South Africa: Competition Tribunal

                                                               Case No.: 46/LM/Jun02

In this case we have to decide whether a series of transactions by Anglo American Holdings Ltd (from now on ‘Anglo’) in which they have purchased up to 34,9% of Kumba Resources Ltd (from now on ‘Kumba’):

1)      

Amount to an acquisition of control by it of Kumba, in which case it is a notifiable merger in terms of the Act; and then, if so,

2)      

Whether the merger should be approved, conditionally approved or prohibited in terms of the criteria set out in section 12A of the Competition Act. In so doing we will have to consider the competition issues raised by the merger in several implicated markets, any efficiency gains it might bring about and the impact on the public interest. All three of these issues are pertinently raised in this case.

From the outset this merger proceeding has been dogged by controversy. Perhaps, this is because the prize, control of the country’s largest iron ore mining company, has been the dream of many suitors. Perhaps, because government at the highest level has been drawn in as the confidante of the players involved, and finally, perhaps, because the prospect of a battle for corporate supremacy, particularly when it is waged in the name of conflicting notions of the national interest, always invites titillation and speculation.

3.      

A brief description of the merger and the events that precipitated it are worth mentioning, not because a long decision requires an overture before it starts, but because some scene setting is necessary in order to appreciate why it is that we have to decide certain issues and why they have taken a particular form.

4.      

Up until March 2001 the firm that is now Kumba Ltd was a division of Iscor On 26 November 2001, Kumba was listed on the JSE.. The integrated firm combined the mining of iron ore with steel making. This arrangement was dictated not by efficiencies, but by political considerations. The economically isolated apartheid government needed to ensure its self-sufficiency in manufacturing by having a domestic steel company. In turn the same logic dictated that the steel company was itself self-sufficient in terms of its supply of iron ore. Hence the integrated Iscor was an industrial giant combining it appears, unusually, both iron ore extraction and steel manufacture.

5.      

The Avmin coup did not happen but the break up of the two companies did. The steel making plants remained in Iscor and through a swap with the IDC, Iscor acquired the whole of Saldanha Steel. The IDC was left with sizable stakes in both the new Iscor (holding 8.8% of the shares) and the mining company to be known as Kumba (holding 14% of the shares). The markets loved the divorce and the value of both shares increased dramatically. As in all friendly divorces the parting spouses, whilst still on civilised terms, negotiated a sophisticated access agreement. As the price for the break up, Iscor negotiated with its erstwhile mining division a complex set of supply agreements from Kumba’s two iron ore mines, Thabazimbi and Sishen. The aim of the agreements was to ensure that Iscor enjoyed as favourable a supply situation post break-up as it had before. Whether it has succeeded in doing so is one of the issues in dispute in this hearing.

8.      

Meanwhile Avmin, concerned over its high gearing, abandoned its ambitious plan for control of the iron ore industry and sold its, not insignificant, stake in Kumba to a foreign owned company known as Stimela. See Financial Mail dated November 9, 2001 article entitled “Menell keeps options open.” The ultimate shareholder in Stimela was Israeli businessman Bennie Steinmetz also a key investor in Arctic Resources, whose shares in Avmin were later purchased by Anglo when it bought Stimela.

9.      

So the stage was set at the beginning of 2002 for the IDC and its partner, Stimela, to take control of Kumba. At that stage their combined holdings would have been 24,5%.

11.     

But it was not to be. In a corporate announcement on 12 March 2002 Anglo American plc announced two significant acquisitions. In the one it had acquired, from Arctic Resources Limited, a 34,9% shareholding in Avmin. In the other, it acquired a significant stake in Kumba by acquiring all the shares in Stimela BVI, which is the ultimate beneficial holder of 10.5% of shares in Kumba, thereby raising its total shareholding in Kumba to 20.1%. Prior to this Anglo had held a 9,6% stake in Kumba. It was, it said, the start of a move to consolidate the Northern Cape iron ore assets of Kumba and Assmang.

12.     

The Anglo blitzkrieg had caught the IDC off-guard and certainly surprised the market. An acrimonious dispute broke out between Anglo and the IDC over the Stimela shares. The IDC suggested that they had been sold to Anglo in breach of their option. Anglo and Stimela’s erstwhile shareholder Murex Holdings S.A, assert that what has been sold is ownership of Stimela not the shares subject to the option. The rights and wrongs of this commercial dispute are not relevant for purposes of our jurisdiction. However it does explain two aspects - why the IDC and Anglo came to have such an acrimonious relationship, and secondly, how the IDC came to feel so strongly that its empowerment ambitions for Kumba, developed over time, had been frustrated. Which of the two set of frustrations to its ambitions more greatly animates the IDC is hard to know, but that these events would make Anglo its’ bete noir is hardly surprising.

13.     

Anglo on 18 June 2002 duly notified two mergers with the Commission, the Avmin and Kumba deals. Although separate transactions, the Commission investigated them jointly and duly made its recommendation to the Tribunal that both mergers be approved without conditions.

14.     

At the first pre-hearing in respect of the mergers held on the 20^th of September 2002, the IDC applied to intervene in the proceedings. Anglo opposed this application vigorously, and ultimately, unsuccessfully. This saga is more fully detailed in other decisions, The two decisions by the Tribunal are: Case No: 45/LM/Jun02 // 46/LM/Jun02 dated 23 October 2003, Tribunal Case No:45/LM/Jun02 // 46/LM/Jun02 dated December 2002 and the two decisions by the Competition Appeal Court are: Case No:24/CAC/Oct02//25/CAC/Oct02 and CAC Case No:26/CAC/Deco2. suffice to say that several Tribunal hearings and two Competition Appeal Court decisions later, the IDC were granted leave to intervene on stipulated grounds. This litigation lasted from September 2002 till 28 March 2003 when the CAC gave its final ruling. The Competition Commission had initially also opposed the IDC’s application to intervene but abandoned this stance after the first CAC hearing.

15.     

Whilst this battle between Anglo and the IDC was continuing in the Tribunal and the CAC there was another twist to the tale. The government and Anglo American announced that they had signed a Memorandum of Understanding in respect of the Northern Cape iron ore fields. Although all parties to this agreement claimed that it is confidential, a public statement indicates common objectives in respect of investment, development of the Sishen – Saldanha railway line, use of another railway line from the Northern Cape to Coega and the advancement of empowerment. Signatories to the agreement from government included three national departments. The IDC, although mentioned in the agreement, is not a party.

16.     

A second pre-hearing was held on the 23 April 2003 and the hearing was set down for 26 to 30 May 2003. As it happened this time was insufficient and the hearing continued on 15,17 and 19 June while closing argument was heard on 6 and 7 August. On the 5^th of May 2003 came a dramatic development. Anglo announced that it was selling its Avmin stake to a consortium comprising Harmony Gold Mining Company Limited and Armgold, citing in its press statement inter-alia as a rationale, the competition concerns that had arisen because of its Kumba acquisition. In its press statement dated 2^nd May Anglo states: “ Subsequently it became apparent to Anglo American that there were concerns about it owning an effective interest in both Assmang and Kumba iron ore assets and combining the iron ore assets in the Northern Cape.” The sale of the Avmin stake reduced the extent of overlaps that would be created by the two mergers originally notified, most notably in iron-ore.

17.     

When we commenced our hearings on 26 May 2003 we were thus asked to consider only the Anglo/ Kumba notification as the Anglo/ Avmin notification had been withdrawn.

18.     

The late development meant that most of the filings, including the reports of the economic experts of the IDC and the merging parties were outdated in many respects as much of the consideration given in them was the significance of overlaps added by the Avmin leg. Nevertheless we are of the view that this deficiency in the record has been rectified by the oral evidence of the witnesses which addressed the Kumba transaction pertinently and from the point of view of fairness by the ample opportunity we have given the IDC to make further filings of expert reports. Since Anglo opposed this latter issue we have had to rule on the admissibility of the late filings and it is to this issue that we turn next.

After the merging parties and the IDC had concluded with their oral evidence and immediately prior to us hearing the closing arguments of the various participants, the IDC filed various additional expert reports which were, we were told, a response to various questions put to their expert economist in cross-examination and to which he had not been given an adequate opportunity to deal. Anglo, for its part, although it had filed further documents in response, objected to the late filings, as the documents were not available at the time neither for comment by its own experts nor for cross-examination.

20.     

The presiding member indicated that the Tribunal would rule on the admissibility of these documents at the time we gave our decision. We have decided to accept this documentation as part of the record. Anglo’s belated announcement that it would be selling its Avmin stake had a profound effect on the emphasis of the record up until that date in respect of the competition issues. The IDC had focussed its opposition to the merger up until then on the creation of what it termed as a local monopoly in iron ore and the merging parties had in turn focussed on the efficiency gains from consolidation in the Northern Cape iron ore industry.

21.     

The announcement thus meant a change in focus by all participants and it was therefore in our view legitimate for the IDC to argue that it required an additional opportunity to address certain of the economic issues. We have decided to admit these additional documents including those filed in response by Anglo to the record.

22.     

That being said we have approached the additional documents with caution, bearing in mind that they were not the subject of testimony during the proceeding and accordingly must be accorded less weight than those documents that were.

23.     

Finally, we have also admitted, and therefore considered, proposals that the IDC made in respect of conditions to be attached to the approval of the merger despite the fact that these proposals were not made during the course of the proceedings. Although the specific conditions were not canvassed, the issues to which they relate were, and for that reason they would have been matters that we, in any event, considered in evaluating the merger.

The merging parties were separately represented during the course of our proceedings. The acquiring firm, Anglo, argued that the transactions amounted to a merger and that the merger should be approved unconditionally. Kumba, which is the target firm, was represented but did not participate in the proceedings in any meaningful manner. Kumba’s legal representative put the issue quite frankly when at the end of proceedings he stated that his client was reluctant to involve itself in a dispute among its shareholders.

25.     

The Commission had initially, when both the Kumba and Avmin transactions had been notified, recommended their approval without conditions. It maintained this position throughout the proceedings and largely made the same case that Anglo did, although the Commission did not call any of its own witnesses. The Commission was also in agreement with Anglo that the series of transactions amounted to a merger.

26.     

The IDC having been given leave to participate in the proceedings, opposed the merger from the outset. This stance did not alter despite the withdrawal of the Avmin leg of the transaction. The IDC was initially of the view that the transactions did not give rise to a merger as there was no evidence that there had been an acquisition of control in the record. It later, after the evidence given during the hearing, altered its position to argue that the ‘notification’ of the merger was defective.

27.     

In the alternative it argued that if the transactions gave rise to a merger this merger should be prohibited on both competition and public interest grounds. Later, as we have already mentioned, the IDC proposed that if we were inclined to approve the merger it should be subject to various conditions, which it suggested.

28.     

Although we have not found for the IDC in this decision, we benefited from its participation as an intervenor given the economic and social significance of this transaction. Our deliberations were enhanced by the fact that information placed before us was subject to vigorous scrutiny and critique, by a party with an adverse interest.

The IDC initially raised a jurisdictional challenge premised on the basis that the transaction contemplated by Anglo did not amount to a merger in as much as the shareholding contemplated did not give rise to a change of control. Later, apparently after hearing the evidence of Anglo’s chief executive officer Mr. Trahar who testified as to the fact that Anglo currently held equity and options amounting to 34,3% of Kumba and that it intended to acquire up to 49%, this attack shifted to the adequacy of the manner in which the transaction was notified and it is this issue which we will now consider. Although Kumba is a new company the evidence concerning the shareholder meetings it has held thus far, suggests that a shareholder with 34,3% equity would command a majority vote at a general meeting. The same would have been true of previous meetings in Iscor in the last few years prior to the unbundling of Kumba.

30.     

When merging parties notify a merger the rules of the Competition Commission require them to do so in accordance with a prescribed form. The form in question, CC 4(2) requires the merging party to provide details concerning the transaction. Question 11 asks for the following information:

In answer to this question Anglo provided the following information:

1)      

On 8 March 2002 AAH (Anglo American Holdings) concluded an option agreement with Murex (attached as Annex C) in terms of which AAH acquired a call option to purchase, and Murex acquired a reciprocal put option to sell, either:

2)      

the shares that Murex holds in Stimela BVI, which is the ultimate beneficial holder of 10.5% of the shares in Kumba); or

3)      

the shares of Stimela SA, provided that Murex ensured that the shareholding in Kumba by Stimela BVI was transferred to Stimela SA; or

4)      

the underlying assets of Stimela BVI, being the shareholder in Kumba;

5)      

in each case, subject to the suspensive condition that the necessary competition approvals were obtained prior to the shares being transferred.

6)      

In terms of a letter dated 10 April 2002, AAH exercised the call option to purchase the Stimela BVI shares. A copy of the letter exercising the call option is attached marked Annexure G….

7)      

AAH has been advised by Stimela BVI that it has concluded a co-operation agreement with the IDC, (the other major shareholder in Kumba, whose current shareholding in Kumba is 14%), in terms of which Stimela BVI and the IDC have agreed to co-operate in respect of certain specific issues with respect to then iron ore assets of Kumba. The parties to the agreement have undertaken to keep the agreement confidential.

8)      

In addition to the Stimela BVI shares, Anglo American has concluded a further option with a third party, in terms of which it has the option to acquire up to an additional 3.8% shareholding in Kumba.

9)      

Anglo American (through ASAC) already holds 9.6% of the equity in Kumba, which it acquired through open market purchases. Furthermore, Anglo American intends to acquire further shares in Kumba as and when opportunities arise. (Our emphasis)

10)     

In summary, when Anglo American has exercised all the relevant options and acquired the relevant shareholdings pursuant to such options, it will hold at least 23.9% in Kumba.

11)     

Accordingly, in the circumstances, Anglo American’s aggregated shareholding together with its board representation would afford it the ability to materially influence the policy of Kumba for the purposes of section 12(2)(g) of the Competition Act as a result of the transaction. (Our emphasis) Suitably paraphrased section 12(2)(g) states that a firm controls another firm if it has the ability to materially influence the policy of a firm in a manner comparable to a person who controls a company either by virtue of owning the majority of its shares or controlling appointment of the majority of the board.

32.     

The IDC argues that this information was incomplete at the time and that, as a result, the notification is a nullity. It argues that information subsequently provided, namely the extent of Anglo’s ambitions about acquiring an interest in Kumba, was not disclosed in the form. Since the information disclosed on the form became the basis on which trade unions and the Minister were notified, they were obliged to respond to incomplete information. In turn, since their responses form part of the Commission’s investigation and subsequent recommendation to the Tribunal, the whole process is tainted by this want of proper compliance.

33.     

Whilst one could not expect firms to notify a merger where their ability to acquire control was at that stage still academic it is not necessary for them to have completed the process of acquisition as a jurisdictional prerequisite to notification. To do so would create burdens on merging firms who would then be faced with the Scylla of not implementing a merger prior to approval, and the Charybdus of not adequately completing a transaction prior to notification. In our view, unless information not notified, materially affects the evaluation of the merger, failure to detail the final series of transactions is not fatal. It might well be that the extent of an acquiring firm’s interest in the target may affect the competition analysis, but that is an issue that may form part of the analysis and does not detract from the adequacy of the initial filing. For instance a firm’s incentives may differ depending on whether it holds 20% or 100% of a company post merger.

35.     

“Anglo American also has prospective interests in other shares in Kumba, which may be acquired in due course…(and later on)...For the purpose of competitive analysis, as the largest single shareholder, Anglo American will be able to exercise material influence over Avmin and Kumba, and those companies in which they have a significant stake (such as Assmang).” See Record page 32.

36.     

We may assume that the diligent reader of the notification will read all the information filed. That reader could have no doubt that Anglo intended through a variety of transactions, some of which had already taken place, to acquire control of Kumba. Granted, Anglo did not disclose how large a stake in Kumba it intended to acquire and that this fact only emerged during the hearing in the evidence of Mr. Trahar, its chief executive officer, but once it had disclosed in the notification its intention to acquire sole control by virtue of some stake in excess of 23.9%, the remaining issues were matters for evaluation of the merger, not to found jurisdiction.

37.     

Thus by way of example, a firm may indicate in its filing that at that stage it has 20% of a company but will have control by virtue of section 12(2)(g). It may be that if the acquirer’s stake did not exceed this figure its incentives would be different than if it owned 100% of the target. For instance the target may be a supplier of the acquirer and the acquirer might have an incentive to use its control of the target to squeeze its margins in favour of itself. This however is a substantive not a jurisdictional issue. If the transaction might give rise to competition concerns only if the firm owned a controlling stake of more than X% of the equity, then the way to cure that problem, might be to either (i) prohibit the merger, on the assumption that the problem may occur later or (ii) to conditionally approve the merger, either by prohibiting the acquiring firm from acquiring a stake above X% or to require it to notify again if it went above X%. In other words, the inherent uncertainty as to the extent of ownership in a notified transaction need not be cured by deciding that the merger is inadequately notified, but rather by the Tribunal appropriately utilising the substantive powers that it has in terms of section 16.

38.     

In this merger, whilst the threshold for Anglo’s ambitions in respect of Kumba is not known with any certainty, Mr. Trahar’s conversation with Minister Erwin notwithstanding there is nothing to bind Anglo legally to this level of shareholding. we have approached the evaluation by not merely evaluating its incentives in respect of its present holding of 34,9% or its proposed holding of 49% but also if it went past that threshold and acquired the remaining equity. Had we felt that the size of the holding would have altered its incentives or behaviour other than as a profit maximising shareholder in Kumba, we would have imposed conditions to this effect. However, we see no reason to do so.

39.     

Whilst there may have been an element of opaqueness in Anglo’s approach to the notification, and even some inconsistency during this process about when control is reached, we are by no means certain that there was information that it did know at the time and that it failed to disclose in order to present the proposal in a more palatable form at the time of filing, lest it stir controversy. On the inconsistency front the IDC makes much of the fact that Anglo had seemingly relied on its existing stake plus the Stimela option and another to reach control yet had a few months later implemented the Stimela option after having obtained a favourable opinion from the Commission that this did not amount to implementation as it did not amount to a change of control. On the lack of adequate disclosure the IDC suggested that an option Anglo had obtained in respect of Deutsche Securities (Proprietary) Limited holdings in Kumba and which was only announced on 22 November 2002, in an announcement by Kumba after the Commission had filed its recommendation, was in existence at the time of the filing of the CC(4) and should have been mentioned then. There is no evidence that this is in fact so and the IDC seem to be relying for this on a remark made by Anglo’s counsel in response to a question from the Tribunal about what Anglo’s holding was at the time of filing. Perhaps the real reason for Anglo’s apparent cloak and dagger behaviour is not an ulterior anti-competitive motive that it wished to conceal from scrutiny during the regulatory process, but a desire to quietly build its stake in Kumba without attracting the attention of rival suitors.

40.     

We find that the merger has been adequately notified, constitutes an acquisition of control, and that the Commission and the Tribunal have jurisdiction over the transaction in terms of the Act. We need not decide various arguments made by Anglo as to whether proper notification is a peremptory requirement in terms of a proper reading of the Act and rules. For the purpose of this decision we have assumed it is, but that it is a requirement that has been met.

Anglo and Kumba’s products overlap in four areas (1) coal (2) mineral sands (3) zinc and (4) iron ore.

42.     

The IDC’s competition concerns related to the latter two markets only, and for this reason although we examine all four, we devote more time to the analysis of the effects of the merger on zinc and iron ore.

Both Anglo and Kumba, through Anglo Coal and Kumba Coal respectively, produce coal. Anglo Coal’s operations in South Africa include eight wholly owned collieries located mainly on the Witbank coalfields. Anglo operates the following coal mines: Bank, Greenside, Goedehoop, Kleinkopje, Landau,
Kriel, New Denmark and New Vaal. Kumba’s operations include three wholly owned collieries, Leeuwpan, Grootgeluk and Tshikondeni.

44.     

Coal is a differentiated product that is categorised according to the degree of transformation of the original plant material to carbon. The ranks of coal from lowest to highest are lignite, sub-bituminous, bituminous and anthracite. Bituminous and anthracite are the only two ranks mined in South Africa, with bituminous accounting for most of the sales.

45.     

Within bituminous coal a distinction is made between thermal coal, also referred to as “steam coal” that is used for power generation and metallurgical coal, referred to as “coking coal, which is used in the production of steel.

46.     

Thermal coal

The four largest producers of thermal coal, excluding Sasol and Eskom’s suppliers, Sasol is excluded because it consumes its entire thermal coal production internally and suppliers to Eskom are excluded because they are locked into long-term contracts. are: BHP Billiton with a market share of 27%, Kumba with a market share of 18%, Duiker (Xstrata) with 11% and Anglo Coal with 9%. Post merger the market share of the merged entity will be 27%.

48.     

Approximately 70% of South Africa’s thermal coal is used to generate electricity or produce synthetic fuels. The two largest consumers of thermal coal in South Africa are Sasol and Eskom and they account for more than 90% of the coal used. Eskom consumes 56% of thermal coal produced in South Africa and Sasol 35%, while Highveld Steel, Iscor and municipal power stations use the remaining 9%.

49.     

It is unlikely that the merger will give rise to a substantial lessening of competition in the thermal coal market. Coal supplies to Eskom are covered by long-term contracts, the prices of which are fixed and indexed to adjust for inflation or cost-plus. Sasol supplies its own coal with BHP Billiton being the only external contractor to also supply Sasol. Subsequent to the filing of this merger Sasol Mining and Anglo Operations, through its coal division, concluded a merger transaction through which Anglo will supply Sasol with Coal from its Kriel opencast Colliery, see Tribunal Case No: 26/LM/May03. Smaller players can source their coal from a number of alternative players such as BHP Billiton, Duiker, Eyesiswe and Kangra.

Metallurgical coal

51.     

Metallurgical coal accounts for less than 3% of coal used in South Africa. Of this percentage 40% is imported.

Firms that use metallurgical coal in their production process, design their operations in a way that requires metallurgical coal with very particular specifications, for instance, Iscor requires and also imports metallurgical coal that has a high measure of “coke strength after reaction”.

53.     

Nearly all of the metallurgical coal sold in South Africa by Kumba is hard coking coal sold to Iscor under long-term contracts. Anglo Coal does not have such reserves and sells 60% of its production to its own related entities. The rest is sold to Siltech and CMI (known as Xstrata-Lydenburg).

54.     

Because of the differentiated use of metallurgical coal there is no direct overlap in this product segment between Anglo and Kumba and they are not regarded as competitors in this product market.

In South Africa three companies produce products from mineral sand mining, namely, Anglo, which owns Namakwa Sands on the west coast of South Africa, Kumba, which owns Ticor SA Ticor SA is a new company that began its operations in 2001. The ownership of Ticor SA is split 60:40 between Kumba and Ticor, an Australian mining company. Kumba’s effective interest in Ticor SA, however, is significantly greater than 60% because Kumba also owns 46.5% of Ticor. on the east coast and Richards Bay Minerals, which is also on the east coast. BHP Billiton and Rio Tinto jointly own Richards Bay Minerals. A fourth player is currently entering the South African market, namely Mineral Commodities, a company from Australia, which will be situated on the east coast.

56.     

The principal products obtained from mining and processing mineral sands are titanium dioxide feedstock, high purity pig iron and zircon. We deal with each one separately.

The three firms mentioned earlier are the only local producers of titanium dioxide. Their market shares are:

Richards Bay Minerals    80%
Namakwa Sands             15%
Ticor                               5%

58.     

Both Richards Bay Minerals and Ticor are situated on the east coast of South Africa, while Namakwa Sands is situated on the west coast.

59.     

According to Anglo its’ Namakwa Sands plant has a locational disadvantage in respect of domestic customers, as it is located on the west coast whilst the major domestic customers like Huntsman, and to a lesser extent Chiesa, are located on the east coast. Presently the only customer domestically that would now find that its choice of suppliers had reduced from 3 to 2 is Afrox, but they do not constitute a major source of consumption, accounting for sales of only R 6 million.

61.     

Whilst the merger will lead to the number of producers declining from three to two this is unlikely to reduce competition substantially given that:

•        

Huntsman accounts for 97% of the consumption and appears from a location point of view committed to RBM as its supplier;
•         ΤηελοχατιονοφΝαμακωαΣανδσσυγγεστσιτισνοταϖιαβλεχομπετιτορφορΡΒΜανδΤιχορ

In South Africa, where Zircon has a limited application, it is used mainly in ceramic applications as a glazer. It is also used in foundries, refractories and TV glass.

63.     

South Africa exports 98% of its Zircon production. In South Africa, Richards Bay Minerals enjoys a market share of 62%, Namakwa Sands 33% and Ticor 5%.

64.     

Post the merger Anglo will have a market share of 38%. However, the merger would have little effect because Richards Bay Minerals will still be the largest producer in the country. Anglo has argued that there is again here limited competition between Namakwa and Ticor not only because location is relevant in respect of this product but also because there are distinctions in the types of zircon produced by the firms making the one firm’s products not suitable as substitutes for the products of the other. Mention is made of Foskor, which requires the particular Zircon produced by Namakwa Sands.

High purity pig iron is an output product from the ilmenite smelting process. Currently only Richards Bay Minerals and Namakwa Sands produce high purity pig iron since Ticor does not have a smelter yet.

66.     

Richards Bay Minerals’ market share is 86% and Namakwa Sands 14%. The price of pig iron in South Africa is at export parity and there are no capacity constraints that limit the ability of South African producers to increase exports. South Africa exports 94% of its high purity pig iron.

67.     

The dispute between Anglo and the IDC on the competition issues associated with this merger starts with zinc.

69.     

Zinc concentrate however is an intermediary product. It needs to be further processed into zinc metal the final product, which is then sold in the market. The process of transformation is achieved through the refining and smelting of the concentrate into metal, which is then sold to customers in various industries.

71.     

The process of refining is not an integrated operation in South Africa and is performed by a single refiner, Zincor. We will for this reason refer to the zinc metal market as a separate downstream market.

72.     

It is common cause between Anglo and the IDC that the merger leads to overlaps between Anglo and Kumba in the zinc concentrate market. It is also common cause that Kumba is, because of its ownership of Zincor, in the downstream market. What is not common cause is whether Anglo is a potential entrant into the downstream market and that whether its choice to enter that market by acquisition rather than independent entry will lead to an increase in the price of zinc metal in South Africa.

73.     

Anglo American operates a zinc/lead mine in the Northern Cape Province, called Black Mountain. It also owns the undeveloped Gamsberg ore reserves. The Gamsberg project has been postponed indefinitely because of the depressed zinc market, which is at a 20 year low. Anglo is also in the process of constructing a zinc refinery in Namibia, called the Skorpion project, as part of an Export Processing Zone. However, this development will not have an effect on South Africa because, according to Anglo, sales to SADC members are prohibited.

74.     

Zincor, a wholly owned subsidiary of Kumba, is the only zinc refiner in South Africa. Zinc is supplied to Zincor by all the South African producers. Kumba also operates the Rosh Pinah lead/zinc mine in Namibia.

75.     

In the upstream market Anglo has a market share of 28% (it owns the Black Mountain mine), Kumba 36% (it owns the Rosh Pinah mine), BHP Billiton 11% (it owns the Pering mine), Metorex 12% (it owns the Maranda mine) and imports 17%. Post the merger Anglo will have a market share of 64% in the upstream zinc market. However, both the Pering and the Maranda mine will stop producing by 2004. Black Mountain is expected to close in 2013 and Rosh Pinah has 8 years of production left.

76.     

Zinc concentrate is not traded on the London Metal Exchange (“LME”) because all of the zinc concentrate produced in South Africa is supplied to Zincor. It is sold by way of long-term contracts between the mine and smelter. For historical reasons mines are paid for only 85% of the zinc in concentrate. This has persisted even though smelters now recover 95% of the zinc in the concentrate. The difference between the 85% paid for and the 95% recovered is referred to as “free zinc”. The zinc that is paid for is called “payable zinc”. The mine pays the smelter a fee known as the treatment charge to transform the concentrate into saleable zinc metal. The treatment charge is set in annual negotiations with reference to a base price, normally an expected SHG price. The treatment charge is typically about 40% of the actual SHG price. The price is, however, tied to the LME benchmark price for Special High Grade zinc (“SHG”). The lack of exports of zinc concentrate from Southern Africa is an indication that the prices paid by Zincor for zinc concentrate have been sufficiently high to make it more economic for independent producers of zinc concentrate to supply Zincor than to sell their product on the export market. We therefore regard the geographic market as international.

77.     

In the downstream market the zinc refinery, in this case Zincor, receives concentrated ore from the various mines and refines it into one of the primary grades of zinc metal. Zincor is regarded as the 3^rd lowest cost producer of zinc metal in the western world, according to international benchmarking by Brook Hunt 2000.

78.     

The local input of zinc concentrate is insufficient to meet Zincor’s demand. Zincor, therefore, imports 4% of its concentrate requirements. It receives 24% of its zinc concentrate from Rosh Pinah, 36% from Black Mountain, 12% from Maranda and 20% from Pering.     

79.     

Nearly all of the output of the Zincor refinery, 87%, is used within South Africa and 13% is exported. The majority of its output, around 24%, is sold to Iscor, its largest customer, and 63% to other domestic customers. Its zinc customers are all located within a 100km radius of its plant, located near Johannesburg. Approximately 10% of domestic consumption is imported, mostly by customers located close to the coast in Cape Town.

80.     

Zinc metal is an internationally traded commodity and is traded on the LME. The price paid for zinc metal in South Africa equals the LME price plus a premium that is individually negotiated between buyers and sellers to reflect the quality of the zinc purchased and logistical issues. Transport only represents 2% of total cost. Thus if the local price is raised by 5%, customers will turn to imports. It is priced at near import parity. Dr Hilmar Rode of Anglo, told the Tribunal that Zincor indicates on its website that it purchased zinc concentrate on international terms, see transcript p. 297. We thus regard the geographical market as international.

81.     

According to the IDC the transaction will have a horizontal effect on the zinc market leading to the diminishing or prevention of future competition in the zinc market.

82.     

Zincor, operated by Kumba, is the only zinc refinery that is operating in the downstream market. According to the IDC there are, from an anti-trust perspective, two possible future entrants that could come into the market and compete with Zincor. The first is Anglo’s Gamsberg project, not yet established, but which in terms of its proposal would have included a refinery, and the second, the Skorpion refinery in Namibia, also being developed by Anglo.

83.     

However, should the Gamsberg project be reactivated it would, according to Anglo, not affect prices in the zinc market.

85.     

86.     

Anglo vehemently deny this and state that the Gamsberg project would not have happened in the present state of the zinc market even absent the merger. Given its location in the Western Cape, well away from the bulk of the domestic customer base for zinc metal which is inland, the project needed to be largely dependent on exports to be viable. Whilst we are in no position to resolve this dispute of fact we need not do so. The evidence turns on whether, if Gamsberg had come on line as a refinery, it would have led to a substantial reduction of zinc metal prices for domestic consumers.

Thus, we find that in light of the relative small affect that a potential entrant would have on prices in the local zinc market, that zinc concentrate can easily be imported The evidence of Dr Rode was that in relation to the value of the product, transport costs were about 3%. and that prices are capped by import parity for zinc concentrate lying between import parity and export parity for zinc metal, competition will not be substantially lessened or prevented in the zinc market.

90.     

If the market is viewed as an international one, then post merger, Anglo would have 3,7% of the zinc concentrate and 2,2% of the zinc metal market, figures of no significance from a competition perspective. By way of comparison with its international competitors in zinc concentrate this is small. Pasminco has 11% and Glencore 12,5%.

92.     

Iron ore is found in a variety of physical forms:

1)      

Fines, which are “sintered” to form sintered fines of 17-18mm average diameter,
2)       Pellets, which are pelletised, to form pellets of 5 -10 mm in diameter, or
3)       Lump, which has a width greater than 76mm.

93.     

Lump ore and pellets can be charged directly to an integrated steel mill’s blast furnace, but fines must be sintered first. The majority of South Africa’s production consists of lump ore and fines.

In South Africa, Kumba and Assmang produce iron ore for sale to steel producers. The other significant player is Anglo American that owns the Mapochs mine, referred to in par. 99 below. Foskor Limited, which is a wholly owned subsidiary of the IDC, produces limited amounts of iron ore as a by-product in the mining of phosphate rock, its primary business. Foskor states that, at present, it is stockpiling the iron ore that it produces for possible future sales.

96.     

Kumba is the largest iron ore mining entity in South Africa. Kumba owns three mines, the Sishen mine in Northern Cape Province and the Thabazimbi mine in the Limpopo Province. The third mine is the Sishen South Mine that is expected to start production in 2004. Kumba also owns the Hope Downs iron ore mine project in Australia.

97.     

The Sishen mine accounts for 90% of Kumba’s production of iron ore and is the 3^rd largest open pit iron ore mine in the world. Most of the output of the Sishen mine, 75%, is exported via Saldanha Bay. The remaining 25% is sold to Iscor Iscor has an undivided share in Sishen mineral rights, which entitles it to 6.25 metric tonnes of iron ore per annum. and Saldanha steel. The volume exported through Saldanha is limited by the capacity of the Orex railway line that connects Sishen with Saldanha. Orex is owned and managed by Spoornet, a division of Transnet. The Orex railway allows for the shipment of 28 million tonnes of iron ore per annum. Kumba has been allocated capacity rights of 23 million tonnes per annum and Assmang 5 million tonnes.

98.     

Thabazimbi sells all of its output to Iscor’s Vanderbijlpark and Newcastle mills and is thus a ‘captive’ mine. The Thabazimbi mine is near the end of its life and the estimated remaining mine life is eight years.

99.     

Currently Anglo American has no direct link to the iron ore mining industry other than its interest in Highveld Steel, which obtains its feedstock ore from its wholly-owned Mapochs Mine. The Mapochs Mine is a vertically integrated operation that produces vanadium rich iron ore, which is consumed entirely by Highveld Steel.

100.    

Post the merger Anglo will have a market share of 78.8%, based on production of iron ore. Assmang will have a market share of 13.6% and Highveld 7.6%.

101.    

Assmang’s iron ore mine is located at Beeshoek in the Northern Cape Province, adjacent to Kumba’s undeveloped Sishen South mine. Assmang exports its iron ore and also sells to Scaw Metals, a subsidiary of Anglo, and to Davsteel an independent steel producer in Gauteng province.

102.    

There are no imports of iron ore into South Africa. The price paid for iron ore in South Africa is approximately 30% less than the import parity level. However, Iscor is protected by two long-term contracts with Kumba, one in respect of Thabazimbi and the other in respect of Sishen. Although the contracts differ, the material term, which is a guarantee that Iscor is supplied iron ore at cost plus 3% is the same for both.

Dr Daniel identified the following competition concerns with the merger:

At a horizontal level he said that the merger would lead to a reduction in competition as the market presently had three players and that post merger there would be only two.

Horizontal issues.

108.    

Dr Daniel examines both the pre-merger and post–merger concentrations in the iron ore market and comes to the conclusion that the markets are highly concentrated pre-merger and even more so post-merger. See p 499 of the Exhibits file where Daniel analyses the HHI results as follows:
•         ΤηειρονορεμαρκετινΣΑισηιγηλψχονχεντρατεδ
         1,800 vs 9,232 in sales
         1,800 vs 7,650 in production
•         ΤηεΗΗΙινχρεμεντισϖερψσιγνιφιχαντ
         50 vs 3,150 in sales
         50 vs 1.198 in production

109.    

This evidence about the concentration level in the industry is common cause. What is novel about his evidence is his theory that Anglo is presently engaged in the iron ore production market through Mapochs and that this has served as a constraint on Kumba. He argued that presently Kumba is constrained from increasing its price to Iscor and others. If prices in steel become uncompetitive and customers switch to Highveld this will decrease iron ore sales domestically for Kumba. Post the merger, however, Kumba will have no such constraint as it will be controlled by Anglo who have the incentive to bring about such an outcome. He further argues that new entrants to the steel market will now be faced with the choice of two iron ore suppliers instead of three.

110.    

Anglo had little difficulty in disposing of this theory, which much like the rest of Dr Daniels evidence, was theoretically sound but lacked any application to the empirical data of the case. The reason that Mapochs is not a competitor to Kumba and Assmang is technical. Its iron ore is not substitutable for that of other iron ore as it has a low content of iron ore and a high presence of other ingredients, which have to be eliminated in the production process of steel. This means that a steel plant using Mapochs iron ore has to be designed to achieve this at great cost. Highveld is the only steel plant in the country at the moment configured to make use of this iron ore. For a new entrant wishing to enter into steel manufacture there is a choice of spending $400m to $500m for a mini mill or $750m for a DRI mini mill, as opposed to one billion dollars for a mill that could use Mapochs ore.

111.    

It appears from the evidence of Dr Mareels that it would only be viable to use Mapochs iron ore if one was to enter the market primarily using vanadium as the core product and steel as a secondary by-product, as Highveld do. In addition it appears that the Mapochs mine ore will last for another 40 years at the current rate of production. If that rate were to be increased to supply another firm this would greatly reduce the life of the mine.

112.    

A new entrant is not faced with a diminished choice in supply. Prior to the merger the entrant would have to choose between a factory designed for Mapochs iron ore, in which case it would be faced by a monopoly supplier or the more conventional plant, in which case it would face a duopoly in Kumba and Assmang. The situation is no different post merger and the fact that Anglo controls Kumba would not alter the equation.

113.    

In our view Mapochs does not compete with Kumba or Assmang because of these technical difficulties, and accordingly the IDC’s objection fails at the first hurdle and we need not examine the issue of incentives here. Incentives are examined later in the section dealing with vertical issues.

115.    

We will firstly consider the foreclosure or raising rivals’ costs theory. It will be recalled that earlier, we stated that post the unbundling of Kumba, Iscor had secured its supply position by entering into long term supply agreements with Kumba for supply from Sishen and Thabazimbi. These contracts are highly favourable to Iscor as they ensure that Iscor is supplied on a cost plus 3% formula. Dr Daniels argues that notwithstanding the existence of these contracts, they allow an Anglo-controlled Kumba enough “wiggle room” to interpret them in a manner that would permit them to raise the level of costs so as to raise the costs of steel production for Iscor. When the Commission did its initial investigation of the merger it met with Mr Van Niekerk, the chief executive officer of Iscor, who according to the Commission’s counsel expressed no concerns about the merger and gave as the reason the fact that Iscor was protected by its contracts with Kumba. This meeting took place in June or July 2002 . On 14 May 2003,approximately one week before the merger hearing commenced, Mr Van Niekerk wrote to the Tribunal to state that Iscor had serious concerns about the merger. Concerns were that under Anglo, Kumba might not invest sufficiently in iron ore expansion at Sishen South and might favour investment in Australia instead. Secondly, he raised concerns that Iscor’s strategic information might become available to its competitors in the Anglo stable. He did not however offer to testify at the hearing nor ask for Iscor to intervene. Iscor was represented throughout the proceedings by its attorneys who had a watching brief. What accounts for the Iscor volte- face is difficult to discern - all of these issues would have been known to Iscor when it met with the Commission in 2002. Anglo suggest, cynically, that the change is explicable by the change in dynamics on the Iscor board- LNM is now the dominant shareholder and Mr Ngqula, the IDC’s chief executive officer, is now Iscor’s chairperson. We are not in a position to comment on this but were Iscor as concerned about the merger as the tone of the letter suggests, we have no doubt that it would have been more robust in communicating this point of view to us. Given that we find that Anglo has no incentive to raise Iscor’s costs, we need not probe this further.

116.    

There was much dispute as to whether the contracts properly interpreted allowed for this flexibility. We need not involve ourselves in this interpretative exercise, as Dr Daniel’s case is dependent on whether Anglo has, in addition to the contractual wiggle room, an incentive to behave in this manner.

Kumba as we have noted produces iron ore from two mines, Sishen and Thabazimbi. As about 75% of Kumba’s total output is destined for the international market, and as the Thabazimbi mine is contracted to sell 100% of its output to Iscor, these exports originate entirely from the Sishen mine and require the incentive structure in both mines to be analysed separately.

118.    

The cost of iron ore to Iscor is calculated on the basis of a cost plus 3% management fee. Dr Daniel is of the view that Anglo indeed has the incentive to raise costs of Iscor by artificially or actually increasing mining costs at both Sishen and Thabazimbi. This increase in cost to Iscor will then compel Iscor to raise the price of steel and hence benefit Highveld and Scaw, either by having consumers switch to their products or enable them to raise their domestic price without losing business to Iscor. Dr Daniel recognizes that if Anglo were to increase the costs incurred at Sishen this would impact adversely on exports that have to be sold at world prices or export parity. However, he was also of the view that the benefits would arise from an increase in profits from higher prices or volumes in the downstream market for Highveld and Scaw and that these benefits would outweigh the costs.

119.    

Anglo’s expert, Dr Stillman, in response to this theory then performed a calculation based on figures that they had to test Dr Daniel’s central proposition. Upon calculating the impact of a hypothetical 10% increase in iron ore costs by Kumba on the upstream export market and the downstream steel market it was shown by Dr Stillman that the costs vastly outweighed the benefits to Anglo of pursuing this particular strategy of raising rivals’ costs. Even though the calculation assumed Highveld would be able to pass on the full cost increase into domestic prices, the loss of sales and profits incurred by Kumba on the export market dwarf the increase in Highveld’s sales into the domestic market. Therefore we can conclude that the costs far exceed the benefits of this strategy to Anglo. The IDC failed to contradict the data presented and therefore we can comfortably dismiss the likelihood of this scenario occurring at the Sishen mine.

120.    

With regard to the Thabazimbi mine on the other hand, the incentive to raise costs to Iscor does not appear to exist. The contract between the mine and Iscor rather provides a strong incentive to reduce costs. Under the contract, if costs are reduced, Kumba gets to keep [confidential information] of the cost reduction. Once again a numerical example demonstrates that under the most stringent assumptions, the incentive to reduce costs outweighs any benefit to Anglo at Highveld of allowing costs to be increased at Thabazimbi.

121.    

Even if Anglo were to operate in such an irrational manner and allow costs to rise at Thabazimbi, given that only a small amount of iron ore is sourced from the mine, the calculations show that a 10% increase in the cost of iron ore sourced from the Thabazimbi mine would only increase total steel making costs at Iscor by 0.3%.

122.    

Finally, given the agreement between the expert witnesses that the marginal cost of steel production in South Africa is determined by export parity, higher iron ore costs would fail to impact on domestic prices charged by Iscor for steel. Therefore, in conclusion, on the basis of evidence before the Tribunal there is little support for the contention that Anglo would follow a strategy of raising rivals’ costs in steel making at either of the two mines.

123.    

The foreclosure argument is a variant of the above argument for raising rivals’ costs. The argument made was that Anglo would raise costs to Iscor in order to benefit Anglo’s Highveld steel operation. This would be achieved by raising costs to other existing steel manufacturers such as Davsteel. According to Dr Stillman, Davsteel accounts for less than 10% of local steel sales. It currently buys iron-ore from Assmang, however, 85% of its feedstock comes from scrap metal. The iron-ore portion of Davsteel’s total cost is less than 1%. (See page 59 of the transcript.) Anglo would also raise the costs of new entry in order to either foreclose such entry or reduce the profitability of such entry.

124.    

Dr Daniel suggested that while complete foreclosure may not be profitable to Anglo, partial foreclosure would be if Anglo were maximizing profits. Unfortunately once the calculations of the benefits and costs of raising costs are done, given the importance of exports in the Kumba portfolio, there is no possibility of even partial foreclosure being profitable. Various theoretical models of partial foreclosure were proposed during the hearing, but once these models were calibrated to the actual data, the model proposed by Dr Daniel became an empirical impossibility. (See page 551 and 552 of the Exhibits file.) Interestingly these calculations suggested that given the global focus in Kumba’s business the major disciplining factor in the iron ore market is the pressure in world markets to keep costs down in order to be competitive.

125.    

126.    

Dr Daniel suggests that the possibilities for collusion are enhanced by the possibilities for information sharing that the Iscor /Kumba supply contracts provide and secondly, that the ability to raise Iscor’ s price via Kumba, gives a post-merger Anglo a weapon for enforcing the cartel if Iscor cheated, a weapon it lacks pre-merger.

Both experts are at least in agreement that three elements are required for a stable collusive agreement. Firstly, the ability of the parties to reach agreement to collude; secondly, the parties need to monitor compliance of the agreement by the members because of the incentive to cheat and; thirdly, there needs to be an ability to enforce the collusive agreement so that if cheating takes place the members would be able to discipline that member.

128.    

Dr Stillman in his analysis, focussed only on the possibility of collusion whereby the members of the cartel agree to restrict output in order to have higher prices in the steel industry for the benefit of the members.

129.    

In order for such an arrangement to succeed, Anglo will have to know how much steel Iscor plans to sell in the local market because exports would be excluded in this scenario. Dr Stillman doesn’t doubt that Anglo could learn from its control of Kumba about Iscor’s expansion