merger.

183. This inference – an inference that, we reiterate, is to be drawn with considerable caution - is reinforced by the fact that both of the parties to this transaction had identical shareholders. While the transaction clearly resulted in a change of control and hence required notification and evaluation, it remains nevertheless that their identical ownership structures undoubtedly compromised the intensity of pre-merger competition between the parties.

184. We are required, then, to examine additional features of the post-merger markets, before concluding definitively whether or not competition has in fact been substantially lessened. We will show that this further examination reveals that whereas in the value and premium markets there is not a substantial lessening of competition, in the proprietary market, on the other hand, competition is compromised by the transaction. We will examine each of these in turn.

The value market

185. In the value segment Distell enjoys a market share of slightly under 30%. The bulk of this share – of the order of 21,4% - was contributed to the merged entity by SFW. The accretion is small but not inconsequential – the merger is clearly removing a competitor of some significance from the value segment.

186. However, there are other factors that ameliorate the impact of the merger on competition.

187. Firstly, Distell is not the pre-eminent presence in this segment. This position is clearly occupied by Snell which owns, inter alia, the most successful brand in value whisky (First Watch), value brandy (Wellington VO), value vodka (Russian Bear), value cane (Cape to Rio) and value Gin (Strettons Deluxe Gin). Snell has, with considerable success, focused on this low value segment. Snell has achieved its powerful position in this market remarkably quickly and, although price rather than branding has been the key determinant of market share in this segment, it appears that it has managed to establish significant brandy and whisky brands which has enabled it to charge a small premium over its competitors’ brands in the value market.

188. However, margins in the value market are extremely tight. Indeed, Snell’s success partly resided in the reluctance of established distillers like Distell and GUDV to participate in this segment. Having established a powerful beachhead in the low value market, Snell is now attempting to migrate some of its successful value brands into the proprietary market where margins are significantly greater.

189. Distell now appears determined to penetrate the value market. It is not clear precisely why Distell has decided to enter this market, although clearly Snell’s success and, particularly, the prospect that this may ease its ability to penetrate the proprietary sector, must have influenced Distell.

190. Distell’s entry into the value market clearly represents a significant threat to Snell. Not only does it represent a direct challenge to Snell on its own terrain, but it increases the risk to Snell of migrating one of its successful value brands into the proprietary market. Snell avers that Distell requires the merger in order to successfully prevent entry to the proprietary market. Snell argues that the value market with its low margins represents a major challenge to a large company supporting a relatively huge overhead structure. Snell argues that the merger will enhance Distell’s market power in the high value proprietary and premium market segments and that the additional returns gleaned from the transaction will subsidise its penetration of the value segment. Snell effectively suggests that the merger will enable Distell to engage in predatory behaviour in the value segment cushioned by the returns on its dominant position in the proprietary and premium segments. Snell argues that aided by its predatory capacity, Distell will ultimately triumph in the value market, while simultaneously preventing Snell from moving its established value brands into the proprietary segment. Upon successfully driving its competitors, notably Snell, out of the value segment, Distell will desert that low margin segment in favour of a committed presence in the prop and premium markets having further limited the prospect of new entry in those segments.

191. However, while we accept that aspects of this scenario are plausible, we do not believe that this portends a lessening of competition in the value segment. On the contrary, it will raise the competitive temperature in that segment. Snell has already demonstrated an ability to raise somewhat the prices of its key brands in the value segment and this will be constrained by Distell’s entry into that market. We do not accept the argument that this will ultimately enable Distell to exercise market power in the value segment. Snell remains a robust competitor in that segment and even if Distell enjoys considerable success in the value market, low entry barriers, clearly established by Snell’s proven ability to establish itself there, will constrain any exercise of market power by Distell, just as it already constrains Snell.

192. We should underline our view that entry barriers in this segment are low principally because, in contrast with the other two market segments, penetration of the value market does not presuppose the considerable expense and lead times required in brand building.

193. We should also note that even though we have ultimately elected to exclude wine from our relevant market definition in the value segment, we are persuaded that the unusually significant production of low value wine in South Africa, will nevertheless constrain pricing practices in the value segment, a segment where price, above all, determines consumer behaviour.

194. We accordingly find that the merger will not result in a substantial lessening of competition in the value spirits market.

The proprietary market

195. However, as already indicated, we are less sanguine regarding the outlook for competition in the proprietary segment.

196. Distell’s post-merger market share in this sector is considerable. We accept that Distillers enjoyed a considerable pre-merger market share of this segment. However, the merger with SFW has added important brands to the Distillers’ armoury, notably the brandy brands, Martell 5 Star and Mellow Wood 5, and Mainstay cane and vodka.

197. Crucially, an analysis of key competitive drivers in this segment, does not, in contrast with the value segment, ameliorate our concerns. We refer in particular to our conclusions regarding barriers to entry and the dynamic features of the market.

198. Barriers to entry in the proprietary spirits market are significant. In particular, successful penetration of this sector clearly requires considerable investment in brand building. Snell’s concerns and the difficulties it has experienced in entering this market are clear evidence of this. Distell’s concern, to keep all of its brands out the hands of its competitors – as evidenced by its resolute defence of its license to distribute the Martell brand – is a further indication of the importance of brand recognition for entry into this segment. Clearly, the owners of the agency brands fear that Distell would use its distribution licence to undermine the brand before it reverts to its owner.

199. Note also that in this segment although successful branding is, in contrast with the value segment, important and the higher margins available in this segment are testament to this, the margins appear to be significantly lower than in the premium segment. The upshot of this is that for successful penetration of the proprietary segment not only is investment in branding important but high volumes are crucial as well. It appears that a premium brand earns such high margins that it is wholly possible to sustain market entry on the basis of very low sales volumes. Where the proprietary brands are concerned it appears that not only must the brands be solid, but they must also achieve considerable sales volumes. This is graphically depicted in Annexure A. This twin requirement makes for very high barriers to entry indeed.

200. We have also considered the dynamic features of this market. In this respect, FABs are the most significant element in our consideration. Both the pricing level and the ‘extrinsics’ associated with FABs, ensure that the ‘demographics’ of its consumers overlap with those of the consumers of proprietary spirits brands. While we have declined, after substantial consideration, to include FABs in the relevant market, we nevertheless do believe that they are capable of acting as a considerable influence on the behaviour of participants in the proprietary spirits market. Many of the strongest FABs are, after all, simply ‘ready-to-drink’ spirits, mixed spirits. Others – for example, the cider FABs - are likely to act as something of a bridge between spirits and beer, while, still others are clearly designed to appeal to wine consumers. FABs are thus significant not merely because certain of the products in this market may already be in direct competition with spirits but because others may ultimately come to build bridges between spirits and wine, and, even, spirits and beer. Part of the extraordinary success of the FABs category is undoubtedly attributable to its ability to draw new consumers – women, it is suggested - into the alcoholic beverages market. But it is equally clear that FABs draws a significant measure of its support from the ranks of spirits, beer and wine consumers.

201. Just as we have determined that inexpensive wine may influence competitive behaviour in the value segment, so, are we satisfied, that FABs similarly influence the behaviour of participants in the proprietary segment. However, far from ameliorating our concerns regarding the impact on competition, this factor serves to exacerbate our concerns. Distell enjoys a very powerful position in the FABs market - a market share of 70.8% in 2000 and 61.8% in 2001. Had we elected to include FABs in the relevant market, Distell’s market share would have been considerably larger, rising, we estimate, to 51,8% of the prop spirits and FABs market.

202. We accordingly find that the merger will substantially lessen competition in the proprietary spirits market.

The premium market

203. Distell’s post-merger market share in the premium spirits market is substantial and the accretion is notably larger than in any other segment. Moreover, the relatively affluent consumers in this category are more likely to conform to the brand-, category- and occasion-loyal customer stereotype identified by the Commission than is any other group of consumers. However, by the same token they are particularly sophisticated and well-informed consumers, characteristics that enable them to resist more easily attempts to exercise market power.

204. We note also that certain of the key Distell brands in the premium market are agency brands. We refer to Martell VO brandy and Scottish Leader whisky. Martell VO is the most successful brand in the premium brandy stable, commanding a 24.2% share on its own, compared to Oude Meester with 23.2%. Scottish Leader is SFW’s only premium Scotch (with a premium whisky market share of 18.8%), which is added to Distillers modest sales of Glenfiddich (3.5% of premium whisky sales), itself an agency brand. Distell’s licenses to distribute Martell VO and Scottish Leader are both subject to sales targets and periodic renegotiation. This not only renders Distell’s competitive position vulnerable to the multinational owners of those brands should they choose to enter the South African market directly – and the lengthy court battles over both the key brandy and gin brands that Distell distributes is indicative of precisely this intention – but it also means that, should we find a substantial lessening of competition in this market, there is no readily available remedy. Certainly, it would not be possible for us to order Distell to dispose of a brand that it does not own.

205. Brands are all-powerful in the premium market and they are owned by powerful multinationals. Brands are, as already noted, also important in the proprietary market but here South African brands – notably, although not exclusively, brand brands – are a significant presence. However, in the premium market there is, to our knowledge, not a single South African brand of note. Even a company with the resources commanded by Distell would have difficulty developing a new premium brand of whisky or brandy or any of the white spirits. Accordingly Distell is only likely to advance in this market through the medium of additional agency agreements. This, as we have already indicated, is only feasible if the large multinational brand owners do not wish to exploit the South African market directly.

206. Moreover, even if the multinational brand owners prefer an agency arrangement with a South African distributor, Distell is, by no means, the only effective distribution agent available in this country and there is no reason to expect that a large multinational would favour Distell, potentially a competitor in the international market, over other efficient South African distributors such as Snell or DGB. The relationship that has developed between Snell and Brown and Forman, the US based owner of, inter alia, Jack Daniels, the powerful premium whiskey brand, is evidence of this. The existence of these effective local distribution channels coupled with the importance of well-established brands means that entry into the premium market is relatively easy.

207. We accordingly find that the transaction is not likely to cause a substantial lessening of competition in the premium spirits market.

208. In summary we have determined that the transaction will result in a substantial lessening of competition in the proprietary spirits market only.

Countervailing Efficiency Gains

209. We are, in the event of an adverse finding on competition grounds, obliged to examine whether there are any efficiency gains arising from the merger that would outweigh the negative impact on competition. The parties have presented us with a massive quantity of evidence regarding efficiencies. The Commission has disputed much of this evidence – in particular it has argued that many of the efficiencies claimed could have been achieved without merging the two entities.

210. However, we do not believe that a finding on efficiencies is necessary in this instance. In our view, the efficiency defence is intended to resurrect an anti-competitive merger that, absent the claimed efficiency gains, may fall to be prohibited. This is clearly not contemplated in this instance. We have found a substantial lessening of competition in one of the many markets implicated in this transaction. Any potential remedy imposed, whether behavioural or structural, will be confined to that market compromised by this transaction. It is well nigh impossible to isolate the efficiency gains that refer to this market specifically, although, suffice to say that we are firmly of the view that none of the efficiencies claimed will be sacrificed by this finding or by any of the remedies that flow from it. Accordingly, this will not be examined further.

Public Interest

Introduction

211. In this case both the merging parties and the unions have relied on the public interest grounds set out in the Act in section 12 A (3). That section states:

When determining whether a merger can or cannot be justified on public interest grounds, the Competition Commission or the Competition Tribunal must consider the effect that the merger will have on –

1. a particular industrial sector or region;

2. employment;

3. the ability of small businesses, or firms controlled or owned by historically disadvantaged persons, to become competitive; and

4. the ability of national industries to compete in international markets.

212. The parties rely on subparagraphs (a) and (d) of subsection (3) - they state in broad terms that the merger creates an internationally competitive firm and, ironically, relying on what some of their competitors have stated during the hearing, and a submission from the DTI, that the merged firm will be good for the industry as a whole.²⁹ This they assert justifies the merger on substantial public interest grounds.

213. The unions rely on subparagraph (b) of subsection (3) and assert that job losses occasioned by the merger are so great that they will have an adverse effect on employment and hence this justifies its prohibition on public interest grounds.

214. Thus the public interest asserted pulls us in opposing directions. However, the legislation expressly allows for this. Just as the legislation may allow the public interest to resurrect a merger that will harm competition it also contemplates a situation where a public interest ground may justify the prohibition of a merger even if a merger does not have an anti-competitive effect.³⁰

215. All of this presupposes that we have determined what the public interest is and that it is not, to borrow a phrase from Robert Bork, a “policy at war with itself”.

216. In this case, multiple public interest grounds are implicated which suggest contradictory outcomes.

217. How do we approach this contradiction? In the first place each public interest ground asserted, should be viewed in isolation to see if it is substantial. Assuming that the answer to that is in the affirmative and that more than one ground exists which contradicts the other/s we need to see, first, if they can be reconciled and, if not, we must then balance them and come to a net conclusion.

218. The language of subsection 12A(1)(b) states:

“..otherwise, determine whether the merger can or cannot be justified on substantial public interest grounds by assessing the factors set out in subsection (3).”

219. This language is consistent with both a balancing approach and a reconciliation approach and we therefore need not choose between either.

220. In most cases, however, we suspect that balancing will be unnecessary as is illustrated by the facts of this case.

221. This is because a contradiction can be of two types. One is when the opposite interests collide head on, like two medieval knights in a jousting contest where only one can emerge as the victor and continue his course. This is where we are faced with a stark choice of whether to prohibit or to approve.

222. The other situation is where the opposite interests avoid one another like two vehicles bypassing each other in opposite directions on a dual lane road. They pass one another without affecting their respective opposite courses.

223. In this case, unless we find that the employment grounds justify prohibition, the unions’ public interest and those of the parties do not collide - hence we have no jousting contest and the opposing interests can travel past one another on the road in their separate lanes.

224. We will examine the employment claims first.

225. Since this transaction was first made known to them, the unions, FAWU and NUFBWSAWU, have been concerned about the numbers of jobs that would be lost. The unions were, for this reason, active participants in the litigation that took place over notification that we referred to earlier in the background section. The unions also intervened in these present proceedings and, with the consent of the merging parties, were present throughout the hearing, put questions to certain witness and made legal submissions.

226. The parties’ evidence is that the actual employment effects are minimal and offer the following statistics.

227. Prior to the merger, as at 30 June 2000, the total number of employees at both firms was 5828.Once final production decisions had been implemented, which would determine the extent of job losses, 1 882 jobs were lost. Of these only 1414 represented actual terminations, the balance representing either jobs that were frozen or not filled once they became vacant or whose occupants were re-assigned to other positions in the Distell group. Thus according to Distell the effective job loss figure is 1 414 and that included in this figure are 630 people who accepted voluntary retirement and 621 who accepted voluntary retrenchment. This left a remainder of 164 people who were forcibly retrenched.

228. Distell goes on further to state that only 16 of the 164 employees forcibly retrenched, challenged this action and declared a dispute. Some cases have been settled, but to date those that have proceeded for determination have all been resolved in favour of Distell.

229. Distell also details a lengthy process of negotiation and consultation with unions over the retrenchments. It states that the retrenchment package, which included payment of four weeks pay for every year of service, exceeded the minimum requirement of the law (one week) and the previous policy of both companies (two weeks). The unions insisted on six weeks.

230. Given this factual background, Distell argues that once one has properly analysed the effects of the terminations one should only take into account the 164 forced job losses, and that this number in the context of the overall merger, does not constitute a substantial public interest ground for prohibiting the merger.

231. The unions did not dispute Distell’s figures or its claim to have followed the procedures outlined above, but they are less sanguine about the prospects of those who took voluntary packages and whom Distell has accordingly removed from the ranks of those adversely affected by the merger. Many employees, argue the unions, took the voluntary package because they believed they had no other choice or needed the money in the short term. Accordingly, they should still be considered as having been adversely affected by the transaction. With the inclusion of this class of employee the numbers of employees adversely affected by the merger is significantly increased and the impact on the public interest concomitantly larger.³¹

232. We have in previous decisions indicated that we do not exercise our public interest determinations in a void.³² Parliament has in many instances enacted legislation that deals quite specifically with the issues referred to in section 12 A (3) and employment is no exception. Indeed we observed in the Shell/Tepco case that in many respects our jurisdiction in these areas is secondary, as these other statutes and the institutions that they create, are better placed and resourced to deal directly and effectively with these issues than are we, given that our discretion is described in section 12 (A) (3) at a high level of abstraction and generality.³³

233. In this case the unions are asking us to consider prohibiting the merger on public interest grounds and, alternatively, to consider a more generous package to those employees who are to be retrenched. They further argue that if we impose a divestment remedy that it be made a condition of such divestiture that the affected employees are transferred with the assets.

234. Let us take the alternative request first. If we are to second guess the fairness of the retrenchments and the appropriateness of the package offered we would be doing precisely what we have previously said we should not – acting as a second- hand regulator, post facto, when the appropriate body, on the evidence before us, has already made determinations and was in a better position to do so.

235. The first request, that is, that the merger be prohibited because it has an adverse effect on employment amounting to a substantial public interest ground is more difficult to evaluate.

236. Beyond requiring that public interest grounds be “substantial” before they qualify for assessment, the legislation offers no criteria as a yardstick for their evaluation, unlike with the competition evaluation, where criteria are enumerated in section 12A(2).

237. Our previous approach, although not expressly articulated in this way, has been to suggest that, where there are other appropriate legislative instruments to redress the public interest, we must be cognisant of them in determining what is left for us to do before we can consider whether the residual public interest, that is that part of the public interest not susceptible to or better able to be dealt with under another law, is substantial.

238. Thus in the case where the public interest asserted is employment, if it could be demonstrated that the merger specific employment effects are so adverse and that no other law or regulator can remedy them, then we would be obliged to intervene to either prohibit or set conditions on an approval.

239. In this case we are advised that, at worst, the merger specific employment losses are 1 414 jobs, if we accept the unions argument that all voluntary retirements and retrenchments should be counted. On the other hand if Distell is right and only forced retrenchments are counted the employment effect may be less than 164.

240. How many jobs must be lost before one has grounds for a substantial public interest? The legislature wisely does not seek to answer that for us, nor can we assume that it should be a uniform figure for all mergers - it would depend on the context. In the context of this merger Distell maintains that the total jobs lost - 164 on its approach - is less than 3 % of the aggregate pre-merger work force.³⁴

241. If, however we take the total number of jobs lost, forced and voluntary, then that figure becomes 24% - certainly significant.

242. However, we know that even if this figure is significant the test is not the number of jobs lost but the substantial effect on employment. That means that the effect can be ameliorated by a retrenchment package and not simply job retention. The package offered by Distell seems reasonable measured against the requirements of the law and its own past practice. The fact that only 16 employees disputed the final package, out of the 1414 terminated, suggests that by and large employees felt the same way. In our view, although the loss of jobs is significant, the overwhelming acceptance of voluntary retirement and retrenchment packages has lessened the adverse effects and we cannot conclude that on balance the final effect on employment constitutes a substantial public interest ground that would justify prohibition of the merger.

243. The fact that the reasonableness of the package was a subject of protracted negotiation and that when opposed by some employees, subsequent adjudication by a specialist body has, on all occasions to date, found in favour of the company, suggests that there are also no grounds for us to impose, in the public interest, any further condition on the company.

244. The remaining question of whether we should insist that employment obligations should follow any divestment will be deferred to our remedies hearing.

Conclusion on Public Interest

245. Here there are potentially two stages of balancing. First within the public interest category, recall the choice of approaches between the joust versus the bypass. The second depends on whether we come to the conclusion that a substantial public interest has emerged in which case we must decide whether that would alter our conclusion on the competition grounds (Section 12 A (2)).

246. In respect of the first stage, it is not necessary for us, given our finding on the employment issues, to consider the parties’ public interest claims as we are not faced with the need to balance or reconcile multiple public interest claims – that is, the employment consequences of the transaction have not led to an adverse finding on public interest grounds and, so, do not have to be balanced against any positive impact that may be found on the parties’ public interest claim.

247. Moreover, we do not need to evaluate the parties public interest claim because, even if it did qualify for recognition, it would not alter the outcome of our finding on the competitive criteria - we have determined that the merger should not be prohibited on competition criteria and hence considerations that, like the public interest claim, go to resurrection are not pertinent.

248. Note also that the public interest benefits to the industry and the country that the parties claim are not negated by any conditional remedies that might be imposed in the affected markets, and, hence, no evaluation of them is needed if no balancing is required between the competitive concerns and the public interest asserted. The same justification for our not evaluating the alleged efficiency gains is equally applicable here.

Finding

249. We find that the transaction will result in a substantial lessening of competition in the proprietary spirits market. For the reasons outlined above, this finding is not affected by considerations of efficiency or public interest.

19 March 2003
D. Lewis Date

Concurring: M Holden and N Manoim

Annexure A

---