EC, December 17, 2008, No M.5046

THE COMMISSION OF THE EUROPEAN COMMUNITIES,

Having regard to the Treaty establishing the European Community, Having regard to the Agreement on the European Economic Area, and in particular Article 57 thereof, Having regard to Council Regulation (EC) No 139-2004 of 20 January 2004 on the control of concentrations between undertakings (1), and in particular Article 8(2) thereof, Having regard to the Commission's decision of 17 July 2008 to initiate proceedings in this case, Having given the undertakings concerned the opportunity to make known their views on the objections raised by the Commission, Having regard to the opinion of the Advisory Committee on Concentrations (2), Having regard to the final report of the Hearing Officer in this case (3), Whereas:

1. INTRODUCTION

1. On 12 June 2008, the Commission received a notification pursuant to Article 4 of Council Regulation (EC) No 139-2004 of 20 January 2004 on the control of concentrations between undertakings (hereinafter "the EC Merger Regulation") of a proposed concentration whereby the cooperatives Zuivelcoöperatie Campina U.A. (hereinafter "Campina") and Zuivelcoöperatie Friesland Foods U.A. (hereinafter "Friesland Foods") merge by way of a full legal merger. Campina and Friesland Foods are collectively referred to as "the notifying parties".

2. After examination of the notification, the Commission adopted a decision on 17 July 2008 (hereinafter "the 6(1)(c) Decision"), concluding that the operation falls within the scope of the EC Merger Regulation and raises serious doubts as to its compatibility with the common market and the functioning of the EEA Agreement and initiated proceedings pursuant to Article 6(1)(c) of the EC Merger Regulation.

3. At the request of the notifying parties, the time period for taking a decision in this case was extended by 5 working days on 18 September 2008, pursuant to the second subparagraph of Article 10(3) of the EC Merger Regulation.

4. On 3 October 2008, a Statement of Objections was sent to the notifying parties pursuant to Article 18 of the EC Merger Regulation.

5. Friesland Foods and Campina replied to the Statement of Objections on 17 October 2008.

6. On 21 October 2008, at the request of the notifying parties, an Oral Hearing took place.

7. At the request of the notifying parties, the time-period for taking a final decision in this case was extended by one working day on 27 October 2008, pursuant to the second subparagraph of Article 10(3) of the EC Merger Regulation.

8. On 28 October 2008, the notifying parties offered commitments with a view to rendering the proposed concentration compatible with the common market. These commitments were modified and the final version of the commitments was submitted to the Commission on 27 November 2008.

2. THE NOTIFYING PARTIES

9. Campina and Friesland Foods are the two largest dairy cooperatives in the Netherlands. Both collect raw milk and process raw milk into consumer and industrial dairy products.

10. Campina, with 6,885 farmers as members (2007) (4) has activities in fresh dairy products, cheese, butter, fresh and long-life flavoured drinks, and emulsions in various countries in Europe, North and South America and Asia.

11. Friesland Foods counts 9,417 members (2007) (5) and sells dairy products for consumers in Europe, the Middle East, Asia and Africa and ingredients for professional and industrial customers worldwide.

3. THE OPERATION

12. The planned transaction would take the form of a full legal merger between the two cooperatives leading to the establishment of a single cooperative. It therefore constitutes a concentration as defined in Article 3(1)(a) of the EC Merger Regulation.

13. The merger agreement was approved by the members of both Campina and Friesland Foods on 7 May 2008 and was signed by the Chairmen of the Boards on 8 May 2008. The new entity would be named FrieslandCampina and would continue all the business activities of Friesland Foods and Campina.

4. COMMUNITY DIMENSION

14. The undertakings concerned have a combined aggregate world-wide turnover of more than EUR 5 billion (Campina: EUR 4 032 million; Friesland Foods: EUR 5 075 million). Each of them has a Community-wide turnover in excess of EUR 250 million (Campina: EUR 3 362 million, Friesland Foods: EUR 3 124 million), and none of the notifying parties achieves more than two-thirds of its aggregate Community-wide turnover within one Member State. The notified concentration therefore has a Community dimension.

5. INTRODUCTORY REMARKS ABOUT THE DAIRY SECTOR

15. The dairy sector comprises a series of interrelated product markets, reflecting the wide variety of milk-based end products. The typical business model for dairy companies, notably dairy co-operatives, is to valorise the raw milk collected from farmers into a wide variety of dairy products. The common raw material - raw milk- means that prices of dairy products follow similar trends.

16. Raw milk consists of several nutritional components: fat, proteins, lactose (milk sugar) and minerals. For some dairy products, only the non-fat components (notably proteins and lactose) are used. Other products, notably butter and cream, are made from the milk fat. Many key products such as cheese and milk contain a mix of fat and non-fat components. Some products - in particular cream, buttermilk and whey - are, in essence, by-products resulting from the production of the primary dairy products such as drinking milk and cheese. The diagram below shows how the fat and non-fat components of raw milk can be used for different applications.

Figure 5-1: The different applications of milk's components

17. Most dairy products are household consumer products, normally sold on the retail market whereas some dairy products are bought by industrial customers to be processed into finished products. The dynamics of these different dairy markets are not identical and competitive conditions may differ depending on product characteristics (for example, perishability) and the existence of strong brands. In many European markets, liquid milk has become a low profit product, with a large proportion of private label products which are sold under distributors' brands.

18. The production and sale of many dairy products within the Community is subject to the Community's common organisation of the market for milk and milk products6. The common market organisation has a particular impact on the production of raw milk and on the imports and exports of dairy products in the Community.

19. For many decades, the objective of the Common Agricultural Policy (CAP) for milk and dairy products was realised by raising the price level for these products (including milk and milk products) above world market price, thus ensuring farmers a direct income from selling agricultural products that would allow a fair standard of living. To this end, a number of basic instruments were used, such as: (i) fixed import duties to protect the desired price level within the Community against lower world market prices; and (ii) export restitutions covering the differences between the Community prices and world prices.

20. The system of milk quotas has been a corner stone of the CAP for milk and dairy products. Quotas are allocated to individual farmers (but administered by buyers of raw milk who, for that purpose, need to be registered by the competent authorities of the Member States). As for 2007, the total quota for the EU-27 amounted to 142 million kg out of which 11.05 million tons were attributed to Dutch farmers. In 2006-2007, deliveries of raw milk in a large number of production regions were below quotas (7). This was the case in the Netherlands where total production in 2007 was roughly 10.94 million tons. As of 1 April 2008, the overall quota was increased by 2% (an increase of 2.5% in 11 Member States). In its meeting of 18-20 November 2008, the Agriculture and Fisheries Council decided on an increase of milk quotas by 1% per year in 2009, 2010, 2011, 2012 and for the marketing year 2013-2014, to prepare for the planned disappearance of the milk quotas in 2015. Two interim reports by the Commission will assess the state of the sector at the latest in June 2010 for the first one and at the latest in June 2012 for the second one.

21. The milk quota regime imposes a ceiling for total Community dairy production and for total production for each individual Member State. However, quotas can be transferred between dairy farmers within a single member State (but not from one Member State to another). This possibility of transferring quotas, at least within one Member State, allows for some flexibility in milk production at farmers' level.

22. Strong demand growth, structural reductions in production and a series of more temporary supply restrictions caused a rise in global dairy prices in 2007 to record levels. Although globally-traded milk volumes only represent 6-7 % of all milk produced, the global market price developments have a crucial impact on all dairy and milk prices. The European Union's role in the global dairy trade has traditionally been linked to the CAP, without which exports of butter and skimmed milk would not have been competitive. As dairy markets are liberalising, the focus can be expected to expand to include cheese, value added ingredients and formulated milk powders. (8)

23. As such, the market conditions in 2007 were in many ways exceptional, as global prices exceeded the Community price level for the first time. These global price developments also resulted in higher milk prices paid to farmers. Throughout Europe in 2007, the price of raw milk increased by about a third compared to 2006. (9) In 2008, worldwide prices of dairy products have dropped down to roughly 2005 levels.

6. PROCUREMENT OF RAW MILK

6.1. Introduction

6.1.1. The collection of raw milk

24. Both Campina and Friesland Foods procure raw cow milk from farmers, which they then process into consumer and industrial dairy products. They are dairy cooperatives and hence mainly collect raw milk from member farmers.

25. Raw milk is milk which has not undergone any treatment (other than cooling) and has a perishable nature. It is produced by dairy farmers, who normally milk cows twice a day. Raw milk is subsequently stored in milk storage cool tanks at the farm which reduces its temperature from 37° C to 4° C. The raw milk is collected at least once every three days by milk trucks which deliver it to processing facilities, where it is stored for a maximum duration of 36 hours prior to further processing.

26. Friesland Foods and Campina use specialized transport companies to carry out the collection and transport of raw milk in isolated (non-refrigerated) transport trucks. The notifying parties have different organisational and logistic structures: Campina plans its daily and weekly truck stops based on the quantity of raw milk requested from factories whereas Friesland Foods makes only an allocation plan per transport company. For Friesland Foods, the average truck collects 160 tons of raw milk daily, which means that the average truck makes four to five trips per day.

27. Raw milk is a homogenous product (10) for which, at least for most of its applications, there are no substitutes. Many dairy products are processed from raw milk but the relative importance of raw milk as input varies. Cheese constitutes by far the most significant primary application for raw milk: 50% of raw milk produced in the Netherlands is used for the production of cheese; 9 kg of raw milk is needed for the production of 1 kg of cheese. With regard to other main downstream products, 13% of the raw milk collected in the Netherlands is used to manufacture milk (fresh and long-life) and milk derivatives and 12% is used to produce milk powder.

28. The chart below shows the split of dairy products processed from raw milk in the Netherlands.

Figure 6-1: Split of dairy products processed from raw milk (NB: EVAP is evaporated milk). Source: Power point presentation from Campina during the site visit in Rijkevoort, 26 August 2008.

6.1.2. Relations between member-farmers and the cooperatives

29. As dairy cooperatives, both Friesland Foods and Campina have with their member-farmers a dual relationship. First, the member-farmers of both Campina and Friesland Foods are the owners of their respective cooperatives since the cooperatives "capital" has been supplied by these members. Most of this "capital" stems from retained earnings.

30. Member-farmers are involved in the governance of the cooperative. According to the notifying parties, farmers would be divided into 21 geographic districts in the planned FrieslandCampina cooperative. Each district would have a district council of at least 10 members who would be elected by farmers. These 210 members would form the Members Council of the cooperative. The Members Council, which consists of member-farmers who have been elected by all the members, would appoint the Board of the Cooperative (the decision-making body) and would have the right to approve or veto certain decisions of the Board (amending articles of association, approving annual accounts and some investments and acquisitions) (11).

31. Furthermore, there would be a Cooperative Council which would act as an advisory body to the Board, in particular with regard to investments, acquisitions and mergers. The graph below shows the details of the cooperative structure post-merger.

Figure 6-2: Structure of the cooperative FrieslandCampina

32. Second, farmers are also suppliers to the cooperatives. According to the respective articles of association, member-farmers are obliged to deliver all their raw milk to the cooperatives and both Friesland Foods and Campina are obliged to collect all the raw milk that their members produce, irrespective of market conditions. The result of this reciprocal obligation is that there is a continuous flow of raw milk from farmers to dairy plants that the cooperatives collect and process into dairy products.

33. Cooperatives may purchase additional milk from non-members but this additional procurement may not replace the milk procured from the members. The total quantity of raw milk collected from non-members in the Netherlands where the procurement activities of the notifying parties overlap, did not exceed [0-5]*% in 2005, 2006 and 2007.

34. In order to assess the impact of the merger regarding procurement of raw milk, the specificities of the cooperative structure between the suppliers (the farmers) and the customers (the cooperatives) must be taken into account. Two points are of particular interest with regard to incentives for farmers to leave or join the cooperative, namely (i) the calculation of the milk price paid by the cooperative to the farmers and how the profits raised in a given year are distributed within the cooperative and (ii) the entrance and withdrawal requirements imposed on farmers to join or leave the cooperative. These issues will be dealt with in section 6.1.3. and 6.1.4.

6.1.3. Calculation of the price paid for raw milk to member-farmers by the merged entity

35. Currently, the price that Friesland Foods and Campina pay to their member-farmers for raw milk is, in principle, calculated ex-post to reflect the financial performance achieved by each cooperative. For Campina, the raw milk price is based on the net result of the cooperative with part of the net profit reserved for re-investment in the cooperative. The remaining net profit is paid to the member-farmers through the same average milk price, irrespective of where these members are located (12). For Friesland Foods, the raw milk price is determined on the basis of an index of four of the largest cooperatives in North-Western Europe (including Campina). In addition to the milk price, Friesland Foods pays a dividend to its current and retired members at the end of each fiscal year. The amount paid depends also on the business results of the cooperative and is the same for each member.

36. The price paid by the merged cooperative to member-farmers would also depend, to a certain extent, on the business results of the cooperative. Clause 9.2 of the Merger Agreement (13) provides that the milk price paid to the member-farmers would consist of two elements:

(a) the guaranteed milk price, which would be independent of the financial performance of the company and would consist of a weighted average of the raw milk price paid by dairy companies in Denmark (Arla), Germany (ZMP prices), Belgium (Milcobel) and the Netherlands (DOC, Cono, Bel-Leerdamer), these milk prices being weighted on the basis of the quantity of processed milk in the entire Member State in question; and

(b) the performance payment (or aggregate supplement), which would equal the net profit, after deduction of the amount that will be added to the reserves of the company. The aggregate supplement would be divided by the total number of kilogrammes of raw milk delivered by member-farmers and then paid to these members in addition to the guaranteed milk price.

37. The performance payment would be designed as follows: the net yearly profit would be calculated on the basis of the guaranteed milk price and after deduction of the interest payment on the Member Bonds and the Member Certificates (see recital 38). 75% of this net profit would be added to the reserves of the company whereas the remaining 25% would be paid to the member-farmers as a performance payment (14). Of those 75% of the net profits that are reserved, 60% per year would just be added to the reserves of the company. The remaining 15% would also be added to the reserves by means of the issuance of new Member Bonds to members.

38. On top of the milk price, farmers would also be granted member bonds and member certificates. Bonds (freely transferrable) and certificates (non-transferrable) are financing instruments which are awarded to member-farmers on the basis of the quantity of milk delivered by them in a given year. The compensation or interest rate payable on such instruments for financing the company is independent of the quantities of milk that such investors deliver to the company. Certificates will be converted into bonds upon termination of membership (compulsory) or on a voluntary basis during membership. In both cases, this conversion would have a fiscal implication for the member and sale of bonds would be possible but not mandatory unless the member terminates his membership (15). Therefore, members would be renumerated for financing the company through annual interest on bonds. They would also have the possibility of cashing the value of these bonds upon termination.

39. It follows that a part of the remuneration paid to farmers (the performance payment and, less directly, member bonds) would depend on the business results of the new entity in the downstream dairy markets where it would be active. This is not surprising given that the purpose of a dairy cooperative is to valorise the raw milk produced by these members by processing it into dairy products. The price paid to farmers cannot therefore be compared to a "normal" market price which would be solely influenced by raw milk supply and demand factors.

40. This method of determination of the milk price creates a strong link between the downstream markets for processed dairy products and price paid to farmers by cooperatives for raw material, which is relevant for the assessment of this case.

6.1.4. Entrance and withdrawal policy

41. Today, dairy farmers who wish to become members of the Campina or Friesland Foods cooperatives must generally pay an entrance fee, that is, a non-recurring amount per approximately 100 kg of raw milk that will be delivered by the individual member in the first year of its membership. In the case of Campina, new members must pay EUR 10.5 per 105 kg of raw milk (out of which EUR 6 are a capital contribution, reimbursable upon exit, while the remaining EUR 4.5 constitutes sunk costs). For Friesland Foods, new members must pay EUR 9.95 per 100 kg. Entrance fees are intended to compensate existing members for the value created by the cooperative by investing the retained profits of the existing members. The notifying parties explain that new members are "boarding a moving train" since former members have created value by investing in the cooperative.

42. Member-farmers who wish to withdraw from either Campina or Friesland Foods have to comply with the resignation requirements, including a notice period (6 months prior to 31 March or 6 months prior to the end of the financial year for Campina; 3 months prior to the end of the financial year for Friesland Foods) (16). Upon withdrawal, a member loses his prorata entitlement to the profits of the cooperative.

43. A similar mechanism will be put in place in the planned FrieslandCampina merged cooperative. As stated in the Merger Agreement, an entrance fee of EUR 4 per 100 kg of raw milk as a one-off contribution would be requested. This entrance fee would constitute sunk costs. In addition, the new member would have to meet objective entrance criteria, which still have to be formulated (a possible criterion might be a proven track record of having had no major problems with quality).

44. With respect to withdrawal requirements, a membership may only be cancelled as of 1 January of each year with a notice period of three months (a member would therefore have to submit a notice of termination before 1 October of the preceding year).

45. The withdrawal and entrance policies exercised by the notifying parties, which would continue after the merger, create barriers for farmers in relation to switching from one cooperative to the other. As the notifying parties themselves stated, "Another aspect of dairy cooperatives which enhances inflexibility of the market for the procurement of raw milk is the fact that cooperatives pursue an entrance and withdrawal policy" (17). The fact that farmers are only allowed to leave the cooperative on a given date and with a notice period prevents farmers from taking advantage of one-off opportunities on the market to deliver their milk.

46. This issue of flexibility for farmers is particularly relevant for the assessment of the merger on the procurement market since it influences farmers' incentives to leave or to join the merged entity and, by extension, influences the decisions of independent farmers or members of competing cooperatives. This can be expected to have an impact on access to raw milk for competitors.

6.2. Relevant product market

47. The notifying parties submit (18) that a separate product market for the procurement of raw milk can be identified, which is in line with the Commission's findings in previous cases (19).

48. The notifying parties also submit (20) that a distinction must be drawn between the procurement of "conventional raw milk" and the procurement of "organic raw milk". "Conventional raw milk" is all milk produced under general legal conditions. Organic raw milk is produced under additional legal conditions, as defined: Council Regulation (EC) No 2092-91 of 24 June 1991 on organic production of agricultural products and indications referring thereto on agricultural products and foodstuffs (21). Qualitative requirements for the production of organic raw milk are laid down in this Regulation and include, inter alia, the following prescriptions: no use of fertilizers, pesticides or GMO (genetically modified organisms); animal welfare requirements; and compulsory grazing of cows in spring, summer and fall.

49. These differences in production methods are reflected in the prices paid to farmers. Farmers who supply organic raw milk are paid a supplement of roughly [20-30]*% (22) of the payout price for conventional raw milk. This supplement is justified by the extra costs incurred by the farmer who produces organic raw milk instead of conventional raw milk.

50. On the demand side, the market investigation has confirmed that, for milk processors, organic raw milk and conventional raw milk are not substitutable. The processing of organic raw milk must be strictly separated from the processing of conventional raw milk. Whilst conventional raw milk may not be used in the production of organic dairy products, the reverse can be true. However, given the price differences between both types of raw milk mentioned above, it is economically unattractive to use organic raw milk to produce non-organic dairy products.

51. On the supply side, organic dairy farmers have no incentives to switch to conventional raw milk production, given the price premium they obtain and the investments they have made to produce organic raw milk. On the other hand, switching to organic raw milk production is an option for a conventional dairy farmer but it requires significant investments in grasslands (more extensive use) and, on average, a 2-year transition period (23).

52. Therefore, it is concluded that the procurement of conventional raw milk and the procurement of organic raw milk constitute separate product markets.

6.3. Relevant geographic market

53. With regard to procurement of raw milk (conventional and organic), the notifying parties' activities overlap only in the Netherlands. In the EEA excluding the Netherlands, Campina collects milk from member-farmers in Germany and Belgium whereas Friesland Foods procures raw milk from non-members in Hungary, Romania and Greece. Campina transports approximately [...]* million kg of raw milk per year from Germany to the Netherlands (mainly to its factory in Veghel) and [...]* million kg per year from Belgium to the Netherlands (mainly to its factory in Eindhoven). These volumes are negligible in comparison with the total volume of raw milk purchased by the notifying parties in the Netherlands (more than 8 000 million kg per year). The merger, therefore, has no significant impact on the procurement market outside the territory of the Netherlands and the assessment thus focuses on the Netherlands. This is in line with the notifying parties' views, although they submit that the relevant geographic market is narrower than national, as explained below.

54. In their response to the Statement of Objections (24), the notifying parties argued that there is no overlap between the notifying parties' activities in the procurement of raw milk within the Netherlands (see recital 85). Therefore, if the fact that there is no overlap in the activities of the notifying parties outside the Netherlands constitutes an argument for the Commission for the delineation of the geographic market, the same argument should apply to the situation within the Netherlands.

55. First, this statement is in contradiction with the position of the notifying parties, as stated in the notification: "Thus, irrespective of whether the geographic scope of the market for the procurement of conventional raw milk is defined as national or regional, the only member State in which the market for the procurement of the conventional raw milk is affected is the Netherlands. Therefore, the parties will not in their analysis address national and regional markets outside the Netherlands" (25).

56. Second, contrary to what the notifying parties claim in their reply to the Statement of Objections, national borders are of relevance in the procurement of raw milk, especially with respect to raw milk quotas. As mentioned, raw milk quotas may be transferred between dairy farmers within a single Member State but not from one Member State to another. Furthermore, these quotas are managed by a governmental or semi-governmental body at national level (26). Consequently, the two-stage reasoning proposed by the notifying parties in the notification (first an identification of the Member States where the notifying parties' activities overlap and then an assessment whether the geographic market in this given country is national or sub-national) and adopted by the Commission is deemed relevant for the delineation of the geographic market in the present case.

6.3.1. Procurement of conventional raw milk: geographic market definition submitted by the notifying parties

57. With regard to the geographic market, the notifying parties argue in the Form CO (27) that the markets for the procurement of conventional raw milk are sub-national in scope and fully coincide with the regional scope ("working area") of the cooperatives. Friesland Foods collects raw milk in the provinces of Groningen, Friesland, Drenthe, Overijssel, Gelderland and Flevoland (eastern and northern parts of the Netherlands) (28). Campina collects raw milk in the provinces of Zealand, North Brabant, Limburg, South Holland, North Holland, Utrecht and Flevoland (southern and western parts of the Netherlands) (29). The map below provides an overview of the provinces of the Netherlands.

Figure 6-3: Map of provinces of the Netherlands

58. Together, the working areas of Friesland Foods and Campina cover the entire territory of the Netherlands, with a small overlap in the province of Flevoland.

59. In arguing that the markets for the procurement of conventional raw milk are sub-national in scope, the notifying parties put forward several arguments to support their views. First, according to the notifying parties, the costs of collecting raw milk are influenced by both the distance which raw milk needs to be transported (and in particular the distance between the first or the last farmer and the processing unit) and the density of farmers in the collecting area, which has a strong influence on the efficiency of the raw milk collection process. These factors limit the scope of the geographic market to a local area within which transport distance is minimized and truck loads optimized.

60. Second, the notifying parties submit that competition conditions are homogenous for dairy farmers in each working area, since competing cooperatives and private buyers set their purchase price by reference to the price paid by the leading cooperative. Thus competition conditions in the Friesland Foods area are homogenous and significantly different to conditions in the Campina area. The notifying parties admit that the merged cooperative would apply the same pay-out price within the merged working area, that is, the entire territory of the Netherlands. This implies that the geographic market would change as a result of the merger.

61. Lastly, the notifying parties contend that the strong links between the cooperatives and the dairy farmers and the lack of flexibility within the raw milk market limit incentives to switch to other buyers. The notifying parties explain that large dairy cooperatives such as Friesland Foods and Campina have no interest in recruiting new members. Attracting new members, in particular outside the existing working area where the concentration of farmers would be smaller and the costs of collection of raw milk higher, would result in a dilution of profits generated for existing members. This is the reason why, according to the notifying parties, Friesland Foods and Campina do not use their pay-out price as a competitive tool to attract new farmers. This is evidenced by the fact that within the last ten years, only one member has switched from Friesland Foods to Campina and not a single member has switched from Campina to Friesland Foods (30).

62. Lastly, the notifying parties indicate that the Dutch Competition Authority (de Nederlandse Mededinggingsauthoriteit or NMa) has conducted an in-depth enquiry into the Dutch market for the procurement of conventional raw milk (31) which concludes that the geographic market could indeed be considered sub-national and limited in scope to the self-defined working area of the cooperatives.

6.3.2. Procurement of conventional raw milk: Geographic market definition retained by the Commission on the basis of the market investigation

63. The results of the market investigation indicate that the relevant geographic market is not limited to the self-defined working area of each cooperative, but rather, is national in scope.

6.3.2.1. Costs of procurement of raw milk are influenced by a number of factors besides distance.

64. The efficient procurement of conventional raw milk is constrained by large distances. The market investigation has shown that collection costs are linked to transport distances and clusters of farmers in a given area. Even if conventional raw milk could theoretically be transported over longer distances, the average distance between farmers and processing units is roughly 40-50 km (32).

65. However, distance is not the only factor which has an impact on collection costs. Another element affecting collection costs relates to the size of the processing or production plants located in the area. Large processing plants can benefit from economies of scale as well as economies of scope in the collection of raw milk. This results in a broader range of efficient collection for large dairy plants. To give an example, one of the cheese competitors only has one very big dairy plant in the north-eastern part of the Netherlands which gives this competitor the possibility of achieving economies of scale which outweigh the higher collection costs stemming from the fact that its farmers are located throughout the Member State.

66. Another important influence on the economies of scale at the level of procurement, which tends to reduce collection costs, is the size of the farms from which raw milk is sourced. As stated by the notifying parties "Depending of the cost structure of these buyers, transport over a longer distance may or may not be an economically attractive option.(...) For third parties, the trade-off between a stable source of supply through a membership relation and transport costs may be different. In particular purchasers of raw milk that aim for rapid growth may seek members and/or long term supply contracts with dairy farmers in a less concentrated area and that are more dispersed. These third parties, however, tend to achieve economies of scale by contracting mainly larger farmers" (33).

67. The market investigation confirmed that some competitors focus on larger farms. One such competitor indicated that it collects on average 640,000 kg of conventional raw milk per farm (34). By comparison, Friesland Foods collected on average [550,000-600,000]* kg per farm in 2007 and Campina procured [500,000-550,000]* kg per farm in 2007 (35).

6.3.2.2. The notifying parties' main competitors are active throughout the Netherlands and the notifying parties can decide to do the same

68. With regard to the procurement areas of competitors, the market investigation showed that they collect conventional raw milk in areas which extend, in general, beyond the respective working areas of Friesland Foods and Campina and cover a substantial part of the Netherlands. This questions the position of the notifying parties that the geographic market is limited to their working areas, since alternative raw milk processors appear to be active on a larger scale and, in any event, do not always restrict themselves to one of the notifying parties' working areas.

69. For example, one competitor sources conventional raw milk in the provinces of Friesland, Groningen and Gelderland (which are included within Friesland Foods' working area) and in the provinces of Utrecht, South Holland and North Brabant (which are included within Campina's working area), as well as in the province of Flevoland (where the notifying parties' activities overlap). This is also the case for a second competitor whose working area covers the provinces of Friesland, Flevoland, Drenthe and Overijssel (Friesland Foods), North Holland and South Holland (Campina) and Flevoland (both). Finally, since 1 January 2008, a third competitor sources raw milk in the provinces of Gelderland (Friesland Foods), Utrecht and North Brabant (Campina) (36).

70. Estimates by the notifying parties of their competitors' market shares in their respective working areas (see section 6.4.1.1.) indeed confirm that these players are active in the whole of the Dutch territory. According to these estimates, Bel-Leerdamer sources onethird of its raw milk requirements in Campina's working area and the remainder in Friesland Foods' working area. The same is true of DOC (one-third in Campina's area and two thirds in Friesland Foods' area).

71. Not only are competitors active in areas which extend beyond Friesland Foods' and Campina's working areas but the notifying parties themselves could decide to enlarge their own working areas. According to Article 20 of the Articles of Association of Campina, the geographic working area of Campina is defined by the board, which divides it into departments. Article 13 of the Articles of Association of Friesland Foods provides for a similar delineation of the working area of Friesland Foods. Thus, the current, limited, working areas only reflect decisions made by the notifying parties themselves- they are selfimposed.

72. There appears to be no significant obstacles to a modification of the current working territories as Friesland Foods and Campina were themselves created through subsequent mergers of smaller cooperatives active in neighbouring working areas over a couple of decades. These cooperatives regrouped the majority of dairy farmers within a small region. The current territorial divisions are a result of these mergers. Consequently, the current working areas of both notifying parties are the result of a historical process and could be easily modified by both notifying parties. Indeed, even if they consider that they have no interest whatsoever in enlarging their working areas, the notifying parties themselves admit that the Netherlands as a whole will be the working area (and the scope of the geographic market post merger) of the new cooperative (37). Hence there is no reason to limit the geographic market to these sub-national working areas.

6.3.2.3. Competitors do not necessarily take the pay-out price paid by the notifying parties as a benchmark

73. The market investigation also showed that competitors in the raw milk procurement market do not necessarily take the price paid out by Friesland Foods or Campina to their farmers as a benchmark to set their own purchase prices in the areas covered by those competitors. As the main competitors of the notifying parties collect raw milk throughout the Netherlands, conditions of competition in Friesland Foods' working area are not significantly different for these players than the conditions in Campina's working area. Cooperatives like DOC or Cono pay the same price to their farmers for raw milk, irrespective of the region where these farmers are located.

74. With regard to the determination of the raw milk price, competitors indicated during the market investigation that they primarily set their purchase prices with reference to the business results that they achieve in the downstream markets (38). When a yardstick system is retained, competitors rather take as a benchmark a weighted average of 3-5 competitors (including, but not limited to, the notifying parties) and add a mark-up to this Benchmark (39). One competitor even submitted that no price differentiation existed between each region for dairy companies in the Netherlands.

75. In their reply to the 6(1)(c) Decision, the notifying parties claim that they have submitted documentary evidence (particularly newspapers' articles (40) related to competitors Noorderlandmelk and Lyempf) of competitors in the procurement market who set their purchase prices for raw milk directly in relation to the pay-out prices applied by large cooperatives in their working areas (Friesland Foods in that case). The notifying parties acknowledge, however (41), that not all competitors act in the same way. This is indeed in line with the Commission's findings during the market investigation. Even if some small new entrants in the procurement market could have used Friesland Foods' or Campina's pay out price as a benchmark, this is not a general feature of the market. To give an example, the price paid by Cono is also used as a reference by one competitor.

6.3.2.4. In previous cases, the procurement market was considered national in scope

76. Finally, with regard to the decision by the Dutch competition authority (NMa) in which it was stated that the market for procurement of conventional raw milk was sub-national in scope, that assessment had specific features. The geographic area in which the target was active fell completely within the area in which the acquirer was active. Therefore, for the purpose of that particular decision, it could be argued that markets could be assessed at a regional level.

77. Previous Commission decisions provide examples where the market for the procurement of conventional raw milk has been considered national in scope. In Arla Foods/Express Dairies (42), the relevant geographic market for the procurement of raw milk was determined to be the United Kingdom. In Lactalis-Nestlé (JV), it was held that the relevant geographic market for procurement of raw milk was France (43). These Member States are larger than the Netherlands and although the competitive conditions of the domestic procurement markets were considered sufficiently different from neighbouring Member States, within France or the UK itself, transport and access infrastructure is comparable to that in the Netherlands.

6.3.3. Procurement of organic raw milk

78. With regard to the procurement of organic raw milk, the notifying parties submit that the geographic market is also sub-national (44). However, since in the case of Campina the working area covers the entire territory of the Netherlands, the notifying parties provided purchase figures at the national level (45). The notifying parties also submitted that transport costs are of less significance for organic raw milk than for conventional raw milk, given the important price difference between both kinds of milk. Furthermore, organic raw milk must be collected over longer distances due to the fact that there are relatively few organic farmers in the Netherlands who are not concentrated in just one area.

79. The market investigation confirmed the main features of the market for the procurement of organic raw milk described by the notifying parties. Therefore, it is concluded that the relevant geographic market for procurement of organic raw milk is national.

6.3.4. Conclusion on the relevant geographic market

80. Based on the above, it is concluded that the relevant geographic markets for the procurement of conventional raw milk and organic raw milk are national in scope and include the whole of the territory of the Netherlands.

6.4. Competitive Assessment

6.4.1. Procurement of conventional raw milk

6.4.1.1. The merger would lead to the creation of a strong buyer in the procurement of raw milk

81. Market shares of the notifying parties and their competitors in the market for procurement of conventional raw milk in 2007 are shown in the table below (46).

Table 6-1: Companies procuring raw milk in the Netherlands and their market positions

82. On the national market, the combined share of the new merged entity would be [70- 80]*% (47) with a substantial overlap (Campina: [20-30]*%, Friesland Foods: [40-50]*%) and competitors are much smaller than the new entity: DOC ([5-10]*%), Bel-Leerdamer ([5-10]*%), Vreugdenhill ([0-5]*%) and Cono ([0-5]*%).

83. Thus, it appears that on a national market the merged entity would hold a very high market share in the procurement of raw milk (48). Coupled with the fact that on the supplyside, farmers supplying raw milk are highly fragmented, this provides a strong indication that the merged entity would hold a dominant position in the procurement market for raw milk. Friesland Foods and Campina procure raw milk from more than 9 000 and 6 000 member-farmers respectively.

84. According to the notifying parties, Friesland Foods and Campina do not compete with each other to attract new members. They submit that in the last ten years, only one member has switched from Friesland Foods to Campina and not a single member has switched from Campina to Friesland Foods. The notifying parties further submit, in their reply to the 6(1)(c) Decision, that adding up the pre-merger national market shares of the notifying parties is a distortion of economic reality. For the notifying parties, the only difference which the proposed merger would bring about is that the working area of the merged cooperative will double in size compared to the working areas of the constituent cooperatives.

85. The notifying parties also submit that there are numerous alternative sources of demand for dairy farmers apart from them. They argue that a milk producer who decides to terminate his membership with the merged undertaking will always be able to find an alternative buyer for his milk. The notifying parties submit that there have been competitors who have, in the recent past, taken over significant volumes from previous members of the notifying parties. For example, DOC, a cheese manufacturer, increased its volume of conventional raw milk by over [60-70]*% in three years, from [...]* kt in 2005 to [...]* kt in 2008. Lyempf, a producer of milk powder, was able to attract [...]* kt of raw milk in one year. The notifying parties also mention Eko-Holland, a producer of organic dairy products and therefore sources only organic raw milk.

86. The notifying parties' view regarding the market position of the merged entity post-merger in the procurement market cannot be supported. First, with respect to the argument that adding up the notifying parties' pre-merger national shares to assess their market position post-merger distorts the economic reality post-merger, it is customary to assess the market position of the merged entity in the market post-merger by aggregating the purchasing volumes in that market of the notifying parties to that merger -to analogously assess a position on the selling side. Even the notifying parties agree that, post-merger, the merged entity would be operating a national market for the procurement of raw milk. In this context, it is therefore difficult to support their claim that aggregating market shares of the notifying parties would distort economic reality. Indeed, the notifying parties propose no alternative assessment of the market position of the merged entity in the Dutch market for the procurement of raw milk.

87. In their reply to the Statement of Objections (49), the notifying parties reiterated their claim that adding up the notifying parties' pre-merger market shares to assess their market position might be customary but is irrelevant in this case for the reasons already stated. They claimed that the merger did not decrease the number of outlets available for dairy farmers to sell their milk, but did not elaborate on the consequences of having a sole entity procuring [70- 80]*% of the raw milk in the Netherlands.

88. Second, with regard to the evolution of market positions over time and the claim that alternative buyers would have taken over a significant volume from the notifying parties, it is noted that the notifying parties' market share has been remarkably stable over the last few years. According to the notifying parties' own estimates, their combined market share was [70-80]*% in 2005, 2006 and 2007. For 2008, on the basis of the milk quota contracted this year, this market share shows a slight decline from [70-80]*% to [70-80]*%, which remains considerable and constitutes, in itself, evidence of a dominant position.

89. Third, as to switches from the notifying parties to other suppliers and the possibility for entry and expansion in the market for the procurement of raw milk, the market investigation did not find that competitors have taken over significant volumes from previous members of the notifying parties in the recent past. With regard to DOC, and according to the notifying parties' own estimates, it appears that its increase in volume of raw milk procured coincides with the exit of Nestlé of the market for the procurement of raw milk in 2005-2006. At that time, Nestlé took the strategic decision to stop production of milk powder in the Netherlands. Consequently, Nestlé sold the plants to Vreugdenhill and Hochwald Nederland and a large number of farmers switched to DOC. Therefore, it does not appear that DOC took over a significant volume of raw milk from the notifying parties, but rather that DOC took over volumes from other competitors who had already left the market.

90. The notifying parties claimed in the reply to the Statement of Objections that the assessment of the Commission on the coincidence between the increase in raw milk procurement by DOC and the market exit of Nestlé is "speculative" and "lacks factual foundations" (50). Since these elements have since been confirmed by DOC in minutes which have been communicated under a non-confidential version to the notifying parties after the Statement of Objections (51), the notifying parties' conclusion cannot be shared. Furthermore according to estimates made by the notifying parties in the reply to the Statement of Objections, DOC would have taken only over 15% of its additional volume from the notifying parties and 85% from other competitors.

91. With regard to Lyempf, it must be noted that Lyempf is a producer of milk powder and cream which has been buying raw milk directly from farmers since 1 January 2008. Until 31 December 2007, Lyempf purchased raw milk from, inter alia, Campina and Friesland Foods. In any event, the volume of raw milk contracted by Lyempf for 2008 (roughly [...]* kt, according to notifying parties' estimates) represents less than [0-5]*% of the total market of the Netherlands.

92. With regard to other competitors, their market shares have also remained stable over the last three years. Bel-Leerdamer had a market share of [5-10]*% in 2005 and it is estimated to hold a share of [5-10]*% in 2008. The same is true for Vreugdenhill ([0- 5]*% in 2005 and 2008), Cono ([0-5]*% in 2005 and 2008) and Rouveen ([0-5]*% in 205 and 2008). As claimed by the notifying parties, there have been new entrants in the market for the procurement of raw milk, such as Noorderlandmelk, Lyempf and Hochwald Nederland, but their market shares ([0-5]*% each).are negligible compared to the notifying parties

93. The notifying parties nevertheless argue that there are no important obstacles in accessing raw milk in the Netherlands and refer to chapters in the reply to the Statement of Objections that deal with fresh dairy products and cheese that show that access to raw milk has never been, and is not expected in the post-merger era to become, a problem for any player in the Dutch market.

94. However, these views cannot be accepted. In section 7 of this Decision, it is demonstrated that access to raw milk as an input material would pose a problem to competitors post-merger, especially smaller ones, as it would be difficult to convince members of cooperatives in particular to re-direct their supply with equal conditions as Campina-Friesland (52). Moreover, non co-operative producers would have to pay a risk premium compared to the notifying parties, as they would not be able to offer a longterm purchase guarantee.

95. Likewise, with respect to cheese, it has been underlined that the supply of milk required for the production of cheese would be locked into the notifying parties. It would therefore be impossible for smaller cheese producers to increase their production of Dutch-type cheese. Even if the market investigation has indicated that all Dutch cheese producers are planning to expand their current production capacity as a result of the expected increase of the existing milk quotas and possible abolition of it in 2015, these expansion plans, and the increase in demand of raw milk thereof, will not materialize within the next two years, the time frame normally taken into account for the purpose of merger control.

96. With respect to the stability of these market shares, the notifying parties underlined in the reply to the Statement of Objections (53) that an undertaking does not seek access to the raw milk market for the sole reason of buying a market position in raw milk but only, and to the extent necessary, in order to carry out its activities on the downstream markets. The fact that a company had a stable share in the procurement market for raw milk indicates only that it had not significantly increased its output in the downstream markets.

97. This statement clearly shows the link between a strong position in the downstream markets for fresh dairy products and cheese and the upstream market for raw milk. Since post-merger competitors in the cheese and fresh dairy markets would not be in a position to exert sufficient competitive pressure on the notifying parties in these downstream markets, they would encounter more difficulties in offering farmers price conditions which would match those proposed by the notifying parties. As will be further explained, these competitors would find it more difficult to secure supply as their farmers would be attracted by more favourable conditions (linked to market power downstream) offered by the notifying parties. This would therefore further strengthen the notifying parties' position upstream.

98. There are therefore strong indications that the merged entity would have a dominant position in the market for the procurement of raw milk. When assessing the impact of a concentration in an upstream market such as the procurement of raw milk, it is necessary to first analyse whether the increase of buyer power in upstream markets may significantly impede effective competition, as it appears to be the case here, by creating or strengthening a dominant position. It should be clarified whether the increased buying power of the notifying parties could have a detrimental effect on competition, by enabling the new entity to obtain lower prices from farmers by reducing its purchase of raw milk, which would in turn lead to lower output also in the downstream markets and thus harm consumer welfare.

99. In this respect, an alternative interpretation of the notifying parties claim concerning the aggregation of market shares is that despite the resulting post-merger overlap in the Dutch procurement market for raw milk, this would not lead to a significant impediment of effective competition, since there was little competition for sourcing from each other's farmers pre-merger.

100. With respect to the latter argument, limited competition between the notifying parties in the procurement of raw milk does not mean that they are not potential competitors, potentially even the closest competitors in the procurement market in the Netherlands. Whilst pre-merger, absence of competition for farmers between the notifying parties could result from a conscious strategy not to procure from each others farmers, the merger, through an irreversible change in market structure, entirely eliminates the possibility that this strategy might have been altered in the future.

101. Notwithstanding the above, the notifying parties also note that should the merged entity have a dominant position on the procurement market, this would in principle mean that the notifying parties could use their market power as buyers to reduce milk procurement prices. The notifying parties further explain that they would have no reason to reduce the pay-out prices paid to their members post-merger. Being dairy cooperatives, the primary aim of the notifying parties is to achieve long-term optimal pay-out prices for their members. According to the notifying parties, a reduction of the pay-out prices due to increased market power vis-à-vis their members would be unsustainable as it would lead to the dismissal of management by member-farmers.

102. The Commission agrees with the premise that ultimately, the merged entity, should it retain its cooperative structure, would be unable to sustainably pursue a policy which results in lower income for its members. There are no reasons to expect the merged entity would cease to be a cooperative in the foreseeable future. Furthermore, the cooperative would retain the current commitment to purchase the milk from its members post-merger. It is thus likely that the merged entity would have neither the ability nor the incentive to reduce the price paid for raw milk purchased from its members such that it would lead to a reduction in input purchases, consequently reducing output and increasing prices further downstream.

103. The merged entity would therefore hold a dominant position in the procurement of conventional raw milk. However, in the context of the present case, the strong market position of the merged entity would not, in itself, allow it to exert buyer power vis-à-vis member-as a result in particular of the cooperative structure of the merged entity

6.4.1.2. Risk of foreclosure of rivals in the downstream dairy markets

104. As mentioned above, raw milk is an input which is used to manufacture consumer and industrial dairy products which are sold downstream to retailers, OOH ("Out of home") wholesalers and industrials customers. It is thus necessary to also assess whether competition in the downstream markets could be adversely affected if, in particular, the merged entity would have the ability and incentive to leverage its strong position in the procurement market for raw milk to foreclose existing or potential downstream rivals (54), by limiting or raising the costs of their access to raw milk.

105. The market investigation reveals that there is a risk that competitors in the Netherlands would be indirectly foreclosed in their access to raw milk (55). The following elements are relevant in that respect.

6.4.1.2.1. The merged entity will have market power on downstream markets

106. It is concluded that the merger would significantly impede competition in some national downstream markets for dairy products processed from raw milk, notably cheese and fresh dairy products.

107. With regard to cheese (50% of the raw milk collected is used in the production of cheese), it was found (see section 8) that the new entity would hold a very high market share (around [60-70]*%) in the Dutch market for Dutch-type cheese sold to supermarkets, which would not have sufficient alternatives to the notifying parties. It was concluded that the new entity would also control the supply of cheese to cheese wholesalers in the Netherlands.

108. With regard to milk and milk products (13% of the raw milk collected), it was found (see section 7) that the merger would significantly impede effective competition by creating or strengthening a dominant position on the Dutch markets for fresh dairy products. It was found that the merged entity would have high market shares and that the transaction would bring together the closest competitors and the only companies to offer the full range of fresh dairy products in sufficient volumes and quality as brands or private label, thereby removing the key driver for effective competition in the market.

109. As a result, the merged entity will become an unavoidable trading partner for a significant part of the demand for cheese and fresh dairy products in the Netherlands and would be able to exercise market power on these markets. This reduction of effective competition in these markets will enable the notifying parties to charge higher prices, and secure higher margins, in the fresh dairy products and cheese markets. This would lead to an increase of the overall profits of the merged entity.

110. Importantly, the ability to exercise market power in the downstream market also implies that the merged entity can generally expect to benefit from any increased barriers to entry into any of those downstream markets. Lack of access to raw milk can be one significant barrier which deters entry across a number of downstream dairy markets, as described in sections 7 and 8.

6.4.1.2.2. The increase in profits linked to downstream market power will be redistributed to farmers, by means of higher pay-out prices

111. As mentioned above, the milk price that would be paid to farmers by the new entity consists of two different elements : the guaranteed milk price which corresponds to an average milk price paid by dairy companies in North-West Europe and is independent from the financial performances of the new entity ; and the performance payment, which is directly linked to the performance of the new entity as it consists of a part of the net profits of the company (25%), after deduction of the amount that will be added to the reserves of the company. 15% of these reserves are also indirectly distributed to farmers as they are added to the reserves by means of the issuance of new member bonds to the members, who can cash these member bonds upon terminating their membership.

112. Profits raised in the downstream markets as a result of the increased market power would be redistributed to farmers who supply raw milk to the merged entity. The market investigation also showed that some competitors expect the merger to result in higher prices paid to farmers, given the increase in business and the fact that the essence of the cooperative is to maximise the valorisation of the milk for its members who own the cooperative.

113. For example, one competitor indicated that the merged entity would increase the prices by 2-5%, as "the merger has to prove themselves against their members" (56). Another competitor stated that "the main target of the new company is paying the farmers a good milk price. Besides that the transaction will result in better business results. This could be beneficial for the milk price paid" (57). Likewise, a third competitor expects a 5% price increase for raw milk

6.4.1.2.3. By paying higher prices for raw milk, the notifying parties will be in a position to attract new members

114. If the notifying parties pay higher prices for raw milk to farmers than their competitors while offering the same kind of cooperative solution to which Dutch farmers are attached (stability of income, commitment to collect all the raw milk produced, longstanding relationships), it is likely that new farmers would apply to join the new entity.

115. The market investigation indeed showed that some competitors expect the new entity to enlarge its farmers' base, thanks to higher milk prices. One competitor expects the new entity to attract more farmers since "Farmers are very sensible for money and security. Big companies have more foundation than small companies" (58). A second competitor believes also that new farmers will join the new entity because of this increase of the payout prices. This is also the opinion of two dairy companies producing Dutch-type cheese and other dairy products and currently not active in the Netherlands, which expressed their concern over the creation of a monopoly in the upstream market for raw milk in the Netherlands (59).

116. A Dutch competitor (60) expressed the same concern - that the new FrieslandCampina cooperative would have such a strong position in the Netherlands for consumer milk products that the price for these products would be relatively high. In that case, this competitor claimed that it would have a positive influence on the merged entity's position with regard to the milk price paid to farmers and the impact of the merger on the market for procurement of raw milk would be negative.

117. The notifying parties claim that as a cooperative, the new entity would be obliged to pay the same milk price to all of its members and, therefore, it cannot target particular farmers and offer them a higher price in order to induce them to switch to the new entity. If prices to some member farmers are increased, prices paid to all member farmers have to be increased.

118. However, it is not claimed that the new entity will apply price discrimination towards new members but rather that additional profits raised in the downstream markets through its increase in market power will enable it to pay higher prices to all memberfarmers. Currently, Friesland Foods and Campina together hold almost [70-80]*% of the procurement of raw milk in the relevant geographic market. Any increase in the pay-out price for raw milk would make members of the post-merger entity less likely to switch to other purchasers of raw milk. As a result, it is also likely that it will attract new farmers from third parties. It would therefore further reinforce the merged entity's position on the market for procurement of raw milk.

6.4.1.2.4. The merger will increase barriers to entry or to expansion in the downstream dairy markets by limiting access to independent sources of raw milk

119. With regard to the upstream markets for procurement of raw milk, it has been demonstrated that the merger will bring the main purchasers of raw milk in the Netherlands together and that its market power downstream would enable it to payer higher prices to farmers and therefore attract farmers from third parties.

120. With regard to the downstream markets, the market investigation showed that alternative suppliers on the Dutch fresh dairy product and cheese markets would need access to Dutch raw milk on a large scale. The market investigation conducted by the Commission in the cheese market showed that Dutch customers have a clear preference for cheese produced in the Netherlands with raw milk originating in the Netherlands. The same is true for fresh dairy products where the market investigation showed that access to Dutch raw milk seems to be a necessary condition to be able to increase supply, as explained below.

121. In a situation where the merged entity would have the possibility of increasing its already very large farmers' base, these competitors would find it more difficult to gain access to independent sources of Dutch raw milk. It would therefore further reduce their ability to compete effectively in their respective Dutch downstream dairy markets by raising barriers to entry or expansion and strengthen the market power of the notifying parties in these markets.

122. During the market investigation, clarification was sought regarding the minimum volume of Dutch raw milk necessary to successfully enter the Dutch market for fresh milk or Dutch-type cheese. To take an example, a Dutch competitor indicated that in order to enter the Dutch market for fresh milk in a competitive manner, the minimum volume of Dutch raw milk that a new entrant would have to source would be 150-250 million kg. For Dutch-type cheese, a new entrant would have to source at least 250-350 million kg (61) for cheese and a comparable volume for fresh milk (150-250 million kg). These competitors have expressed concerns with regard to the possibilities of entry into the Dutch dairy markets in this specific market context.

123. The quantities of raw milk that competitors would have to collect in order to successfully enter the Dutch dairy markets are larger than the additional volumes that the notifying parties' competitors have been able to secure in the last three years, according to notifying parties' estimates (62). Contrary to what the notifying parties claim in a separate submission (63), it does not appear that the availability of raw milk to third parties is evidenced by the development of volumes of raw milk supplied to third parties. With the exception of DOC (and for the specific reasons already mentioned), their shares have remained stable in the last three years.

124. During the second phase investigation, the notifying parties submitted a memorandum and recently published articles with regard to expansion plans from competitors which would allegedly show that competitors see opportunities arising from the contemplated merger (64) (by expanding their production) and that both expansion of production as well as entry into the production process is easy (65).

125. Some of these articles are related to expansion plans of producers of organic dairy products such as Eko-Holland which procure organic raw milk. They are therefore not relevant for the assessment of the impact of the merger on the market for procurement of conventional raw milk. With regard to expansion plans of Katshaar, a small Dutch player which recently secured a private label contract of fresh milk with retailer Albert Heijn, this company plans to grow up to [...]* kt of milk, which is slightly more than [0-5]* % of the market (which is admitted by the notifying parties in the reply to the Statement of Objections) and considerably less than the notifying parties control or what is necessary to effectively enter the main downstream markets. The same is valid for Globe Milk, a company which is the result of cooperation between Vrebamel and a milk processor and which will process the milk collected by Vrebamel ([...]* million kg of raw milk collected in 2007).

126. The article regarding Cono, sent during the second phase investigation, does not refer to any expansion plans of this company but rather to the fact that Cono's farmers produce raw milk to support Unilever's expansion plans in Europe with respect to the Ben and Jerry's ice cream. Indeed, in the reply to the Statement of Objections (paragraph36), the notifying parties refer to an earlier article with respect to the new cheese plant of Cono (66). However, the increase in processing capacity of raw milk of Cono will be roughly [...]* million kg, which accounts for less than [0-5]* % of the volume procured by the notifying parties only in the Netherlands.

127. In their reply to the Statement of Objections (67), the notifying parties indicated that the planned abolishment of the milk quota regime in 2015 will have a tremendous impact on the Dutch dairy industry. In a report prepared by two scholars in the dairy industry and submitted by the notifying parties, it is stated that "Raw milk production in the Netherlands is even predicted to increase with about 20%" (68), which as the notifying parties claim, amounts to more than 2 000 million kg of raw milk available in the Netherlands in the next 5 to 6 years (more precisely 2 220). Other sources mention a potential increase of raw milk production of 16% up to 2015 in the Netherlands.

128. However significant these figures may seem, they should be taken in the context of the Dutch raw milk market. Together, the notifying parties hold, on the basis of the 2008 quotas, a market share of [70-80]*% in the market for the procurement of raw milk in the Netherlands. In the most conservative approach, this share will remain stable in the coming years. This means that up to [20-30]* % of this increase in milk production of 2220 million kg will be available for competitors -roughly [...]* million kg in the next 6 years or [...]* million kg per year. Friesland Foods, the smallest party of fresh milk, sells approximately [...]* million kg of fresh milk per year in the Netherlands. Therefore, the annual volume which would be " available " (69) in the Dutch raw milk market would be far from sufficient to cover the overlap in the sole market for fresh milk and to exert on this market competitive pressure which would be comparable as the one exerted by Friesland Foods on Campina pre-merger. Even if raw milk production would increase by 20% between now and 2015, this would not be sufficient to restore effective competition conditions on this market.

129. Finally, the notifying parties argued (70) that the Commission's reasoning was highly circular. Due to the market power that the merged entity will obtain in certain specific downstream markets, the merged entity will be able to increase pay-out price to farmers, which will in turn increase barriers to entry and expansion for competitors in those downstream markets, which in turn will allow the notifying parties to further increase their profits in those downstream markets, which will lead to higher pay-out prices and so on, until the merged entity will obtain a monopoly position in the procurement of raw milk and downstream markets for Dutch-type cheese and fresh basic dairy products.

130. While the Statement of Objections did not suggest that the effects of the increased market power in the downstream markets on the upstream market would eventually result in a monopoly, it is clear from this section that the increase of market power in the downstream markets resulting from the proposed concentration would strengthen the position of the merged entity on the upstream market, which would in turn aggravate the negative effects on the downstream markets.

6.4.1.3. Conclusion of the competitive assessment: procurement of conventional raw milk

131. In conclusion, the merger would bring the two main purchasers of raw milk in the Netherlands together, which would control roughly [70-80]*% of the market. The market power that they would have on downstream markets would enable them to raise additional profits and therefore pay higher prices to farmers. Consequently, the merged entity would be in a position to attract more farmers and maintain and/or strengthen its farmers' base. This situation would increase barriers to entry and/or to expansion on the primary downstream dairy markets where Dutch raw milk is needed to compete effectively.

132. For the reasons set out above, it is concluded that the notified concentration is likely to significantly impede effective competition in the market for the procurement of raw milk in the Netherlands, which is a substantial part of the common market by limiting access to independent sources of raw milk.

6.4.2. Organic raw milk

133. When looking at a national market for procurement of organic raw milk, the new entity would hold a market share of [40-50]*% in volume in 2007 (Campina: [30-40]*%, Friesland Foods: [10-20]*%). Main competitors are Eko-Holland ([10-20]*%), Rouveen ([10-20]*%) and Op Boerderij ([5-10]*%). Unlike conventional raw milk, the combined market share of the notifying parties has significantly decreased since 2005 (when it was [50-60]*%) and it is expected to decrease again in 2008 since, on the basis of the quotas of organic raw milk contracted by the notifying parties for 2008, the market share of the new entity in 2008 would be [30-40]*%. The award of a contract for organic raw milk by one important retailer to competitor Vecozuivel at the end of 2007 has not yet impacted Vecozuivel's share in procurement of organic raw milk. Consequently, it can be expected that the notifying parties' market share will continue to decline.

134. As for conventional raw milk, organic raw milk is a vital input for the production of organic downstream dairy products where the notifying parties are active. As discussed in section 7-3, the transaction is not likely to lead to a significant impediment of effective competition for organic fresh basic dairy products (organic fresh milk, organic fresh buttermilk, organic plain yoghurt and organic custard) in the Netherlands. This is due, among other reasons, to the decline of the notifying parties' share in organic downstream markets, the volatility of the market, the possibility for customers to switch to existing suppliers as well as the possibility of entry by new suppliers.

135. Therefore, in the light of the situation in organic downstream dairy markets, the continuous decline of the notifying parties' market share in upstream procurement markets and the fact that competitors confirmed that they do not experience problems with access to organic raw milk in the Netherlands (see section 7-3), it can be concluded that the impact of the merger on the procurement of organic raw milk would be limited.

136. For these reasons, it is considered that the transaction is not likely to lead to a significant impediment of effective competition in the market for procurement of organic raw milk.

6.5. Overall conclusion

137. In the light of the above, it is concluded that the transaction is likely to lead to a significant impediment of effective competition in the market for the procurement of conventional raw milk. It is not likely to lead to a significant impediment of effective competition in the market for procurement of organic raw milk.

7. BASIC DAIRY PRODUCTS

138. Campina and Friesland Foods are both active in the production and supply of basic dairy products comprising milk, buttermilk, plain yoghurt and custard. According to the notifying parties basic dairy products can be categorised into fresh and long-life products. A distinction can also be made based on whether organic or non-organic raw milk is used as raw material.

139. The market investigation confirmed the distinction between fresh and long-life dairy products. It revealed that despite improvements in the taste for long-life basic dairy products, customers still perceive long-life basic dairy products as different from fresh basic dairy products because of taste, shelf-life, storage possibilities and placement in the supermarket (71).

140. Both fresh basic dairy customers and competitors confirmed the submission of the notifying parties that fresh basic dairy products based on organic raw milk should be distinguished from those produced from conventional raw milk. Respondents in the market investigation (72) argued that there would only be limited substitutability between conventional and organic products because (i) the price for organic products is significantly higher, (ii) organic products are targeting consumers with specific preferences (protection of nature, animals and food production) and (iii) the production process is different as particular conditions have to be fulfilled. (73)

141. Therefore, this Decision distinguishes "fresh basic dairy" and "long-life basic dairy" as well as "non-organic" and "organic basic dairy" products (74). Where no specific reference is made, it is assumed that non-organic dairy is referred to.

7.1. Fresh basic dairy products

7.1.1. Relevant Product Markets

142. The notifying parties define fresh basic dairy products as fresh milk, fresh buttermilk, plain yoghurt (in gable top packaging) and custard. (75)

143. Fresh milk is liquid milk with a standardised fat content, which is produced from raw milk that is first skimmed and subsequently pasteurised at 72ºC for 15 seconds. Pasteurised fresh milk is still perishable and can be held for 8-9 days if stored and transported refrigerated. By adding a bactofugation process - a specific centrifuge separating micro-organisms from the milk - in combination with aseptic filling equipment this shelf life can be increased to 12 days. An extension to 22 days can be achieved with microfiltration - a separation technology that passes the milk over a membrane filter at high speed - eliminating almost all spores (between 99.5-100%).

144. The shelf life is extended to 30 days if the milk is produced with Thermic Extended Shelf Life ("T-ESL"), which slightly affects the taste as a result of heating it at 128ºC. According to the notifying parties the great majority of German producers use TESL technology, while in the Netherlands bactofugation is prevailing. (76)

145. Fresh buttermilk is a liquid fermented dairy product produced from raw milk with a characteristic sour taste. It is produced by adding lactic acid bacteria to pasteurised and partially skimmed milk. Fresh buttermilk is perishable and can be held for 12-15 days if stored and transported refrigerated.

146. Plain yoghurt is a fermented product also produced from raw milk, which has been partially skimmed and pasteurised and fermented by adding special yoghurt cultures. It has a less sour taste than buttermilk and higher viscosity with a shelf life of 15-20 days if stored and transported refrigerated.

147. Custard or "vla" is a sweet dairy product produced from raw milk, which is first partially skimmed and pasteurised. Subsequently, aromas, sugar and thickening agents are added which provide taste and increase viscosity. It is a spoonable product with a shelf life of 16-25 days if stored and transported refrigerated.

148. Each of these products (fresh milk, fresh buttermilk, plain yoghurt and custard) produced by the notifying parties is packed in carton gable top packaging ranging from 0.5 to 1.5 litres, covering about [80-90]*% of production.

7.1.1.1. Fresh milk, fresh buttermilk, plain yoghurt and custard each form a single product market

7.1.1.1.1. Product market definition proposed by the notifying parties

149. Friesland Foods and Campina submit in their notification that fresh milk, fresh buttermilk, plain yoghurt (filled in gable top) and custard form one single product market because "suppliers are able to switch production to the relevant products and market them in the short term without incurring significant additional costs or risks in response to small and permanent changes in relative prices". (77) In particular they explain (78) that:

(a) The production process for these products is largely the same. The processing (standardising, skimming and pasteurizing) is to a large extent the same and the packaging equipment is identical for milk and buttermilk on the one hand and for yoghurt and custard on the other hand.

(b) In addition, a switch on one production line from one product to a different one is possible at low cost (below EUR 1 million), a short lead time (maximum 3 months) and a downtime of one week. Converting a packaging line would be possible with an investment of less than EUR 100 000 and a downtime of one week.

(c) This switching opportunity is supported by the contracting practice of retailers who tender their supply once a year and grant sufficient lead time to allow companies to adjust their production facilities in order to participate in the tender.

150. The notifying parties did not investigate the possibility of demand-side substitution. However, based on their experience the notifying parties indicated that demand-side substitutability between the four fresh basic dairy products would be limited. In particular, fresh custard would have different consumption patterns than the other products and customers would rather consider it as a dessert. Moreover, while some customers might be willing to substitute buttermilk with milk, not all consumers of fresh milk appreciate the sour taste of buttermilk. (79) However, according to the notifying parties, the lack of demand-side substitutability would not matter for the definition of a single relevant product market given the existence of supply-side substitutability.

151. In their reply to the Article 6(1)(c) Decision, the notifying parties re-iterated their claim that supply-side substitutability would prevail despite switching costs and lead time. In their response to the Statement of Objections, no further arguments have been put forward. (80)

7.1.1.1.2. Assessment of the Commission

152. The market investigation has not confirmed the market definition proposed by the notifying parties whereby fresh milk, fresh buttermilk, fresh yoghurt and custard form part of the same relevant product market.

153. The relevant product market is typically defined as all those products and/or services which are regarded as interchangeable or substitutable by the consumer. The Commission Notice on the definition of the relevant market for the purposes of Community competition law (81) ("Relevant Market Notice") attributes a prominent role to demand substitutability in the definition of the relevant product market.

Demand-side substitution

154. With respect to the demand-side, the majority of customers agreed with the notifying parties' submission that plain yoghurt in gable top packing should be viewed as distinct from plain yoghurt in portion packs or yoghurt with additions (flavour and/or fruit) (82) because of different consumption habits and significant price differences.

155. Respondents also confirmed (83) the lack of or limited demand-side substitutability between fresh custard, fresh milk, plain yoghurt and fresh buttermilk: no respondent grouped these four products together; very few respondents saw the possibility of demand-side substitution between fresh milk and yoghurt or fresh buttermilk and yoghurt. Respondents indicated that consumption pattern, the taste (sour) and the thickness would differ among those products.

156. Only a minority of replies received indicated that fresh milk and fresh buttermilk could be exchangeable. (84)

157. It is therefore concluded that there is no demand-side substitution between fresh milk, fresh buttermilk, plain yoghurt and custard. (85)

Supply-side substitution

158. Immediate and costless entry may render a small but permanent increase in price unprofitable thus leading to broader markets or undermining the relevance of market shares as indicative of market power. According to the Relevant Market Notice, supplyside substitution may also be taken into account if suppliers are able "to switch production to the relevant products and market them in the short term." (86) Thus, supplyside substitution involves entry at a low cost and without incurring irreversible investments and implies the ability to substitute production of one product for another at short notice in response to relative price variations (87). Only if these conditions are met will supply-side substitutability have an impact in terms of effectiveness and immediacy equivalent to the demand substitution effect.

159. Supply-side substitution translates into market aggregation, that is, a broadening of market boundaries to include a larger group of products or geographical area. However, the Relevant Market Notice also points out that aggregation only makes sense when production substitution among a group of products is found to be technologically feasible and economically viable for most, if not all, firms selling one or more of those products. (88)

160. In light of the market investigation, therefore, the notifying parties' view that fresh milk, fresh buttermilk, basic yoghurt and custard are supply-side substitutes for the purposes of relevant product market definition cannot be supported.

161. Firstly, in 2007, only one out of 17 competitors who replied to the market investigation was supplying all the four fresh basic dairy products to customers. The majority of respondents focussed on two products usually produced on different production lines and only a few supplied three categories, mainly fresh milk, fresh buttermilk and either yoghurt or custard. (89)

162. Secondly, a majority of competitors replied (90) that they are either not able to switch between the different fresh basic products (91) or indicated that such a switch would only be possible after additional investment because the processing and packaging lines would be different depending on the product. For example, one competitor argued that investments for a pre-mix tank to be able to switch from yoghurt to custard would be around EUR 300 000. (92) It follows that the requirement that supply-side substitution is nearly universal is not met in this case.

163. Moreover, while some replies seem to confirm the switching cost estimate put forward by the notifying parties in their submission, these costs assume that there would be enough space in the building currently being used and that the filling lines would be suitable for all kind of products. In the Netherlands the standard packaging format is the 1 litre gable top, but this is generally not the case for other neighbouring Member States where many suppliers currently do not supply in such a format (93). Furthermore, some respondents indicated that the addition of a processing or filling line would require a new building and significant investment costs and take up to 1-1.5 years. (94) These factors suggest that for a significant number of competitors supply-side substitution among fresh dairy products is not instantaneous or without cost.

164. Finally, even if an immediate switch was technically feasible, some competitors argued that differences in product composition would remain between German and Dutch yoghurt in particular. They also pointed to the lack of knowledge about the composition of custard. (95)

165. Accordingly, even if the notifying parties were able to produce the four products in question using the same production equipment, it is not the case that most competing suppliers of any of such products have the flexibility to reallocate production and sales interchangeably and without cost to other fresh dairy products. Indeed, even the notifying parties concede that adding a new product into an existing production site implies additional costs and a lead time of at least three months.

166. Hence, it can be concluded that the conditions laid down in the Relevant Market Notice that supply-side substitution is to be immediate and effective are not fulfilled. In particular, it is not the case that most, if not all, suppliers would be able to switch production to the relevant products and market them in the short term without incurring significant costs or risks. This is even more the case for branded products, where there are costs and lead times (in terms of advertising, product testing and distribution) before products can actually be sold. (96)

167. Given the lack of substitution by customers and the lack of supply-side substitutability, it is concluded that fresh milk, fresh buttermilk, plain yoghurt and custard are separate relevant product markets. While fresh milk, fresh buttermilk and plain yoghurt are treated in this section, custard will be discussed in section 11 on fresh dairy desserts.

7.1.1.2. Private label and supplier brands belong to the same relevant upstream market

7.1.1.2.1. Product market definition proposed by the notifying parties

168. The notifying parties agree that a vertical segmentation of a particular product market for consumer products can be made between an upstream market for the sourcing of products by retailers and OOH wholesalers serving hotels, restaurant, catering services and filling stations) and a downstream market for the sale by retailers to consumers, for two reasons.

169. This approach recognises that on the retail-to-consumer level, private label products compete with (producer) branded products, therefore the market share of private label products can be attributed to the retailers that own the private label brands at the downstream level.

170. It also recognises that the competitive conditions on the markets on which retailers and OOH wholesalers source their products can be fundamentally different from the competitive conditions on the retailer-to-consumer markets. (97) The distinction between private label and branded products at the upstream level must be carefully analysed.

171. The notifying parties agree (98) that the procurement of private label products and the procurement of branded products by retailers (and OOH wholesalers) might constitute two neighbouring, but separate markets, in the event that competitive conditions under which retailers source private label products and branded products are fundamentally different. This may be the case in situations where specific branded products are perceived as "must carry" products by retailers, in the sense that significant profits might be lost in the event of delisting such branded products.

172. In the context of the market for fresh basic dairy products the notifying parties claim that both Campina and Friesland Foods are under constant pressure from private label products. This is exemplified by the consistent decline of branded products over the years and the possibility that retailers carry out delistings of branded products or retaliation in the area of private label. (99)

173. Thus according to the notifying parties, private label and branded products would also belong to the same product market at the upstream level.

7.1.1.2.2. Assessment of the Commission

174. All fresh basic dairy producers sell their products to retailers and/or OOH wholesalers which, in turn, sell these products to final consumers. (100) Therefore, there are two stages in the supply chain: the upstream level, where fresh basic dairy products are produced and supplied to retailers/OOH wholesalers and the downstream level of supply to consumers. Campina and Friesland Foods are only active on the production and supply upstream (to retailers and/or OOH wholesalers).

175. In a recent consumer goods case (101), the Commission distinguished the upstream level where retailers source their products from a downstream level where the products are sold to the final customer. The same approach has been followed for the markets for fresh milk, fresh buttermilk and plain yoghurt.

176. Fresh basic dairy products are available in two broad categories: brands owned by the dairy manufacturer and private label products which are marketed by the retailer (102). Both Campina and Friesland Foods produce both branded and private label fresh basic dairy products.

177. Customers and competitors (103) explained in the market investigation that the procurement of branded products follows a different procedure than the one for private label products. In the case of branded products, suppliers and retailers agree in bilateral negotiation on the gross price, discounts, listing fees, and promotions. For private label products, usually a tender procedure is employed, followed by a selection of preferred suppliers which agree with the retailer on a net price in final negotiations.

178. Whether private labels and supplier brands belong to the same product market upstream also depends on factors other than the procurement process.

179. In particular, the following must be assessed:

(a) whether both types of brands, in general, compete closely with each other from the perspective of the end-customer; and

(b) the extent to which upstream suppliers of private label and/or brands as well as the purchasing retailers, take into account in their upstream negotiations the competitive pressure that private labels and supplier brands mutually exert on each other at consumer level.

180. Demand from retailers for a product is derived by the purchases and revealed preferences of end consumers. Thus, retailers will determine the volumes of a product to source for private label as well as their reservation price for it by taking into account the expected profitability of sales of this product under private labels relative to sales of supplier brands. This profitability in turn depends on the pattern of substitution between brands at the level of the end customer.

181. Moreover, depending on the relative importance of private label sales, suppliers cannot ignore the competitive pressure that private label exerts on their brands. This is especially the case when the suppliers of branded products are also the suppliers of private label products and both compete in the downstream market. In situations where the bulk of both the private label and the branded products are supplied by the same firms, it can thus be presumed that these firms would take into account, when supplying retailers, the substitutability of private label and supplier brands among end customers. For fresh milk, fresh buttermilk and plain yoghurt in the Netherlands Campina and Friesland Foods both offer well-known brands and private label products representing more than 60% of the respective market (104).

182. For fresh milk, fresh buttermilk and plain yoghurt there is significant evidence suggesting that private label and supplier brands compete in the downstream market and exert on each other a competitive constraint sufficiently relevant for the purposes of market delineation.

183. In their responses to the market investigation the majority of customers and competitors confirmed that private label and branded products compete at the retail level and that the quality of private label products is to a large extent similar to branded products (105).

184. In addition, on the downstream markets for fresh milk, fresh buttermilk and plain yoghurt the share of private label products in 2007 is at least 40% in value terms and has increased slightly over the past few years: fresh milk 63%, fresh buttermilk 40.5% and plain yoghurt 54.7 %. (106) This can also be observed in the IRI market-level data, provided by the notifying parties. According to IRI data, supplier brands have lost some market share in the recent past to private labels (see Figure 7-1 below). IRI data needs to be interpreted with some caution since it excludes discounters which largely source private label products and account for around 12% of the market in value terms for 2007, thereby overestimating the share of branded products (see also Table 7-9).

Figure 7-1: Evolution of Value Sales by Brand Category - Fresh Milk Source: IRI DATA

Figure 7-2: Evolution of Value Sales by Brand Category - Fresh Buttermilk Source: IRI DATA

Figure 7-3: Evolution of Value Sales by Brand Category - Plain Yoghurt. Source: IRI DATA

185. Furthermore, the margins submitted (107) by Campina for selected fresh milk products over the period 2005-2007 indicate that although Campina is able to achieve higher margins for its branded products than the private label products it sells to retailers, the difference in margins has declined significantly over the last two years, indicating competitive pressure on the branded segment.

186. Other features of the products also suggest that private labels and supplier brands compete within the same market. First, no differentiation can be observed with respect to packaging formats and size in fresh milk, fresh buttermilk and plain yoghurt. Branded products do not distinguish themselves by more expensive packaging like PET-glass bottles or deviating sizes. Campina and Friesland Foods fresh milk brands are sold primarily in gable top format in containers of 1, 1.5 or 2 litres (between [90-100]*% of total volume), with 1 litre accounting for around [50-60]*% in both cases (108). A similar pattern can be observed for private label brands in the aggregate as shown in the graph below.

Figure 7-4: Package Sizes by Brand Category - Fresh Milk.

Source: IRI DATA

187. In fresh buttermilk as well as in plain yoghurt, the vast majority of products are also sold in the standard 1 litre gable top packing. Only Albert Heijn has some sales in the 1.5 litre format for buttermilk, while in plain yoghurt half litre packaging is sold in addition to the dominant 1 litre format (See graphs below).

Figure 7-5: Package Sizes by Brand Category - Fresh Buttermilk Source: IRI DATA

Figure 7-6: Package Sizes by Brand Category - Plain Yoghurt. Source: IRI.

188. Finally there is almost no differentiation between supplier brands and private labels across segments, with 95% of all sales in the "regular" segment and only limited sales in the "weight management" segment (109) for both brand categories.

189. This broad analysis of IRI market-level data suggests that there is a sufficient degree of competitive interaction between supplier brands and private labels for fresh milk, fresh buttermilk and plain yoghurt for both to be considered as belonging to the same relevant product market at retail level.

190. Upstream suppliers of private label and the purchasing retailers take into account in their upstream negotiations the competitive pressure that private labels and supplier brands mutually exert on each other. In this segment it is significant that private labels account for a significant proportion of overall sales - at least 50.5% -, which, furthermore, has been growing over time. Given the relative importance of private label sales (equal or, in the case of fresh milk, close to double that of supplier brands), suppliers cannot be expected to ignore the competitive pressure that private labels exert on their brands. This is especially the case when the suppliers of branded products are also the suppliers of private label products and they both compete in the downstream market. In the present case, the bulk of both the private label and the branded products are supplied by the same firms, namely the notifying parties. It can thus be presumed that the notifying parties would take into account the substitutability of private label and supplier brands among end customers in supplying retailers (110), since this will affect the sales and relative margins they can expect to achieve in their negotiations vis-à-vis retailers.

191. Supermarket chains that offer private label products would also consider the competitive interaction downstream between private label and branded products. The major retailers in the Netherlands have two private labels in addition to the premium brands offered by Campina and Friesland Foods. For example, in the case of fresh milk, IRI data shows between 2004 and 2007 that Albert Heijn sold a high-level brand, coded "ABJ" at an average of EUR 0.66/l and also a low-level brand, coded "DNSH" at slightly below EUR 0.5/l a price difference of around 30%. The same is true for Laurus and Superunie as shown in Table 7-1 below concerning fresh milk. At the same time the brands of Campina and Friesland Foods fluctuated mildly during the same period at comparable prices between EUR 0.68 and EUR 0.70/l (see Figure 7-13 in section 7.1.3) (111). The existence of a high and low-level private label indicates that retailers consider and condition their private labels in relation to supplier brands. (112)

Table 7-1: Average Weighted Prices by Brand - Fresh Milk. Source: IRI.

192. For fresh buttermilk and plain yoghurt the picture is slightly different, as can be seen in the tables below, since not all retailers have two private label brands in addition to the brands of the notifying parties. Albert Heijn as well as Superunie uses two private label brands (high level and low level), while Laurus and Schuitema have only one positioned at the lower end.

Table 7-2: Average Weighted Prices by Brand - Fresh Buttermilk. Source: IRI.

Table 7-3: Average Weighted Prices by Brand - Plain Yoghurt. Source: IRI.

193. In fact, supermarket chains appear to pursue different strategies with respect to the positioning of their private labels vis-à-vis branded suppliers (see Table 7-4). For example, Albert Heijn only promotes its own private labels and only sells negligible volumes of supplier branded fresh milk (113). In contrast, Laurus primarily carries Campina and its own private labels with shares of sales around [50-60]*% and [40-50]*% respectively (with only a [5-10]* of Friesland Foods). Schuitema sells both Campina and Friesland Foods brands but the majority of its sales are private label (around 65%). Finally purchasing group Superunie sells slightly below 40% of private label, around [30-40]*% Campina, and [10-20]*% Friesland Foods. Superunie is also the only chain that carries Inexco (around 10%).

Table 7-4: Value Market Shares by Supermarket Chain - Fresh Milk. Source: IRI.

194. For fresh buttermilk, in addition to its own private labels [80-90]*%), Albert Heijn only offers Campina ([10-20]* %). In contrast, Laurus relies on both brands (Campina [40-50]*%, Friesland Foods [20-30]*%) and achieves one third of its sales through its own private label. Schuitema sells both Campina and Friesland Foods brands and 40% of its sales are private label. Finally purchasing group Superunie sells 35% of private label, around [30-40]*% Campina, and [20-30]*% Friesland Foods.

Table 7-5: Value Market Shares by Supermarket Chain - Fresh Buttermilk. Source: IRI.

195. For plain yoghurt, in addition to its own private labels [70-80]*%), Albert Heijn offers Campina ([5-10]*%). In contrast, Laurus relies on both brands (Campina [40- 50]*%, Friesland Foods [10-20]*%) and achieves one third of its sales through its own private label. Schuitema sells both Campina and Friesland Foods brands and 35% of its sales are private label. Finally purchasing group Superunie sells 35% of private label, around [30-40]*% Campina, and [20-30]*% Friesland Foods. Contrary to the other segments, fringe suppliers achieve higher shares on average.

Table 7-6: Value Market Shares by Supermarket Chain - Plain Yoghurt. Source: IRI.

196. Taken together, the elements discussed in this section allow the conclusion that private label and branded products belong to the same product market upstream.

7.1.1.3. It can be left open whether the markets for fresh milk, fresh buttermilk and plain yoghurt have to be further separated according to distribution channels

7.1.1.3.1. Product market definition proposed by the notifying parties

197. Friesland Foods and Campina both sell fresh basic dairy products to the retail and the foodservice channel OOH. The OOH segment serves restaurants, cafes, hotels, catering services, hospitals but also small businesses like bakers or other food processing entities. According to the notifying parties the OOH segment is typically served by cashand- carry businesses or foodservice wholesalers who deliver at the doorstep of the customer. In the Netherlands direct sales to OOH users would constitute only a very small proportion of the entire market segment (114).

198. In previous decisions (115) a distinction was made between these sales channels because of differences in services, sales force, price structure and packaging sizes as well as health and safety regimes. The notifying parties argue that such a distinction is no longer necessary, for several reasons: (i) most of the products sold through the OOH segment are more or less identical to those in the retail segment; (ii) both channels partially supply the same customers who cross over depending on the circumstances; (iii) in part, players in the segments overlap; (iv) prices in both segments show similar developments and the absence of private label products in OOH would not be determinative; (v) the logistics of supply and additional services are largely the same.

7.1.1.3.2. Assessment of the Commission

199. Several respondents during the market investigation indicated that differences between the retail and OOH segments continue to exist. First, it has become apparent that the distribution in the OOH segment is different from the retail segment in the sense that dairy producers try to target end customers directly and use OOH wholesalers mainly as logistic providers (116). Friesland Foods and Campina also confirmed (117) that they target end-customers directly and these direct contractual relationships currently represent [60- 70]*% of Friesland Foods turnover and about [70-80]*% of Campina's. These contracts cover bonus fees, promotional budgets, but also price, category management. OOH wholesalers are in charge of delivery. Thus, contrary to the retail segment where the end customer is approached through the retailer, in the OOH segment a significant subset of final consumers is targeted directly by dairy manufacturers.

200. In addition, the market investigation (118) showed that packaging sizes differ compared to the retail segment. They also need special packages, which are not available in the retail segment like small portions consumable immediately after purchasing or larger sizes to be used in the catering or hotel business. For these products, which according to OOH wholesalers represent up-to 25% of their turnover alternatives, a cross-over to the retail segment does not seem to be possible. Moreover, logistic services differ as clients in OOH demand daily deliveries including the weekend

201. The service requested from an OOH customer has direct implications for the wholesaler. Since OOH customers expect timely and flexible delivery from their wholesaler, the wholesaler has to look for a supplier who is able to mirror the customers' demand. This implies that unless products can be stored over longer periods (like longlife dairy or butter) suppliers in the vicinity are favoured.

202. Moreover, order volumes in the OOH segment are significantly lower than in the retail segment - notifying parties estimate that 10% of the overall market is covered by the OOH segment. Thus, orders by wholesalers are usually small and shipments over longer distances rarely occur as transport costs per unit are too high. This fact seems to limit sourcing possibilities from abroad.

203. Contrary to the argument of the notifying parties, prices in the retail segment would not constrain prices in the OOH segment as OOH clients would not be willing to cross-over and buy from a retailer because the additional services will not be offered by the retailer. Demand therefore seems to be inelastic as customers have no alternative.

204. Several competitors as well as customers have argued that the notifying parties could potentially offer "package deals" (119). Entering the market with only one dairy product would therefore be difficult as OOH wholesalers would like to get as much as possible from one supplier for logistical reasons - achieving economies of scale in deliveries in particular.

205. Competitors already present in the Netherlands in the retail segment indicated that additional investments into a distribution network and logistics would be necessary to supply the OOH segment (120).

206. Consequently, there are indications that the markets for fresh milk, fresh buttermilk, and plain yoghurt could be separated according to the distribution channels in OOH and retail. However, given the strong presence of the notifying parties on a combined market of the supply of retail and OOH and the alleged small size of the OOH segment of roughly 10% in value, the question whether the OOH segment should be considered as a separate market for the supply of fresh milk, fresh buttermilk and plain yoghurt can be left open as even on the broader market the view is taken that the proposed transaction would significantly impede effective competition.

7.1.1.4. Conclusion on relevant product market

207. It is therefore concluded that within non-organic fresh basic dairy there are separate relevant product markets for fresh milk, fresh buttermilk, plain yoghurt and custard. Custard should rather be characterised as a dessert and will be discussed in section 11. A further division into private label/branded products is not appropriate for fresh milk, fresh buttermilk and plain yoghurt. A possible distinction with regard to the distribution channel into retail/OOH can be left open.

7.1.2. Relevant Geographic Market

7.1.2.1. The relevant geographic markets for fresh milk, fresh buttermilk, and plain yoghurt are national in scope

7.1.2.1.1. Relevant geographic market proposed by the notifying parties

208. The notifying parties submit (121) that the relevant geographic market for fresh milk, fresh buttermilk and plain yoghurt at the downstream level is not wider than national while at the upstream level it is wider than the Netherlands and in any event includes Germany, Belgium and Denmark. This is based on several factors including (i) similar consumption patterns and homogeneity of the products, (ii) advances in logistics due to centralised distribution by retailers (iii) extended shelf life allowing for transportation over longer distances, (iv) procurement done on an international scale and (v) the penetration of the Dutch market by foreign competitors in particular from Belgium and Germany. All these elements would not allow suppliers in the Netherlands to increase prices profitably by 5-10% as they would lose significant volumes to foreign competitors.

7.1.2.1.2. Assessment of the Commission

209. In previous decisions (122) it has been found that fresh basic dairy markets are national in scope. Also, the market investigation has yielded strong indications that the relevant geographic market is narrower than submitted by the notifying parties.

Supply side substitution

210. The market shares of competitors (123) in the Netherlands, Germany, Belgium and Denmark vary significantly. While in the Netherlands the notifying parties both have strong positions in the upstream level, Friesland Foods is not active outside the Netherlands in fresh basic dairy while Campina has a presence in Germany (through its German subsidiaries) and Belgium, but no activities in Denmark where Arla seems to be by far the strongest player. In contrast to the highly concentrated Danish and Dutch markets, the German market is more fragmented with several local brands and producers. The uneven presence of suppliers across Member States has also been admitted by the notifying parties (124).

211. Moreover, the market investigation shows that suppliers serving the broader candidate geographic market cannot, at short notice and without cost, divert sufficient volumes of fresh dairy products sold in one region to another region, thereby offsetting any small and non-transitory divergences in relative prices (see also recital 159 above) - in accordance with the criteria laid down in the Relevant Market Notice. Strong evidence suggests that the necessary criteria for supply-side substitution across the Netherlands and neighbouring Member States are not met with respect to the upstream markets for fresh milk, fresh buttermilk and plain yoghurt.

212. First, and according to the notifying parties' own submission, prices at the upstream level vary from Member States to Member State and are on average 10% higher in Belgium and Germany than in the Netherlands (125). This already gives indications that the conditions of competition are different in different Member States and casts doubt on the ability and incentive for suppliers in these neighbouring Member States to enter the Dutch market at short notice in the event of a small, but significant permanent relative price increase.

213. Respondents to the Commission's market investigation have also confirmed this view. In particular, competitors in Germany informed the Commission that they are not active in the Netherlands or even outside their local supply area (126). Indeed, manufacturers based in Germany argued that their business is local and did not consider themselves in competition with the notifying parties; this would not change in the case of a price increase as a result of the merger.

214. In addition, the Commission asked competitors in its second phase market investigation if they would enter the Dutch market in case of a price increase of 5-10%. The replies were generally negative: while competitors agreed that a 5-10% price increase would allow them to transport milk over longer distances, they all conditioned any permanent entry, let alone uncommitted entry. For example, one competitor argued that a whole portfolio must be offered to be able to compete with the notifying parties, which requires additional investment costs. Another competitor rejected the possibility that entry would be profitable (127).

215. Customers also argued that only looking at the impact of a price increase on the ability to transport over longer distances would not be sufficient because several additional aspects constrain the reallocation of sales from other areas into the Netherlands. In particular, shipments over longer distances would have an impact on freshness. Products coming from Germany for example would arrive later in the shelves than comparable products originating in the Netherlands (128). Extended shelf life would not be a reliable solution since the T-ESL predominantly used in Germany would affect the taste of fresh milk reducing its acceptability for Dutch consumers (129), a fact confirmed by the notifying parties (130). Moreover, foreign suppliers would need to meet strict requirements by retailers, which lead to additional costs, thereby reducing the profitability of costless and immediate entry:

(a) For example, one customer argued (131) that it does not have enough warehouse capacity in its distribution centre; all fresh basic dairy products are therefore currently delivered directly to the supermarkets. Only Campina and Friesland Foods have the logistics in place to be able to handle such a direct delivery. At present, Campina/Friesland Foods deliver a certain portfolio on a daily basis to the more than [...]* stores which cannot be done by any other supplier.

(b) Another customer explained (132) that German suppliers - assuming they would be able to supply several products like the notifying parties - would have to adapt their supply chain to the needs of the retailers. While it might be possible that a German supplier currently delivers dairy products to a limited number of distribution centres, a delivery to a Dutch retailer would eventually have to be made to more centres implying additional investments in roll in containers and trucks as well as the hardware and software to ensure a tailor-made delivery.

(c) A third customer elaborated (133) that the perishable nature of all fresh dairy products would influence the logistical process and therefore sourcing opportunities. Currently the customer is supplied three times a day, which requires suppliers to be in the proximity as a longer distance would not only affect the costs, but also the freshness of the products.

216. Based on these elements - in particular, no clear indication of immediate entry in case of a price increase and the effect of distance on costs, logistics and freshness - it is concluded that supply-side substitution across the Netherlands and its neighbouring Member States is not a given.

Demand side substitution

217. The market investigation revealed that demand characteristics differ across the four Member States. In the Netherlands fresh basic dairy products represent more than 80% of total basic dairy consumption (fresh basic dairy and long-life basic dairy taken together) compared to less than 5% in Belgium and around 30% in Germany. Products like plain yoghurt in gable tops are hardly consumed outside the Netherlands (134).

218. As explained in section 7.1.1 on the delineation of the relevant product markets, demand characteristics and aggregate preferences will generally condition the purchases of supermarkets and other retailers at the upstream level, both with respect to private label products and the listing of supplier brands. In the Netherlands, consumers have a clear preference for products of Dutch origin. While during the second phase market investigation suppliers of fresh basic dairy products have supported the notifying parties' view that milk with Dutch origin has more or less the same properties as milk from elsewhere, it has also been stated that contrary to the notifying parties' claim Dutch consumers expect fresh milk, fresh buttermilk and plain yoghurt to have Dutch origin and to be packed in 1 litre gable tops.

219. According to the main Dutch retailers who supply roughly 75-80% of the Dutch market with fresh basic dairy products, the farming community "is embedded deeply in the Dutch culture" (135) and Dutch milk has long been promoted by the Dutch Dairy Organisation (Nederlandse Zuivel Organisatie, NZO (136)) that has launched several campaigns since the 1970s to promote dairy products, in particular fresh milk (137).

220. Retailers explained that this is further supported by the fact that the notifying parties organise special events to improve their image and the importance of Dutch origin. (138) Indeed, Campina and Friesland Foods both target children on their website for example and Campina organizes a farmers day (139) to allow customers, in particular families, to gain a better understanding of the daily life of a farmer and the origin of Campina's products, thereby creating a link between the product and its origin. (140)

221. Moreover, both Friesland Foods and Campina have in the recent past launched new campaigns to link their products, in particular fresh milk products, to health, sustainable production, care for animals and environment as well as excellent quality of the milk.

222. The new standard by Campina was introduced in 2006 and 2007 for their A-brands (linking the preservation of Dutch nature to dairy products) (141), one of the pillars is the concept of "dichtbij" (nearby), ensuring the awareness of the consumers that the "that the milk which arrives with the Dutch consumer, is also made by Dutch cows. (142)

223. In 2007 Friesland Foods introduced the Weidemelk (143) concept with a clear Dutch connotation; it is premised on the fact that the cows producing the raw milk have spent a substantial amount of grazing time in the open air. Although it does not explicitly state that the milk has to originate from Dutch cows (144), the fact that this concept is based on an initiative of the Dutch Ministry for Agriculture with the explicit aim of improving the image of the dairy sector vis-à-vis its citizens and consumers clearly shows that the concept implicitly supports milk from Dutch cows (145).

224. Moreover, both Friesland Foods and Campina refer explicitly to the Dutch origin of several of their products. In the case of Campina, 19 out of 25 fresh milk products make such a reference. (146) Such references are less common for Friesland Foods' products. According to the notifying parties (147) only 23 of the 83 Friesche Vlag products specify the origin of the product. However, 60 of those products use the Weidemelk logo which has a Dutch connotation.

225. While it is correct that the strategy has been imitated by Albert Heijn for its AH PL (but not for its discount PL), contrary to the claim of the notifying parties in their reply to the Article 6(1)(c) (148) Decision, consumer preferences for milk of Dutch origin are not a result of marketing efforts by retailers. Thus, the marketing or labelling of a wide majority of Dutch dairy products offered today in the Netherlands either explicitly or implicitly refers to their Dutch origin.

226. These factors have reinforced the preference of Dutch consumers for Dutch fresh dairy products, independently of the "real taste". Consequently, the main retailers in the Netherlands supplying around 75% of fresh basic dairy do not regard fresh dairy suppliers from other neighbouring regions as viable alternatives to fresh dairy of Dutch origin. They explained that foreign supply would not only increase the transport/logistic costs and freshness, but in addition "could influence the acceptance by Dutch consumers" (149).

227. Since most of the other suppliers which have responded during the market investigation are hardly active in the Netherlands, they were not able to indicate whether consumers have a specific preference for Dutch origin. Among those supplying the Dutch market, several indicated that their consumers indeed have a preference for Dutch products and indicated that retailers would face difficulties in replacing them by foreign produce. (150)

228. With respect to procurement, competitors and customers have not confirmed the view of the notifying parties that tenders for basic fresh dairy products are usually applied covering an area that is wider than national (151). German and Belgian competitors informed the Commission that they participate in tenders organised by discounters such as Dutch subsidiaries of Aldi and Lidl on a regular basis, but hardly participate in those offered by the other retail chains operating in the Netherlands. Therefore, the role of imports is limited and is mainly focused on the hard-discounters or low priced private labels. This has widely been confirmed by Dutch retailers and is mirrored in the supply data the Commission has received during the market investigation. Most of the non-Dutch supply in 2007 went to the hard-discounters or low priced private label, representing roughly 10% of total supplies, significantly less than the [20-30]*% share of imports submitted by the notifying parties (152).

229. In addition to these elements, retailers and OOH wholesalers strongly argued that because of freshness, transport costs and flexibility of deliveries - described in more detail in recital 215 - switching would be further limited.

7.1.2.2. Conclusion on the relevant geographic markets

230. In view of the above elements, it is concluded that the relevant geographic market is national for the upstream market of fresh milk, fresh buttermilk and plain yoghurt.

7.1.3. Competitive Assessment

231. The notifying parties estimate (153) that the combined size of the markets for fresh milk, fresh buttermilk, and plain yoghurt on the upstream level in the Netherlands was EUR 457.7 million in 2007 with a volume of 981.5 million litres, of which the most part was sold to retailers (90%). Fresh milk was clearly the most important product with 76% of the turnover, 14% contributed by plain yoghurt and the remaining 10% by fresh buttermilk.

7.1.3.1. Non-coordinated effects in the Fresh Milk market

232. During the market investigation the Commission received several complaints, in particular from customers concerned that the merger will ultimately lead to higher prices for fresh milk products. Indeed, a number of factors indicative of significant noncoordinated effects are present in the fresh milk market in the Netherlands.

7.1.3.1.1. The notifying parties have high market shares

233. According to the notifying parties the downstream market for fresh milk covering the Netherlands had a total value of EUR 311 858 000 in 2007, 70.4% of which is covered by private label products and the remaining 29.6% by supplier brands. Both segments have experienced a decline in volume and value, but supplier brands have lost relatively more than private labels since 2005, leading to an increase of the private label share by [0-5]* -percentage points.

234. According to the notifying parties' submission the Campina brand (with [10-20]*% market share) was the strongest supplier brand in 2007, followed by Friesland Foods 'Friesche Vlag' with [0-10]*%, and Inex (Belgium) [0-10]*%. The leading retailer in the Netherlands Albert Heijn was able to achieve a share of [30-40]*% with its two private label brands: AH and De Zaanse Hoeve.

Table 7-7: Market Shares Downstream Market for Fresh Milk, Retail, The Netherlands - Source: Form CO.

235. The Commission has relied on scanner data provided by the notifying parties to cross-check the figures above. As shown in the Table 7-8 based on the IRI data, excluding discounters, Campina brands obtained a market share of [20-30]*% and Friesland Foods brand obtained a market share of [0-10]*%. Adding the sales of discounter brands obtained from Gesellschaft fuer Konsumforschung ("GfK") data leads to adjusted market shares of [20-30]*% and [0-10]* % respectively.

Table 7-8: Market Shares for Fresh Milk at the Retail Level, Netherlands - Source: IRI.

Table 7-9: Value Sales of Discounter Brands, Netherlands - Source: GfK.

Table 7-10: Market Share Table including Discounter Brands, Netherlands - Source: IRI, GfK, own calculations.

236. The discrepancies between the market shares in value reported by the notifying parties and those computed on the basis of IRI and GfK data are not of an order of magnitude that would significantly alter the competitive assessment. These differences are mainly due to the fact that the notifying parties have not estimated the value of the downstream market on consumer prices (like IRI and GfK do), but rather based their value figures on the prices the notifying parties charge retailers. (154) Nonetheless, for the descriptive analysis of the downstream market, in particular prices, it is necessary to rely on the IRI data.

237. The picture of a potential OOH market for fresh milk looks rather different as no private label is present here. According to the notifying parties their two brands achieved a combined share of [90-100]* % in 2007 in the downstream market - no change since 2005 - with Campina being the dominant supplier with a market share of [70-80]* % followed by Friesland Foods with a market share of [10-20]* %.

Table 7-11: Market Shares Downstream Market for Fresh Milk, OOH, The Netherlands - Source: Form CO.

238. Concerning the upstream market, where retailers/OOH wholesalers source fresh milk from dairy producers, the situation is as described in Table 7-12 according to the notifying parties:

Table 7-12: Market Shares Upstream Market for Fresh Milk, Retail, Netherlands - Source: Form CO.

239. The notifying parties achieve a combined market share of [60-70]*% - basically unchanged since 2005 - because of their strong position in the branded segment (Campina [60-70]*%, Friesland Foods [10-20]*%) as well as in the private label segment. In the private label segment Campina had a [30-40]*% market share in 2007, Friesland Foods had a [20-30]*% market share, Inex had a [10-20]*% market share, and the German suppliers Humana and Molkerei Ammerland had a market share of [5-10]*% and [5-10]*% respectively.

240. Based on the notifying parties own submission, therefore, their combined market shares in the upstream market was just below [70-80]*% with only negligible variations over the past 3 years. Such high market shares suggest that the merged entity is likely to enjoy significant market power in the absence of any mitigating factors. Furthermore it is worth noting that the overlap between the notifying parties is also significant with Campina holding a [40-50]*% market share pre-merger and Friesland Foods a [20-30]*% share. This is indicative of strong non-coordinated effects in the form of higher prices or reduced quality and choice. There are no other suppliers with significant presence in either the private label or the branded segments, let alone in both simultaneously.

241. Finally, since there is no private label available in the OOH segment, the market shares at the upstream level are similar to those for the downstream one. Thus on a potential OOH market, the merger would lead to a monopoly as Campina has [70-80]*% and Friesland Foods [10-20]*% of the market (value, in 2007).

7.1.3.1.2. The notifying parties are the closest competitors

242. In addition to pointing to high market shares, all complainants (155) explained that the notifying parties would be the closest competitors. In particular, with respect to the upstream market complainants indicated that the notifying parties are the only undertakings currently able to offer the full range of fresh basic dairy products in sufficient volumes and quality. (156) Moreover, they are the only firms able to supply both supplier brands and private label products. Customers often source from both notifying parties and considered the rivalry between Friesland Foods and Campina as an essential source of competition. Thus, the merger would eliminate the primary source of competitive pressure that prevails in the fresh milk market.

243. The notifying parties currently compete directly both in the supply of branded products and also in the supply of private labels to supermarkets. Furthermore private labels and supplier brands compete in the downstream markets as discussed section 7.1.1. As a result, Campina and Friesland Foods mutually exert competitive pressure on each other by three means:

(a) First, supermarkets willing to carry a private label are currently able to source a large part of their products from either of the notifying parties, thereby enjoying a certain degree of countervailing buyer power.

(b) Second, the supplier brands of the notifying parties are the closest competitors in the downstream market thereby allowing supermarket chains that wish to carry at least one brand of fresh milk to play one against the other in listing and pricing negotiations.

(c) Third, supermarket chains can also threaten to source private label from Campina in negotiating prices for Friesland Foods brands, and vice-versa. The significance of this competitive constraint depends on the degree of substitutability of private label and supplier brands among end-customers.

244. By bringing together the main competitors in the supply of both brands and private labels the merger would undermine all three means of competitive interaction in the fresh milk market and impede the exercise of any significant degree of countervailing buyer power.

7.1.3.1.3. Pricing data indicate that the notifying parties mutually exert competitive pressure

245. With respect to private labels, internal data supplied by the notifying parties suggests that two of the four major supermarket chains ([...]* and [...]*) are supplied primarily by Campina. [...]* as well as [...]*, particularly since April 2007, are supplied by Friesland Foods. The data also shows that wholesale prices for brands as well as for private labels increased in the past year. It appears that supermarkets have increased the purchase of private label (in particular [...]*) whereas purchases of brands have remained relatively stable or have slightly declined.

246. It is interesting to compare wholesale prices charged by the notifying parties to supermarket chains for private labels and for brands. In the case of Campina, prices of private label to [...]* were [...]* between January 2005 and April 2007, within a EUR [...]* to [...]* cent band. Prices of private labels to [...]*, however, are up to EUR [...]* lower at the beginning of the period although the difference narrowed to around EUR [...]* after January 2006. (157) With respect to [...]*, prices are at the level of [...]* up until January 2006 but then they fall slightly resulting in a price differential with [...]* and expanding the price differential with [...]*. Such price differences suggest Campina is pricing strategically taking into consideration the total profitability of sales of brands and private labels at each individual supermarket chain.

[...]*

Figure 7-7: Average Prices of Campina to Retailers for private label fresh milk - Source: internal data submitted by Campina.

[...]*

Figure 7-8: Average Prices of Campina to Retailers for branded fresh milk - Source: internal data submitted by Campina.

247. It is interesting to note the sudden but temporary increase in prices in the last quarter of 2005. Prices of private labels to [...]*, [...]* and [...]* all increased by approximately [...]*% and then quickly fell back. The same price development can be observed in the prices of supplies of Campina branded fresh milk in particular for [...]* and [...]*. No such price increase was observed for Friesland Foods, suggesting that the source of this price spike was a supply shock affecting only Campina. It appears this increase in Campina's wholesale prices for fresh milk did not affect retail prices of either supplier brands or private labels at any of the affected supermarkets. This implies that retailers were not able to pass on to the end-customer the increase in the wholesale price. This suggests that strong competitive pressure pre-merger is being exerted by Friesland Foods upon Campina in both segments.

248. It also appears that Campina's strategy towards [...]*compared to [...]* changed as of the last quarter 2005. During 2005, the difference in the wholesale price of Campina branded fresh milk and private labels was around EUR [...]* for [...]* but only EUR [...]* for [...]*. However, as of February 2006 the Campina lowered the price of its brand to [...]* by an average of EUR [...]* while at the same time increasing the price of private label by EUR [...]*. This probably influenced the increase in the sales of the Campina brand from [20-30]*% in 2005 to [20-30]*% in 2006 at [...]* supermarkets.

249. While it is acknowledged that prices and quantities are the result of the competitive interaction between suppliers and customers, the above observations provide confirmation that Campina, as would be expected, implements a co-ordinated pricing strategy for branded and private label supplies with respect to each individual supermarket chain. (158) Of further relevance is the fact that Friesland Foods appears to have significantly and abruptly reduced the price of its brand to [...]* slightly earlier, in September 2006, by EUR [ ]*. This might also explain the reduction in the price of Campina's brand a few months later. Campina's decision to increase the price of private labels to [...]* can be explained as a way to offset the loss in margin in the sale of brands resulting from the increase competitive pressure exerted by Friesland Foods brands on Campina.

[...]*

Figure 7-9: Average Prices of Friesland Foods to Retailers for private label fresh milk - Source: internal data submitted by Friesland Foods.

[...]*

Figure 7-10: Average Prices of Friesland Foods to Retailers for branded label fresh milk - Source: internal data submitted by Friesland Foods.

250. It is also worth analysing the relative prices paid by [...]*. Campina supplies around [...]* as much branded milk as Friesland Foods (a fact confirmed by the IRI data). However, it appears that Friesland Foods is the main supplier of private labels to [...]*, with Campina supplying only limited quantities up to early 2006. Indeed, Friesland Foods' prices of private labels to [...]* are only slightly below its own brand prices. It is thus not surprising that in the face of two sources of competitive pressure from Friesland Foods (brand and private label), Campina offers its brand to [...]* at a lower price than to any of the other supermarket chains. Moreover, branded prices to [...]* were affected the least by the generalised increase in raw milk prices as of April 2007.

251. Finally it is also interesting to assess the evolution of prices and quantities supplied to [...]* by the two notifying parties. As mentioned in recital 193 [...]* only stocks its own private label, primarily supplied by the notifying parties. As of April 2007, the price of raw milk increased significantly presumably leading to an increase in demand for private labels as customers saw prices rise increasingly above their reservation price for branded milk. As a result of this increase in demand for private labels the prices for supplier brands increased relatively less than that for private labels. This would tend to favour [...]* in particular, with its strong private label and its strategy of carrying no supplier brands. Indeed, the purchases of [...]* from Campina appear to have increased as of April 2007, despite the fact that the wholesale price of Campina (the largest supplier of [...]*) increased approximately EUR [...]* as of April 2007. [...]* could have threatened to switch more volumes to Friesland Foods which would have, in normal circumstances, constrained the wholesale price of Campina. Indeed, at that point [...]* significantly increased its purchases from Friesland Foods (see figures 7-11 and 7-12). However the price paid appears to have been around EUR [...]* more than EUR [...]* cents higher than Campina. Despite this significant increase in purchases from Friesland Foods, [...]* could not prevent the price from increasing to EUR [...]* cents by the end of 2007. This evolution in prices and volumes suggests that [...]* was unable to find alternative suppliers of private labels at a time when both its own demand and market prices were increasing.

[...]*

Figure 7-11: Quantities of private label fresh milk supplied to retailers by Campina - Source: Internal data submitted by Campina.

[...]*

Figure 7-12: Quantities of private label fresh milk supplied to retailers by Friesland Foods - Source: Internal data submitted by Friesland Foods.

7.1.3.1.4. Pricing data indicate that the notifying parties' brands are perceived as the closest competitors in the retail market

252. An assessment of the retail market further suggests that the Campina and Friesland Foods brands mutually constrain each other, thus allowing supermarket chains such as Superunie, which carry both brands, to obtain low prices. First, the two brands of the notifying parties achieved a combined share of [90-100]*% in the downstream market in 2007. Furthermore, the notifying parties' brands are both positioned as premium brands with prices historically [30-40]*% higher than low-end private labels. It should be noted, however, that this gap has narrowed in the past year as a result of the significant increase in raw milk prices which have affected private labels relatively more than supplier brands.

253. The graph in Figure 7-13 depicts the price evolution of the notifying parties' brands and that of the major private labels. Together the brands represented account for more than [90-100]*% of the market.

[...]*

Figure 7-13: Weighted average prices for fresh milk downstream level, Netherlands - Source: IRI.

254. Since fresh milk is a rather homogeneous product, the observed price differences can be regarded as a proxy for the degree of product differentiation. The same conclusions can be drawn from an analysis of price comparisons over time within each supermarket chain. Comparing price levels for the different private label brands and the supplier brands clearly shows that Campina and Friesland Foods are the only two premium supplier brands currently available in the market with a significant volume. Unilever is hardly active, while Inex (Belgium) prices around EUR 0.[20-30]* below the notifying parties.

255. In addition to the premium brands of the notifying parties (with the exception of [...]*), major supermarket chains usually offer two private label products. Figure 7-13 further confirms that while some brands are positioned at the lower-end of the market (for example, Albert Heijn's DNSH or SuperUnie DKR) others seem to be closer to the premium brands of Campina and Friesland Foods (for example Albert Heijn's ABJ or Laurus SU8).

256. The Commission attached a set of econometric results on estimation of retail demand systems to its Statement of Objections. These results complemented the other, larger set of qualitative evidence and descriptive statistics. The goal of the econometric model was to assess whether supplier brands exert a greater competitive constraint on each other than private labels. In their response to the Statement of Objections, the notifying parties raised a number of criticisms regarding the interpretation and the robustness of the econometric results. (159)

257. Most of the criticisms put forward by the notifying parties are invalid or based on a misunderstanding of the methodology. Moreover, a number of criticisms regarding the robustness of the results can be addressed by extending the econometric model. Such an extension is justified on economic grounds albeit for reasons other than those raised by the notifying parties. Depending on the market in question such extension leads to different outcomes. In the case of Fresh Flavoured Dairy Drinks a dynamic specification is used that accounts for habit formation and inertia in the purchasing behaviour of end customers, as explained in more detail in the Annex 1. In this case the extended specification leads to results qualitative similar to those presented in the Statement of Objections and thus the conclusions drawn are maintained (see section 11.2.3. and Annex 1). In the case of Long-Life Dairy Drinks, inventory behaviour, as argued by the notifying parties, will likely lead to biased estimates. Neither the static specification presented in the Statement of Objections nor a more dynamic specification appears appropriate to correct for this circumstance and thus the results presented in the Statement of Objections cannot be used for the purposes of this Decision (see in particular 11.3.3.1.2 and Annex 1). Finally, in the case of fresh basic dairy habit formation is also a concern and thus the dynamic specification is also preferable. The results obtained from this dynamic specification are economically valid and also econometrically more robust than those presented in the Statement of Objections. However the results do not support the initial finding that supplier brands constrain each other more than private labels. (160)

258. It is important to emphasise that this conclusion does not in any way contradict or affect the merits of the other, qualitative and quantitative evidence put forward in this Decision in relation to each of the affected markets. Overall the analysis confirms the qualitative evidence and the inferences made from the high market shares of the notifying parties that they are in fact the closest competitors on the market. The brands of the notifying parties appear to mutually constrain each other. The competitive constraint exerted by private labels on supplier brands is important.

259. Given the strong position of the notifying parties in the supply of branded as well as private label fresh milk, it is evident that since the merged entity will supply both branded products and private label products to supermarket chains, when considering an increase in the wholesale price for the Campina and Friesland Foods brands it will take into consideration not only the fact that customers would switch to private labels but also that switching would increase demand in the private label segment, allowing the notifying parties to also increase the price of private label supplies to supermarket chains. Of course the converse is also true.

7.1.3.1.5. Customers have limited possibilities of switching suppliers

260. Supermarket chains do not have the possibility to switch to an alternative brand with the required level of recognition to compete with those of the notifying parties. In principle, given the high level of penetration of private label fresh milk in the Netherlands the large supermarket chains could consider switching further volumes to private labels. However, as explained, the notifying parties are also the main suppliers of private labels. Even [...]*, which essentially only carries its own private label, obtains the majority of its supplies from the notifying parties. Indeed, the market investigation has revealed that in the past - until the announcement of the merger - retailers usually shifted between Friesland Foods and Campina to source their supplies for private labels. (161)

261. One retailer (162) argued in its response to the market investigation that "substituting Campina or Friesland would only be possible with several suppliers given the volume constraint by producers like Katshaar or Vecozuivel. This, however, would imply that costs in handling deliveries would increase." Consequently, even if enough spare capacity were available on the market, such a contract would have to be divided, lowering the benefits for the retailer coming from economies of scale in production and logistics, thereby increasing the costs and finally the price to customers. (163)

262. The market investigation found little evidence that competitive pressure from dairy producers outside the Netherlands would be able to constrain the notifying parties postmerger. According to the notifying parties, prices in neighbouring MemberStates are normally higher than in the Netherlands, thus the economic incentives to supply customers in the Netherlands also appear weak. Although respondents to the market investigation indicated that prices have, in the meantime, become more aligned across Member States they also indicated that transport costs, different product composition and a particular packaging system requested by some retailers in the Netherlands (so called roll in containers) would make expansion costly. This is supported by the main Dutch retailers who note that foreign suppliers could not deliver products of Dutch origin.

263. Based on the above elements, in particular the smaller scale of competitors as well as the additional costs related to specific needs of Dutch consumers, the possibility of switching to other branded or private label suppliers would be limited after the merger.

7.1.3.1.6. Competitors are unlikely to increase supply if prices increase

264. In addition, the market investigation showed that it is unlikely that competitors would significantly increase supply in the case of a price increase. First, several main suppliers of fresh milk, fresh buttermilk or plain yoghurt have no additional capacity available to be able to supply the Dutch market. (164) Smaller companies (165), in the Netherlands in particular, have either no spare capacity or would face difficulties handling the large volumes required - for example Friesland Foods currently has a contract with a retailer to supply of up to [...]* million litres of fresh milk. While this can easily be handled in a typical Friesland Foods factory which has a capacity of for [...]*-[...]*million litres per year, smaller companies have total available capacities of around 100 million litres per year.

265. In addition, given the consumer preferences for Dutch origin fresh milk products, alternative suppliers need access to Dutch raw milk on a large scale, which might be difficult. One supplier, 49.9% owned by Campina and currently sourcing its raw milk from Campina explained that it would be extremely difficult to substitute current volumes were Campina to halt its deliveries. (166)

266. Even those Dutch suppliers which have indicated spare production capacity in the range of 25% allowing them to increase output for fresh basic dairy products would need sufficient amounts of raw milk for such an expansion. Given the preferences of Dutch consumers for fresh basic dairy products of Dutch origin as described in section 7.1.2 on the relevant geographic market, access to Dutch raw milk seems to be a necessary condition to be able to increase supply. Since the notifying parties are in a position to control more than [70-80]*% of Dutch raw milk, access to raw milk could pose a problem for competitors unless they were to enter or also expand in the market for the procurement of raw milk. This concern was raised during the market investigation by some customers (167) who argued that switching would only be possible on a limited scale because competitors would not be able to obtain sufficient amounts of raw milk. (168)

267. While in Germany and Belgium access to raw milk does not seem to be a problem, fresh basic dairy manufacturers based in the Netherlands, in particular smaller competitors, noted that "entering the market for procurement of milk would be difficult since it would not be easy to convince farmers, in particular members of cooperatives, to re-direct their supply against equal conditions as Campina/Friesland" (169) and that producers not organised as a co-operative would have to pay a risk premium compared to the notifying parties as they would not be able to offer a long-term purchase guarantee. (170)

268. Moreover, due to its 49.9% participation in and its deliveries of raw milk to Farm Dairy, Campina seems to be able to significantly influence the ability of Farm Dairy to expand in fresh milk. According to an internal document (171), Farm Dairy currently supplies [...]* to whom Campina would like to supply more. One possible course of action proposed is to "[...]*."

7.1.3.1.7. Lack of countervailing buyer power

269. The notifying parties argue that countervailing buyer power from the largest retail chains (for example AHOLD, SuperUnie, Aldi, Lidl etc) would restrain them from raising prices, because of the retailers' strong demand for private label products relative to brands.

270. According to the Horizontal Merger Guidelines, countervailing power is "the bargaining strength that the buyer has vis-a-vis the seller in commercial negotiations due to its size, its commercial significance to the seller and its ability to switch to alternative suppliers". The guidelines suggests three ways in which countervailing buyer power may offset the negative impact of an upstream merger: "...if the buyer could immediately switch to other suppliers, credibly threaten to vertically integrate into the upstream market or to sponsor upstream expansion or entry for instance by persuading a potential entrant to enter by committing to placing large orders with this company." (172)

271. Size is also seen as important in this context: "It is more likely that large and sophisticated customers will possess this kind of countervailing buyer power than smaller firms in a fragmented industry". (173)

272. In Commission Decision SCA/Metsa Tissue, (174) claims that buyers had sufficient strength to resist any attempt to raise prices above the competitive level after the merger were dismissed. It was argued that buyers had limited switching alternatives as the notifying parties would have around [80-90]*% of current sales in the markets for toilet tissue and kitchen towels and the brands with the strongest brand recognition in Sweden. It was also argued that the notifying parties could price discriminate between different customers. Hence, even if the largest customers would be able to exercise some countervailing buyer power this would not protect smaller customers.

273. In this case, the notifying parties are the dominant producers of the main retailers' private label products with a combined market share in the Netherlands of [60-70]*%. Furthermore, it is not sufficient that buyer power exists prior to the merger - it must also exist and remain effective following the merger. This is because a merger of two suppliers may reduce buyer power if it thereby removes an important alternative source of supply.

274. Furthermore, the countervailing buyer power defence rests on the presumption that powerful buyers will be able to protect themselves - and thus, ultimately, also final consumers - against adverse changes in the terms of supply following the notified merger. However, not all buyers in the fresh milk market are large and powerful (for example Albert Heijn or Superunie are almost five times larger than Super de Boer), and OOH wholesalers lack a sufficient scale since the OOH market only represents 10% of total fresh milk turnover. In this potential market the largest wholesaler has a 20% market share, which means it represents 2% of the overall market which raises the question of whether, and under what circumstances, the presence of (few) powerful buyers also shields other, smaller or less powerful buyers from the imposition of more adverse terms of supply.

275. First, retailers purchasing small quantities of fresh milk relative to the total sales of the merged entity are very unlikely to exert any significant degree of countervailing buyer. If the merged entity fails to negotiate successfully with a small buyer, it will find it relatively easy to sell its freed-up capacity to other buyers. What is more, as the quantity supplied to each of the remaining buyers increases only by a little, this leads to only a small reduction in total profits.

276. Second, even if a large retailer were to derive some degree of bargaining power from their large size or a sophisticated purchasing strategy, there is no generally convincing reason why other retailers should be also be positively affected. The merged entity would not have to grant smaller buyers similar discounts as those offered to larger buyers to prevent them from switching. In fact, the market investigation suggested that the retail market for fresh milk is relatively competitive. This competition would induce larger retailers to take steps that protect only themselves, if such a possibility existed, from a lessening of competition between suppliers, as these firms may stand to gain from an increase in their rivals' costs after the merger.

277. Thus, the analysis focuses only on the alleged ability of large buyers to exert countervailing purchasing power vis-à-vis the merged entity post-merger.

7.1.3.1.7.1. Large Retailers' ability to switch to other suppliers

278. The main source of buyer power, which also represents the key constraint on suppliers' ability to behave anticompetitively, is the ability to substitute away from any given supplier's input. The merger of Campina and Friesland Foods will remove one of the two main suppliers considered to be the closest competitors in the market. The market investigation confirmed that there are no alternative suppliers in the Netherlands of comparable size to that of either of the notifying parties. Thus, the merger reduces the buyers' available choices and will thereby erode any existing ability to discipline the notifying parties through the threat of switching from one to the other. This is true both when considering private label supplies as well as supplies of branded products.

279. In general the profitability and thus the credibility of substitution should increase with the retailer's relative size. The larger the retailer's intended purchase, the more profitable and thus the more credible it will be to switch suppliers or even to take preemptive steps such as (strategic) multiple sourcing in response to any price increase. This argument suggests that commanding a larger purchasing volume essentially reduces the "lock-in" with a particular supplier as it becomes more profitable (and thus more credible) to switch (175).

280. Bargaining power, however, depends, in part on the relative damage to each party's profits that a failure to reach agreement can cause. The evidence suggests that the merged entity would enjoy considerably greater bargaining power in the fresh milk market than any of the large Dutch retailers or the discounters. This is due to the following factors:

(a) While size might appear to benefit the buyer, a larger buyer may also have more difficulty replacing purchases from the notifying parties than would a smaller buyer. Indeed, in the present case, a large retailer will find it more difficult to access the spare capacity of any available alternative suppliers. As already mentioned, there are no suppliers of fresh milk, whether branded or private label, with enough capacity to serve the requirements of the larger supermarket chains in the Netherlands other than the notifying parties. Companies such as Farm Dairy or Katshaar, two potential alternative suppliers, have not only a total capacity in the fresh milk market which is significantly lower than that of the notifying parties, but also lack access to raw milk and are partially owned by Campina (in the case of Farm Dairy).

(b) Second, the notifying parties supply large retailers not only with fresh milk but also with a number of other dairy products where the merged entity will enjoy considerable market power, for example in the branded market for flavoured dairy drinks. The merged entity can credibly threaten to limit supplies or increase the price of such other branded products in order to increase its leverage in negotiations with retailers with respect to branded or private label fresh milk.

(c) Third, in the event of a breakdown of negotiations, the merged entity would have a more profitable outside option relative to large retailers. Fresh milk is considered by many supermarket chains as an essential product to ensure customers will make their regular purchases in their shops. In some instances, it can be expected that retailers will in fact consider fresh milk as a loss leader, that is, a product sold at a low price (at cost or below cost) to stimulate other, profitable sales. In contrast to retailers, the notifying parties are not dependent on selling fresh basic dairy as they can reduce the capacity allocated to the production of fresh milk in the medium to long-term and channel raw milk to other dairy products. In the event that the negotiations relate only to private labels the notifying parties can reallocate the production of fresh milk to brands, and vice versa.

281. Even today, where large retailers can threaten to switch from one of the notifying parties to the other, they all rely on the notifying parties to different degrees, as in the case of [...]* for private labels or for both supplier brands and private labels (for [...]*, [...]* and [...]*).

282. Another factor highlighting the weak bargaining power of retailers relative to the merged entity is that no retail chain has a "gatekeeper" status to any segment of the retail fresh milk market where it would enjoy a monopoly or near monopoly (176). The merged entity could therefore rely on other large retailers to cover the whole market in the event of a breakdown in negotiations with any of those retailers. As a result the merged entity would take into consideration the fact that generally, a discount to one buyer - whether in the supply of supplier brands or private labels - may erode the market share of other competing buyer(s), which in the case where these buyers purchase from the same supplier, would reduce that supplier's profits.

283. Economic theory on bargaining also suggests that asymmetric information affects the distribution of bargaining power. What is found is that typically, a party can obtain a better deal if it has more information about the other side's outside option or, more generally, more information about the incremental surplus that is on the table. If a larger buyer has more opportunity and more incentive to gather such information, this could further enhance its bargaining power. Already the notifying parties can be expected to have access to more accurate information concerning its buyer's needs than the other way around. This is partly due to the fact that the notifying parties purchase IRI, Nielsen and GfK scanner data which allows fresh milk suppliers to reconstruct the market and monitor the pricing, sales, positioning, evolution and strengths and weaknesses of, in particular, large retailers. Second, the maturity of the fresh milk market also implies that that the merged entity can more easily identify and react to new developments or changes in the strategy of retailers, such as attempts to source from other potential suppliers. Third, the merger will combine the knowledge of the notifying parties and their market intelligence concerning the strategies and requirements of the large retailers as well as discounters.

7.1.3.1.7.2. Sponsoring entry or backwards integration in the processing of fresh milk

284. The threat of integration is particularly powerful when the buyer produces, in part, the input in question but relies on suppliers to cover its demand. This makes the threat to switch to internal production more credible. It also provides the buyer with adequate information about input production costs which can be useful in negotiating with suppliers. However, backwards integration does not appear to be a credible threat in the context of this case. There are no instances in the fresh milk market of retailers vertically integrating backwards in the processing of milk or, for that matter, acquiring ownership of or establishing exclusive supply relationships with farms directly.

285. Furthermore, backwards integration may only be a feasible and therefore credible alternative for buyers that are not only large but also quite sophisticated. Supermarket chains in the Netherlands follow a different business model and there is no indication that they have the ability or incentive to enter the dairy processing market since this is only one category out of hundred other product categories sold in their shelves. Sourcing from outside the Netherlands is made difficult by the fact that there are significant barriers for permanent entry as discussed in section 7.1.3.1.8.

286. Furthermore, as shown in section 7.1.3.1, no suppliers have been identified in the course of the market investigation with the ability and incentive to enter the Dutch fresh milk market. There are also no reasons to believe that large retailers are in a position to undermine all the existing barriers to entry identified such that new suppliers would enter the market effectively, timely and to a sufficient degree so as to offset the noncoordinated effects likely to the result from the merger.

287. The notifying parties have agued that even if it would be difficult to expect new entry in the supply of branded fresh milk, large retailers can sponsor new entry by placing large orders for private labels. They point to the fact that Albert Heijn has obtained private labels supplies from Katshaar in the recent past. The notifying parties also claim that discounters, such as Aldi and Lidl, can establish long term supply relationships with fresh milk suppliers in Germany through their home market headquarters.

288. However, the notifying parties fail to recognise that all of the restrictions to entry identified apply to potential suppliers of private labels as well as potential suppliers of supplier brands. Moreover, it is unlikely that any such expansion of private labels sourced from abroad would exert any significant constraint on the notifying parties' abilities to raise prices in the Netherlands.

289. First, discounters enjoy a limited market share of the retail market in the Netherlands in particular compared to other neighbouring Member States. Also, established large retailers already have a strong presence in private labels in the Netherlands, with even two types of private labels, including a low-end private label which limits further expansion by discounters.

290. Second, even if large retailers managed to reduce their reliance on the merged entity for a large part of their supplies of private labels the fact that the notifying parties are the dominant supplier of brands suggest that they will be largely unconstrained in raising the price of supplier brands.

7.1.3.1.7.3. Conclusion

291. It is therefore concluded that the customers of the notifying parties will not be in a position to exert countervailing buyer power to a sufficient degree so as to offset or mitigate the non-coordinated effects of the merger.

7.1.3.1.8. Entry unlikely to occur

292. Despite the high market shares the notifying parties argue that in case of a price increase domestic or foreign competitors would constrain the notifying parties. While the market investigation has shown that certain entry barriers can be overcome (roll in containers do not represent a significant barrier to entry and gable tops are also available from some foreign producers), others remain. Transport costs, Dutch origin and freshness are also obstacles for foreign producers. Longer supply distances would increase costs for the retailer (and their customer) and put foreign suppliers at a disadvantage concerning freshness because Dutch suppliers would always be able to supply fresher products as they are located closer to the distribution centres of the retailers. Since Dutch customers consider freshness as a key variable in their purchasing decision, Dutch incumbents have a clear competitive advantage.

293. The discussion in section 7.1.2 on the geographic product market definition must also be considered in this respect. Foreign competitors did not assert an immediate entry in case of a 5-10% price increase in the Netherlands.

7.1.3.1.9. Conclusion on the competitive assessment

294. For the reasons set out in this section, it is concluded that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the market for fresh milk in the Netherlands, which is a substantial part of the common market regardless of whether this market should be further segmented according to distribution channels.

7.1.3.2. Non-coordinated effects in the Fresh Buttermilk market

295. During the market investigation in both phases the Commission received several strong complaints, in particular from customers concerned that the merger will ultimately lead to higher prices. Indeed, a number of factors indicative of significant non-coordinated effects are present in the fresh buttermilk market in the Netherlands.

7.1.3.2.1. The notifying parties have high market shares

296. According to the notifying parties, the downstream market for fresh buttermilk covering the Netherlands had a total value of EUR 47 411 000 in 2007, 40.5% of which was covered by private labels and the remaining 59.5% by supplier brands. Both segments have experienced a decline in volume and value since 2005.

297. According to the notifying parties' submission, the Campina brand was the strongest supplier brand in 2007 with [20-30]*% market share, followed by Friesland Foods 'Friesche Vlag' with [10-20]*%, and Zuivelboerderij (Netherlands) [0-5]* %. The leading retailer in the Netherlands - Albert Heijn - was able to achieve an [10-20]*% share with its two private label brands AH and De Zaanse Hoeve.

Data and Netherlands Source: Parties' volume based on Campina and Friesland Foods data. Total Market Data and Competitor Data estimated by Friesland Foods and Campina. Competitors with a market share below 1% are grouped together in other competitor branded.

Table 7-13: Market Shares Downstream Market for Fresh Buttermilk, Retail, The Netherlands - Source: Form CO.

298. The Commission has relied on scanner data provided by the notifying parties to cross-check the figures above. As shown in the table 7-14 (based on the IRI data, excluding discounters), Campina brands obtained a market share of [20-30]*% and Friesland Foods' brand obtained a market share of [10-20]*%.

Table 7-14: Market Shares Downstream Market for Fresh Buttermilk, Retail, the Netherlands - Source: IRI.

299. The discrepancies between the market shares in value reported by the notifying parties and those computed on the basis of IRI are not of an order of magnitude that would significantly alter the competitive assessment. The difference is probably due to the fact that the scanner data does not include the sales of Aldi, Lidl and Koopconsult, sellers of private label products and are based on the prices retailers charge their customers, while the notifying parties have based their estimates on the prices they charge retailers.

300. A potential OOH market for fresh buttermilk would look rather different as no private label is present on this market. According to the notifying parties, the two brands of the notifying parties achieve a combined share of [90-100]*% in the downstream market in 2007 - no change since 2005 - with Campina being the dominant supplier with [70- 80]*% followed by Friesland Foods with [20-30]*%.

Table 7-15: Market Shares Downstream Market for Fresh Milk, OOH, The Netherlands - Source: Form CO.

301. Looking at the upstream market, where retailers/OOH wholesalers source fresh buttermilk from dairy producers, the situation is as follows according to the notifying parties:

Table 7-16: Market Shares Upstream Market for Fresh Milk, Retail, The Netherlands - Source: Form CO.

302. Merged, the notifying parties would achieve a combined market share of [80-90]*% - significantly higher than the [70-80]*% market share in 2005 - because of their strong position in the branded segment (Campina [60-70]*%, Friesland Foods [20-30]*%) as well as in the private label segment. In the private label segment Campina had [40-50]*% in 2007, Friesland Foods [40-50]*%, with Inex [0-5]*%, and the German suppliers Humana ([0-5]*%) and Molkerei Ammerland ([0-5]*%) as other competitors.

303. Thus, based on the notifying parties own submissions, the combined market shares of the notifying parties in the upstream market will be above [80-90]*% with significant increases over the past 3 years. Such high market shares suggest the merged entity is likely to enjoy significant market power in the absence of any mitigating factors. Furthermore, it is worth noting that the overlap between the notifying parties is also significant with Campina holding a [40-50]*% market share pre-merger and Friesland Foods a [30-40]*% share. This is indicative of strong non-coordinated effects in the form of higher prices or reduced quality and choice. There are no other suppliers with significant presence in either the private label or the branded segments, let alone in both simultaneously.

304. Finally, since there is no private label available in the OOH segment, the market shares at the upstream level in that segment are similar to those for the downstream one. Thus, on a potential OOH market, the merger would lead to a monopoly as Campina has [70-80]*% and Friesland Foods [20-30]*% of the market (value, in 2007).

7.1.3.2.2. The notifying parties are the closest competitors

305. In addition to such high market shares, all complainants (177) explained that the notifying parties would also be the closest competitors. In particular, with respect to the upstream market, complainants indicated that the notifying parties are the only ones currently able to offer the full range of fresh basic dairy products in sufficient volumes and quality. (178) Moreover, they are also the only firms able to supply both in the form of supplier brands and private label. Customers often source from both notifying parties and considered the rivalry between Friesland Foods and Campina as an essential source of competition. Thus, the merger would eliminate the primary source of competitive pressure that prevails in the fresh buttermilk market.

306. Currently, the notifying parties compete directly both in the supply of branded products and also in the supply of private labels to supermarkets. Furthermore, private labels and supplier brands compete in the downstream markets as discussed in section

7.1.1.2. As a result Campina and Friesland Foods mutually exert competitive pressure on each other by three means:

(a) First, supermarkets willing to carry a private label are currently able to source a large part of their products from either of the notifying parties, thereby enjoying a certain degree of countervailing buyer power.

(b) Second, the supplier brands of the notifying parties are the closest competitors in the downstream market thereby allowing supermarket chains to carry at least one brand of fresh buttermilk to play one against the other in listing and pricing negotiations.

(c) Third, supermarket chains can also threaten to source private labels from Campina in negotiating prices for Friesland Foods brands, and vice-versa. The significance of this competitive constraint depends on the degree of substitutability of private label and supplier brands among end-customers.

307. By bringing together the main competitors in the supply of both brands and private labels the merger would undermine all three means of competitive interaction in the fresh buttermilk market and impede the exercise of any significant degree of countervailing buyer power.

308. An assessment of the retail market further suggests that the Campina and Friesland Foods brands mutually constrain each other, thus allowing supermarket chains such as Superunie, which carry both brands, to obtain low prices. First, the two brands of the notifying parties achieve a combined share of [70-80]*% in the downstream market in 2007. Furthermore, the notifying parties' brands are both positioned as premium brands with prices historically [30-40]* % higher than low-end private labels. It should be noted, however, that this gap has narrowed in the past year as a result of the significant increase in raw milk prices which have affected private labels relatively more than supplier brands.

309. The graph in Figure 7-14 depicts the price evolution of the notifying parties' brands and that of the major private labels. Together, the brands represented account for more than [90-100]*% of the market.

[...]*

Figure 7-14: Evolution of Average Weighted Prices, Fresh Buttermilk, Retail level, Netherlands - Source: IRI.

310. Since fresh buttermilk is a rather homogeneous product, the observed price differences can be regarded as a proxy for the degree of product differentiation. Comparing price levels for the different private label brands and the supplier brands clearly shows that Campina and Friesland Foods are the only two premium supplier brands currently available in the market with a significant volume.

311. In addition to the premium brands of the notifying parties (with the exception of Albert Heijn) major supermarket chains usually offer two private label products. Figure 7-14 further confirms that while some brands are positioned at the lower-end of the market (for example, Albert Heijn's DNSH or SuperUnie DKR) others seem to be closer to the premium brands of Campina and Friesland Foods (for example Albert Heijn's ABJ or Laurus SU8).

312. The Commission attached a set of econometric results on estimation of retail demand systems to its Statement of Objections. These results complemented the other, larger set of qualitative evidence and descriptive statistics. The goal of the econometric model was to assess whether supplier brands exert a greater competitive constraint on each other than private labels. The notifying parties in their response to the Statement of Objections have raised a number of criticisms regarding the interpretation and the robustness of the econometric results. (179)

313. Most of the criticisms put forward by the notifying parties are invalid or based on a misunderstanding of the methodology. A number of criticisms regarding the robustness of the results can be addressed by extending the econometric model. Such extension is justified on economic grounds albeit for reasons other than those raised by the notifying parties. However, since the model was not helpful in the analysis of the fresh buttermilk market as a disaggregation was not possible, the criticism of the notifying parties does not in any way affect the merits of the qualitative evidence put forward in this Decision. (180)

314. Despite the less than satisfactory results, based on the qualitative evidence received during the market investigation - in particular the sourcing pattern of retailers, the pricing of the brands, and the market shares of the notifying parties and the perception of the customers - the brands of the notifying parties mutually constrain each other. This provides a strong indication that the merged entity would be in a position to increase the price of branded fresh buttermilk.

315. Since the merged entity will supply both branded products and private label products to supermarket chains, when considering an increase in the wholesale price for the Campina and Friesland Foods brands, it will take into consideration potential switching behaviour of customers to private labels, allowing the notifying parties to also increase the price of private label supplies to supermarket chains. The converse is also true.

7.1.3.2.3. Customers have limited possibilities of switching suppliers

316. As the general structure of the market as well as the competitive interaction for fresh buttermilk is similar to the one of fresh milk, the arguments put forward in recitals 260 - 263 concerning switching possibilities also apply for fresh buttermilk.

317. The market investigation found few indications that competitive pressure from dairy producers outside the Netherlands would be able to constrain the notifying parties postmerger. According to the notifying parties, the buttermilk market has been declining in volume and value since 2005, thus the economic incentives to supply customers in the Netherlands appear weak. In addition, different product composition and a particular packaging system requested by some retailers in the Netherlands (so called roll in containers) would make expansion costly. This has been complemented by the claim of the main Dutch retailers that foreign suppliers could not deliver products of Dutch origin.

318. Based on the above elements, in particular the smaller scale of competitors as well as the additional costs related with specific needs of Dutch consumers, it is concluded that the possibility to switch to other branded or private label suppliers would be limited after the merger.

7.1.3.2.4. Competitors are unlikely to increase supply if prices increase

319. In addition, the market investigation showed that the main suppliers of fresh milk, fresh buttermilk or plain yoghurt have no additional capacity available to supply the Dutch market. Smaller companies have either no spare capacity or would face difficulties to handle the large volumes required.

320. As the general structure of the market as well as the competitive interaction for fresh buttermilk is similar to the one of fresh milk, the arguments put forward in section

7.1.3.1.6 concerning increase of supply by competitors also apply for fresh buttermilk and are not repeated here.

7.1.3.2.5. Lack of countervailing buyer power

321. The notifying parties also argue that countervailing buyer power from the largest retail chains (for example Albert Heijn, SuperUnie, Schuitema, or Super de Boer) would restrain them from raising prices, because of the retailers' strong demand for private label products relative to brands. "In this respect", the notifying parties note (181), "any postmerger price increase sought by the Parties would be retaliated by retailers who will be able to rapidly expand the share of PL products or delist brands."

322. As the general structure of the market as well as the competitive interaction for fresh buttermilk is similar to the one of fresh milk, the arguments put forward in section

7.1.3.1.7. concerning buyer power also apply for fresh buttermilk.

323. Since the notifying parties are perceived as the closest competitors and customers would have only limited switching possibilities post-merger, an immediate switch to alternative sources of supply can be excluded. In addition, the notifying parties are the dominant producers of the main retailers' private label products with a combined market share in the Netherlands of [80-90]*%, thus threatening to switch to private label products would be a limited option for retailers as supply alternatives in that segment are limited as well.

324. Furthermore, it is not sufficient that buyer power exists prior to the merger, it must also exist and remain effective following the merger. This is because the merger of two suppliers may reduce buyer power if it thereby removes a credible alternative. The merger of Campina and Friesland Foods will remove one of the two main suppliers considered to be the closest competitors in the market, since alternative suppliers with a similar size could not be identified in the course of the market investigation.

325. Therefore, it is concluded that countervailing buyer power post-merger would not be sufficient to off-set potential adverse effects of the merger.

7.1.3.2.6. Entry unlikely to occur

326. The notifying parties argue that in the case of a price increase domestic or foreign competitors would constrain the notifying parties. While the market investigation confirmed that some of the entry barriers mentioned in the market investigation could be overcome (roll in container do not represent a barrier to entry and gable tops are available also from foreign producers) others such as transport costs, Dutch origin and freshness are also obstacles for foreign suppliers to step in. Longer supply distance would increase costs for the retailer (and their customer) and put foreign suppliers at a disadvantage with regard to freshness because Dutch suppliers would always be able to supply fresher products as they are located closer to the distribution centres of the retailers. Since Dutch customers consider freshness as a key variable in their purchasing decision, Dutch incumbents have a clear competitive advantage.

327. Moreover, as has already been mentioned in section 7.1.2 on the geographic product market, foreign competitors did not indicate an immediate entry in case of a 5-10% price increase in the Netherlands.

7.1.3.2.7. Conclusion on the competitive assessment

328. For the reasons set out above, it is concluded that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the market for fresh buttermilk in the Netherlands, which is a substantial part of the common market, regardless of whether this market should be segmented according to the distribution channel.

7.1.3.3. Non-coordinated effects in the plain yoghurt market

329. During the market investigation in both phases the Commission received several complaints, in particular from customers concerned that the merger will ultimately lead to higher prices. Indeed, a number of factors indicative of significant non-coordinated effects are present in the plain yoghurt market in the Netherlands.

7.1.3.3.1. The notifying parties have high market shares

330. According to the notifying parties, the downstream market for plain yoghurt covering the Netherlands had a total value of EUR 61 724 000 in 2007, 54.7% of which was covered by private label and the remaining 45.3% by supplier brands. Private labels have achieved an increase in value by 3.9-percentage points since 2005.

331. According to the notifying parties' submission, Campina was the strongest supplier of brands in 2007 with [20-30]*% market share, followed by Friesland Foods 'Friesche Vlag' with [10-20]*%, and Den Elder (Netherlands) [0-5]* %. The leading retailer in the Netherlands Albert Heijn was able to achieve a share of [20- 30]*% with its two private label brands AH and De Zaanse Hoeve.

Table 7-17: Market Shares Downstream Market for Plain Yoghurt, Retail, The Netherlands - Source: Form CO.

332. The Commission has relied on scanner data provided by the notifying parties to cross-check the figures above. As shown in table 7-18 based on the IRI data, excluding discounters, Campina brands obtained a market share of [20-30]*% and Friesland Foods brand [10-20]*%.

Table 7-18: Market Shares Downstream Market for Plain Yoghurt, Retail, The Netherlands - Source: IRI.

333. The discrepancies between the market shares in value reported by the notifying parties and those computed on the basis of IRI, are not of an order of magnitude that would significantly alter the competitive assessment. The difference is probably due to the fact that the scanner data do not include the sales of Aldi, Lidl and Koopconsult, sellers of private label products and are based on the prices retailers charge their customers, while the notifying parties have based their estimates on the prices they charge retailers.

334. A potential OOH market for plain yoghurt would look rather different as no private label is present on the OOH market. According to the notifying parties, their two brands achieve a combined share of [90-100]*% in the downstream market in 2007 - almost no change since 2005 - with Campina being the dominant supplier with [70-80]*% followed by Friesland Foods with [10-20]* %.

Table 7-19: Market Shares Downstream Market for Plain Yoghurt, OOH, The Netherlands - Source: Form CO.

335. Looking at the upstream market, where retailers/OOH wholesalers source plain yoghurt from dairy producers, the situation is as follows according to the notifying parties:

Table 7-20: Market Shares Upstream Market for Plain Yoghurt, Retail, The Netherlands - Source: Form CO.

336. Merged the notifying parties would achieve a combined market share of [80-90]*% in 2007 because of their strong position in the branded segment (Campina [50-60]*%, Friesland Foods [30-40]*%) as well as in the private label segment. In the private label segment Campina had a share of [40-50]*% in 2007, Friesland Foods had a [30-40]*% share, with Katshaar holding [0-5]*%, the Belgian supplier with Inex [0-5]*% and Molkerei Ammerland with [0-5]*%.

337. Thus, based on the notifying parties own submission, the combined market shares of the notifying parties in the upstream market will be above [80-90]*%. Such high markets suggest the merged entity is likely to enjoy significant market power in the absence of any mitigating factors. Furthermore, it is worth noting that the overlap between the notifying parties is also significant with Campina holding a [50-60]*% market share premerger and Friesland Foods a [30-40]*% share. This is indicative of strong noncoordinated effects in the form of higher prices or reduced quality and choice. There are no other suppliers with significant presence in either the private label or the branded segments, let alone in both simultaneously.

338. Finally, since there is no private label available in the OOH segment, the market shares at the upstream level in that segment are similar to those for the downstream one. Thus on a potential OOH market, the merger would lead to a monopoly as Campina has a [70-80]*% share and Friesland Foods has a [10-20]*% share of the market (value, in 2007).

7.1.3.3.2. The notifying parties are the closest competitors

339. In addition to pointing towards high market shares, all complainants (182) explained that the notifying parties would be the closest competitors. In particular, with respect to the upstream market complainants indicated that the notifying parties are the only undertakings currently able to offer the full range of fresh basic dairy products in sufficient volumes and quality. (183) Moreover, they are also the only firms able to supply both in the form of supplier brands and private labels. Customers often source from both notifying parties and considered the rivalry between Friesland Foods and Campina as an essential source of competition. Thus, the merger would eliminate the primary source of competitive pressure that prevails in the plain yoghurt market.

340. Currently the notifying parties compete directly both in the supply of branded products and also in the supply of private labels to supermarkets. Furthermore private labels and supplier brands compete in the downstream markets as discussed in section

7.1.1.2. As a result Campina and Friesland Foods mutually exert competitive pressure on each other through three means:

(a) First, supermarkets willing to carry a private label are currently able to source a large part of their requirements from either of the notifying parties, thereby enjoying a certain degree of countervailing buyer power.

(b) Second, the supplier brands of the notifying parties are the closest competitors in the downstream market thereby allowing supermarket chains to carry at least one brand of plain yoghurt to play one against the other in listing and pricing negotiations.

(c) Third, supermarket chains can also threaten to source private labels from Campina in negotiating prices for Friesland Foods brands, and vice-versa. The significance of this competitive constraint depends on the degree of substitutability of private label and supplier brands among end-customers.

341. By bringing together the main competitors in the supply of both brands and private labels the merger would undermine all three means of competitive interaction in the plain yoghurt market and impede the exercise of any significant degree of countervailing buyer power.

342. An assessment of the retail market further suggests that the Campina and Friesland Foods brands mutually constrain each other, thus allowing supermarket chains such as Superunie, which carry both brands, to obtain low prices. First, the two brands of the notifying parties achieved a combined share of [80-90]*% in the downstream branded segment in 2007. Furthermore, the notifying parties' brands are both positioned as premium brands with prices historically [30-40]*% higher than low-end private labels. However, this gap has narrowed in the past year as a result of the significant increase in raw milk prices which have affected private labels relatively more than supplier brands.

343. The graph in Figure 7-15 depicts the price evolution of the notifying parties' brands and that of the major private labels. Together, the brands represented account for more than [90-100]*% of the market.

[...]*

Figure 7-15: Evolution of Average Weighted Prices Plain Yoghurt, Retail, The Netherlands - Source: IRI.

344. Since plain yoghurt is a rather homogeneous product, the observed price differences can be regarded as a proxy for the degree of product differentiation. Comparing price levels for the different private label brands and the supplier brands clearly shows that Campina and Friesland Foods are the only two premium supplier brands currently available in the market with significant volumes.

345. In addition to the premium brands of the notifying parties, major supermarket chains usually offer two private label products. Figure 7-15 further confirms that while some brands are positioned at the lower-end of the market (for example, Albert Heijn's DNSH or SuperUnie ZVLM) others seem to be closer to the premium brands of Campina and Friesland Foods (for example Albert Heijn's ABJ or Laurus SU8).

346. The Commission attached a set of econometric results on estimation of retail demand systems to its Statement of Objections. These results complemented the other, larger set of qualitative evidence and descriptive statistics. The goal of the econometric model was to assess whether supplier brands exert a greater competitive constraint on each other than private labels. The notifying parties raised a number of criticisms in their response to the Statement of Objections regarding the interpretation and the robustness of the econometric results. (184)

347. Most of the criticisms put forward by the notifying parties are invalid or based on a misunderstanding of the methodology. Moreover, a number of criticisms regarding the robustness of the results can be addressed by extending the econometric model. Such an extension is also justified on economic grounds albeit for reasons other than those raised by the notifying parties. However, the results of the extended regressions do not support the initial finding that supplier brands constrain each other more than private labels. (185)

348. It is important to emphasise that this conclusion does not in any way contradict or affect the merits of the other, qualitative and quantitative evidence put forward in this Decision in each of the affected markets.

349. It follows that the brands of the notifying parties mutually constrain each other. This provides an indication that, as would be expected on the basis of the market share overlap, the merged entity would be in a position to increase prices post merger, of supplier brands in particular. This will allow them to increase wholesale prices of private label to the various supermarkets, since, as in the case of fresh milk, the merged entity would supply both branded products and private label to supermarket chains.

7.1.3.3.3. Customers have limited possibilities of switching suppliers

350. Supermarket chains do not have the possibility of switching to alternative brands with the required level of recognition to compete with those of the notifying parties. In principle, given the high level of penetration of private labels in plain yoghurt in the Netherlands the large supermarket chains could consider switching further volumes to private labels. However, the notifying parties are also the main suppliers of private labels. Even Albert Heijn, which essentially sells the majority of plain yoghurt under its own private label, obtains most of its supplies from the notifying parties. Indeed, the market investigation has revealed that in the past - up to the announcement of the merger - retailers usually shifted between Friesland Foods and Campina to source their supplies for private label. (186)

351. One retailer (187) argued in its response to the market investigation that other suppliers would either lack the necessary volumes or would - in the case of foreign dairy companies - not be able to provide the taste demanded by Dutch consumers. Consequently, even if enough spare capacity were available on the market, such a contract would have to be divided, lowering the benefits for the retailer coming from economies of scale in production and logistics, thereby increasing the costs and finally the price to customers. (188)

352. The market investigation found few indications that competitive pressure from dairy producers outside the Netherlands would be able to constrain the notifying parties postmerger. In addition, different product composition and a particular packaging system requested by some retailers in the Netherlands (so called roll in containers) would make expansion costly. This has been complemented by the claim of the main Dutch retailers who noted that foreign suppliers could not deliver products of Dutch origin.

353. Based on these elements, in particular the smaller scale of competitors as well as the additional costs related with specific needs of Dutch consumers, it is concluded that the possibility to switch to other branded or private label suppliers would be limited after the merger.

7.1.3.3.4. Competitors are unlikely to increase supply if prices increase

354. In addition, the market investigation showed that the main suppliers of plain yoghurt have no additional capacity available to supply the Dutch market. Smaller companies have either no spare capacity or would face difficulties to handle the large volumes required.

355. As the general structure of the market as well as the competitive interaction for plain yoghurt is similar to the one of fresh milk, the arguments put forward in section 7.1.3.1.6 concerning increase of supply by competitors also apply for plain yoghurt and are not repeated here.

7.1.3.3.5. Lack of countervailing buyer power

356. The notifying parties also argue that countervailing buyer power from the largest retail chains (for example Albert Heijn, SuperUnie, Schuitema, or Super de Boer) would restrain them from raising prices, because of the retailers' strong demand for private label products relative to brands. "In this respect", the notifying parties note (189), "any postmerger price increase sought by the Parties would be retaliated by retailers who will be able to rapidly expand the share of PL products or delist brands."

357. As the general structure of the market as well as the competitive interaction for custard is similar to the one of fresh milk, most of the arguments put forward in section

7.1.3.1.7 concerning buyer power do also apply for plain yoghurt.

358. Since the notifying parties are perceived as the closest competitors and customers would have only limited switching possibilities post-merger, an immediate switch to alternative sources of supply can be excluded. In addition, the notifying parties are the dominant producers of the main retailers' private label products with a combined market share in the Netherlands of [80-90]*%, thus threatening to switch to private label products would be a limited option for retailers as supply alternatives in that segment are limited as well.

359. Furthermore, it is not sufficient that buyer power exists prior to the merger; it must also exist and remain effective following the merger. This is because a merger of two suppliers may reduce buyer power if it thereby removes a credible alternative. The merger of Campina and Friesland Foods will remove one of the two main suppliers considered to be the closest competitors in the market, since alternative suppliers with a similar size could not be identified in the course of the market investigation.

360. Therefore, it is concluded that countervailing buyer power post-merger would not be sufficient to off-set potential adverse effects of the merger.

7.1.3.3.6. Entry unlikely to occur

361. The notifying parties argue that in the case of a price increase domestic or foreign competitors would constrain the notifying parties. As explained in the assessment for fresh milk (see recital (292)) and fresh buttermilk (see recital (326)) the market investigation confirmed that roll in containers do not represent a barrier to entry and gable tops are also available from foreign producers. Transport costs, Dutch origin and freshness would nevertheless also be a barrier to entry if foreign suppliers would like to step in. Longer supply distances would increase costs for the retailer (and their customers) and put the foreign suppliers at a disadvantage concerning freshness because Dutch suppliers would always be able to supply fresher products as they are located closer to the distribution centres of the retailers. Since Dutch customers consider freshness as a key variable in their purchasing decision, Dutch incumbents have a clear competitive advantage.

362. Moreover, as has already been mentioned in section 7.1.2 on the geographic product market, foreign competitors did not indicate an immediate entry in case of a 5-10% price increase in the Netherlands.

7.1.3.3.7. Conclusion on competitive assessment

363. For the reasons set out in this section, it is concluded that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the market for plain yoghurt in the Netherlands, which is a substantial part of the common market, regardless of whether this market should be segmented according to distribution channels.

7.1.3.4. Overall conclusion on competitive assessment fresh basic dairy

364. Based on the above, it is concluded that the proposed merger would significantly impede effective competition in the markets for (i) fresh milk, (ii) fresh buttermilk and (iii) plain yoghurt in the Netherlands regardless whether these markets are segmented according to distribution channels.

7.2. Long-life basic dairy products

7.2.1. Relevant Product Markets

365. According to the notifying parties long-life basic dairy products include long-life milk, long-life buttermilk and long-life custard. Plain yoghurt is not produced by the notifying parties. Friesland Foods and Campina are both active in the production and sale of long-life milk. Only Friesland Foods produces long-life buttermilk while Campina produces long-life custard. (190)

366. Long-life milk is liquid milk which is produced from raw milk that is first (partially) skimmed. Subsequently, two different methods can be applied in order to produce longlife milk:

(a) Sterilisation: First the milk is pre-sterilised for 3-4 seconds at 130-140°C. It is then cooled down and poured into glass or polyethylene bottles. These bottles undergo a period of final sterilisation at 110-120°C for 10-20 minutes. Sterilised milk can be kept for months at room temperature, if not exposed to sunlight.

(b) Ultra High Temperature ("UHT"): The milk is subjected to a temperature of 135- 150°C for only 2-5 seconds and put into separately sterilised packaging material, normally brick packs. The short heat treatment means that the original taste of the milk remains virtually unaffected. UHT milk can be kept at room temperature for months.

367. Long-life buttermilk is a liquid fermented dairy product with a characteristic sour taste produced from fresh buttermilk. To produce long-life buttermilk fresh buttermilk undergoes the UHT treatment and packaging as described in recital (366).

368. Long-life custard is fresh custard that has undergone an UHT treatment and packaging as described in recital (366).

7.2.1.1. Long-life milk, long-life buttermilk, and long-life custard each form a single product market

7.2.1.1.1. Product market definition proposed by the notifying parties

369. Friesland Foods and Campina submit that long-life milk, long-life buttermilk, and long-life custard form a single product market in their notification because "suppliers are able to switch production to the relevant products and market them in the short term without incurring significant additional costs or risks in response to small and permanent changes in relative prices". (191) In particular they explain (192) that:

(a) The production process for the above-mentioned products is largely the same. The processing (standardising, skimming and UHT-treatment) is to a large extent the same and the packaging equipment is identical for longlife milk and buttermilk.

(b) In addition, a switch on one production line from one product to a different one is possible at low cost (below EUR 1 million), a short lead time (maximum 3 months) and a downtime of one week.

(c) This switching opportunity is supported by the contracting practice of retailers who tender their supply once a year and grant sufficient lead time to allow companies to adjust their production facilities in order to participate in the tender.

370. In addition, the notifying parties indicated that long-life milk and long-life buttermilk would be largely interchangeable as a drink, although not all customers of milk appreciate the taste of buttermilk. Finally, the notifying parties argue that the question of demand and supply-side substitutability could be left open since the only overlap in longlife basic dairy would occur in long-life milk. (193)

7.2.1.1.2. Assessment of the Commission

371. The market investigation has not been able to confirm the product market definition proposed by the notifying parties whereby long-life milk, long-life buttermilk and longlife custard form part of the same relevant product market.

Demand-side substitution

372. Respondents confirmed (194) the lack of or limited demand-side substitutability between long-life custard, long-life milk, and long-life buttermilk: no respondent grouped these three products together; hardly anybody saw the possibility of demand-side substitution between long-life milk and long-life custard or long-life buttermilk and custard. Respondents indicated that consumption patterns, the taste (sour) and the thickness would differ among these products. (195)

373. Only a minority of replies were of the opinion that long-life milk and long-life buttermilk could be exchangeable. (196)

374. Moreover, the market investigation revealed that long-life buttermilk and long-life custard are hardly present in the market. Several retailers do not offer these products at all or sell only insignificant volumes, which further supports a lack of demand-side substitutability. One customer for example explained (197) that "long-life buttermilk does not exist in The Netherlands and long-life custard is nearly gone and will soon not exist anymore in the Dutch markets." Similar another retailer (198) "[who] is not active in branded or private label products in the category long-life buttermilk and long-life custard."

375. It is therefore concluded that there is no demand-side substitution between the different long-life basic dairy products.

Supply-side substitution

376. As explained in detail in section 7.1.1.1, immediate and costless entry may render a small but permanent increase in price unprofitable, thus leading to broader markets or undermining the relevance of market shares as indicative of market power. Supply-side substitution involves entry at a low cost without incurring irreversible investments and implies the ability to substitute production of one product for another at short notice in response to relative price variations (199). In addition, production substitution among a group of products is found to be technologically feasible and economically viable for most, if not all, firms selling one or more of those products. Only if these conditions are met will supply-side substitutability have an impact in terms of effectiveness and immediacy equivalent to the demand substitution effect.

377. With these considerations in mind, the notifying parties' view that long-life milk, long-life buttermilk, and long-life custard are supply-side substitutes for the purposes of relevant product market definition have not been supported by the responses to the market investigation.

378. Firstly, only one out of 10 competitors who replied to the market investigation was supplying all the three long-life basic dairy products to customers in 2007. The majority of respondents focus on long-life milk only. (200)

379. Secondly, a majority of competitors replied (201) that they are either not able to switch between all the different long-life basic dairy products (202) or indicated that such a switch would only be possible after additional investment because the processing and packaging lines would be different depending on the product. It follows that the requirement that "most suppliers" would be able to switch between the products is not met in this case.

380. Furthermore, even if the notifying parties were able to produce the three products in question using the same production equipment, it is not the case that most competing suppliers of any such products have the flexibility to reallocate production and sales interchangeably and costlessly to other long-life dairy products. Indeed, even the notifying parties concede that adding a new product into an existing production site implies additional costs and a lead time of at least three months.

381. Hence, it can be concluded that the conditions laid down in the Relevant Market Notice are not fulfilled. In particular, it is not the case that most, if not all suppliers should be able to switch production to the relevant products and market them in the short term without incurring significant costs or risks. In any event, since Campina is not producing any long-life buttermilk and Friesland Foods is not active in long-life custard, the only overlap in a narrower product market definition would occur in long-life milk, which represents roughly 98% of all long-life basic dairy sales in the Netherlands or even a wider region including Belgium, Germany and the Netherlands. Therefore, even if supply-side substitutability were to apply, the overall size of the market would hardly be different from the one for long-life milk only.

382. Given the lack of substitution by customers and the lack of supply-side substitutability, it is concluded that long-life milk, long-life buttermilk, and long-life custard are separate relevant product markets. Because the only overlap between the notifying parties occurs in long-life milk, the other two products will not be discussed any further. (203)

7.2.1.2. Private label and supplier brands belong to the same relevant upstream market.

7.2.1.2.1. Product market definition proposed by the notifying parties

383. As explained in relation to fresh milk (see section 7.1.1.2) the notifying parties agree that the vertical segmentation of a particular product market for consumer products like long-life milk can be made between an upstream market for the sourcing of products by retailers (and OOH wholesalers serving hotels, restaurant, catering services and filling stations) and a downstream market for the sale of foodstuffs by retailers to consumers for two reasons.

384. This approach recognises that on the retail-to-consumer level private label products compete with (producer) branded products. Therefore, the market share of private label products can be attributed to the retailers that own the private label brands at the downstream level.

385. The approach also recognises that the competitive conditions on the markets on which retailers and OOH wholesalers source their products can be fundamentally different from the competitive conditions on the retailer-to-consumer markets. (204) The distinction between private label and branded products at the upstream level must be carefully analysed.

386. The notifying parties agree (205) that the procurement of private label products and the procurement of branded products by retailers (and OOH wholesalers) might constitute two neighbouring, but separate markets, in the event the competitive conditions under which retailers source private label products and branded products would be fundamentally different. This may be the case in situations where specific branded products are perceived as "must carry" products by retailers, in the sense that significant profits might be lost in the event of delisting such branded products.

387. In the context of the market for long-life basic dairy products, the notifying parties claim that both Campina and Friesland Foods are under constant pressure from private label products. This is exemplified by the fact that the market is dominated by private label products. (206)

388. Thus, according to the notifying parties, private label and branded products would belong to the same product market at the upstream level.

7.2.1.2.2. Assessment of the Commission

389. All long-life milk producers sell their products to retailers and/or OOH wholesalers which, in turn, sell these products to consumers. Therefore, there are two stages in the supply chain: the upstream level, where long-life milk is produced and supplied to retailers/OOH wholesalers and the downstream level of supply to final consumers. Campina and Friesland Foods are only active on the production and supply upstream (to retailers and/or OOH wholesalers).

390. In a recent consumer goods case (207), the upstream level where retailers source their products has been distinguished from a downstream level where the products are sold on to the final customer. The same approach has been followed for the market for long-life milk.

391. Long-life milk is available in two broad categories: brands owned by the dairy manufacturer and private label products which are marketed by the retailer (208). Both Campina and Friesland Foods produce both branded and private label long-life milk.

392. As explained in detail in section 7.1.1.2 whether private labels and supplier brands belong to the same product market upstream also depends on several factors.

393. In particular, the following must be assessed:

(a) whether both types of brands, in general, compete closely with each other from the perspective of the end-customer; and

(b) the extent to which upstream suppliers of private label and/or brands as well as the purchasing retailers, take into account in their negotiations upstream the competitive pressure that private labels and supplier brands mutually exert on each other at consumer level..

394. In the case of long-life milk there is evidence suggesting that private label and supplier brands compete in the downstream market and exert a competitive on each other constraint sufficiently relevant for the purposes of market delineation.

395. In their responses to the market investigation the majority of customers and competitors confirmed that private label and branded products compete at the retail level and that the quality of private label products is to a large extent similar to branded products (209).

396. In addition, on the downstream market for long-life milk the share of private label in 2007 was 53.6% in value terms and has increased by 1.5 percentage points over the past two years. (210)

397. Other features of the products also suggest that private labels and supplier brands likely compete within the same market. First, Campina and Friesland Foods long-life milk brands are sold primarily in 1 litre containers (between [80-90]*%- [80-90]* % of total volume) (211). A similar pattern is observed for private label brands.

398. Secondly, there is almost no differentiation between supplier brands and private labels across segments, with more than 99% of all sales in the "regular" segment and only limited sales in the "weight management" segment (212) for both brand categories.

399. This broad analysis of IRI market-level data suggests that there is a sufficient degree of competitive interaction between supplier brands and private labels for long-life milk for both to be considered as belonging to the same relevant product market at retail level.

400. Looking at the upstream level the evidence has been mixed First, customers and Competitors (213) explained that the procurement of branded products follows a different procedure than the one for private label products. In the case of branded products suppliers and retailers agree in bilateral negotiation on the gross price, discounts, listing fees, and promotions. For private labels, a tender procedure is usually used, followed by a selection of preferred suppliers which agree with the retailer on a net price in final negotiations.

401. This in itself would not be decisive if the same suppliers are active in both segments. The market presence of competitors and the notifying parties varies to some extent when it comes to branded and private label long-life milk. According to the notifying parties' estimates German competitors like Humana, Immergut, and Milchunion Hocheifel mainly supply private label in the Netherlands and only offer their brands to a limited degree, while the Belgian competitor Inza is active in both segments.

402. On the other hand, Campina has recently reduced its presence in the supply of private label long-life milk in the last years and Friesland Foods only has a small presence in this segment and achieves roughly [90-100]*% of turnover with its branded product.

403. Despite this asymmetry in the conditions of competition (the branded producers could constrain the private label producers, but the opposite does not seem possible as a private label producer would have to invest into a brand and marketing before being able to compete), it should be remembered that private labels account for a significant proportion - 53.6% - of the long-life milk market which has been growing over time. Given the relative importance of private label sales suppliers cannot be expected to ignore the competitive pressure that private labels exert on their brands.

404. Also, retailers in the Netherlands have private labels in addition to the brands offered by Campina, Friesland Foods or Inza. The existence of high and low-level private labels indicates that retailers consider and condition their private labels in relation to supplier brands. (214)

405. Taken together, all the above elements, in particular the growing importance of private label long-life milk relative to branded products and the competitive interaction between the two brands that affects the negotiations upstream, allow for the conclusion that in the context of the present case private label and branded products belong to the same product market upstream.

7.2.1.3. It can be left open whether the market for long-life milk has to be further separated according to distribution channel

7.2.1.3.1. Product market definition proposed by the notifying parties

406. Friesland Foods and Campina both sell long-life milk to the retail and the foodservice channels OOH. The OOH segment serves restaurants, cafes, hotels, catering services, hospitals but also small businesses such as bakers or other food processing entities. According to the notifying parties, the OOH segment is typically served by foodservice wholesalers who deliver at the doorstep of the customer and by cash-and-carry business. In the Netherlands, direct sales to OOH users would constitute only a very small proportion of the entire market segment (215).

407. In previous decisions (216), these sales channels were distinguished because of differences in services, sales force, price structure, packaging sizes and health and safety regimes. The notifying parties argue that such a distinction would no longer be necessary for several reasons: (i) most of the products sold through the OOH segment are more or less identical to those in the retail segment; (ii) both channels partially supply the same customers who cross over depending on the circumstances; (iii) in part, players in the segments overlap; (iv) prices in both segments show similar developments and the absence of private label in OOH would not be determinative; (v) the logistics of supply and additional services are largely the same.

7.2.1.3.2. Assessment of the Commission

408. Contrary to the replies for fresh dairy, the Commission has received minimal feedback during the market investigation indicating significant differences between the retail and the OOH segment for long-life milk products.

409. The most important difference between the retail and the OOH segment seems to be the way in which the end customer is approached. In the OOH segment, a significant subset of final consumers is targeted directly by the dairy manufacturers as has been confirmed by Friesland Foods and Campina (217). They stated that they target endcustomers directly and these direct contractual relationships currently represent [60- 70]*% of Friesland Foods turnover and about [70-80]*% for Campina. These contracts cover bonus fees, promotional budgets, but also price and category management. OOH wholesalers are in charge of the delivery.

410. The market investigation (218) also received indications that packaging sizes and services for long-life milk differ compared to the retail segment. While in the retail segment the one litre format dominates, five litre packaging was also mentioned for OOH in the market investigation. Deliveries seemed to be targeted directly to the outlets in OOH and less to distribution centres as in the case of retailers.

411. However, in contrast to fresh products where the perishable nature of the products makes transport over longer distances difficult and deliveries have to be more frequent, long-life milk can be stored over longer periods in an ambient environment, thus transport, storage and less frequent deliveries are possible. (219)

412. Although order volumes in the OOH segment are significantly lower than in the retail segment - parties estimate that 11% of the overall market is covered by the OOH segment - shipments over longer distances do occur because long-life milk can easily be stored and is not perishable, thus contrary to fresh products orders can cover the supply not only of days, but weeks. This fact seems to extend the sourcing possibilities from abroad.

413. Finally, only a few competitors argued that prices would differ, while more were of the opinion that prices would be the same in both markets (220) - one competitor even argued that the "OOH market is sometimes used as a dump market". (221)

414. On the basis of these responses to the Commission's market investigation, there are no clear indications that the market for long-life milk should be separated according to the distribution channel in OOH and retail. In any event, as will be shown in the competitive assessment, the question whether the OOH segment should be considered as a separate market for the supply of long-life milk can be left open, as even on the narrower market the view is taken that the proposed transaction would not significantly impede effective competition.

7.2.1.4. Conclusion on relevant product market

415. On the basis of the elements discussed above, it is concluded that there is a separate relevant product market for long-life milk. No separation into private label/branded products is necessary for long-life milk. A possible distinction with regard to the distribution channel into retail/OOH can be left open.

7.2.2. Relevant Geographic Market

7.2.2.1. The relevant geographic market for long-life milk is wider than national in scope

7.2.2.1.1. Relevant geographic market proposed by the notifying parties

416. The notifying parties submit (222) that the relevant geographic market for long-life milk at the downstream level is national in scope while at the upstream level it is wider than the Netherlands and includes Belgium and Germany. This is due to several factors including (i) significant penetration of the Dutch market by Belgian and German suppliers, (ii) largely homogeneous product (iii) long shelf life allowing for transportation over longer distances, and (iv) the fact that the entire supply of the Netherlands would come from abroad. All these elements would not allow suppliers in the Netherlands to increase prices profitably by 5-10% as they would lose significant volumes to foreign competitors.

7.2.2.1.2. Assessment of the Commission

417. In a previous Commission Decision (223) indications were found that the relevant geographic market for long-life milk might be wider than national. The market investigation in the present case allows for the conclusion that the relevant geographic market is wider than the Netherlands.

418. First, the market investigation confirmed that several Belgian and German producers are already supplying Dutch retailers and OOH wholesalers with long-life milk. In particular, the Belgian producers Inza and Inex as well as the German suppliers Milchunion Hocheifel and Nordmilch deliver significant volumes into the Netherlands. (224)

419. Second, a majority of customers as well as competitors also informed the Commission that long-life milk could be considered a homogeneous product with no particular difference in taste, quality or packaging. (225) Unlike for fresh milk where Dutch raw milk is considered a key input, the wide majority of retailers who responded to the market investigation explained that the origin of the raw milk would not be important in the case of long-life milk. (226)

420. Third, the fact that long-life milk is not a perishable product that can be shipped in an ambient environment would facilitate trade between Member States. Indeed, one retailer (227) stated that "because of longer shelf-life, transport is not as crucial as for fresh dairy". Almost all Dutch customers source from Belgian and German companies. Even Campina and Friesland Foods supply their Dutch customers from their production facilities in Belgium or Germany.

421. Finally, the majority of competitors participating in the market investigation replied that the prices for long-life milk are similar across Belgian, Germany and the Netherlands, indicating that the conditions of competition do not differ between these Member States.

7.2.2.1.3. Conclusion on the relevant geographic markets

422. Based on these elements - in particular, the current supply patterns, the number of foreign suppliers and similar prices - it is concluded that the relevant geographic market for long-life milk is wider than the Netherlands and includes Belgium and the Germany as well.

7.2.3. Competitive Assessment

423. The notifying parties estimate (228) that the size of the market for long-life milk on the upstream level in Belgium, Germany, and the Netherlands was EUR 2 100 511 000 in 2007 with a volume of 4 314 483 000 litres, of which the most part was sold to retailers (80%). In Germany the turnover of long-life milk was EUR 1 744 855 000, followed by Belgium (EUR 257 014 000) and the Netherlands (EUR 98 642 000).

7.2.3.1. Proposed transaction would not result in non-coordinated effects in the long-life milk market

424. During the market investigation, the Commission received only a few complaints from customers indicating that the merger might lead to higher prices, while most customers as well as competitors were not expecting a significant impact of the proposed transaction on the market. A number of factors indicative of significant non-coordinated effects are not present in the long-life milk market.

7.2.3.1.1. Market shares

425. According to the notifying parties, the downstream market for long-life milk covering the Netherlands had a total value of EUR 98 642 000 in 2007, 53.6% of which is covered by private label and the remaining 46.4% by supplier brands. Both segments have experienced an increase in value, but supplier brands have gained relatively less than private labels since 2005, leading to an increase of the private label share by 1.5- percentage points.

426. According to the notifying parties' submission, the Friesche Vlag brand (Friesland Foods) with a [20-30]*% market share was the strongest supplier brand in 2007, followed by Campina with [5-10]*%, and Inza (Belgium) [5-10]*%. The two leading retailers in the Netherlands Superunie and Albert Heijn were able to achieve a share of [5-10]* % (Superunie) and [5-10]* % with their private label brands, respectively.

Table 7-21: Market Shares Downstream Market for Long-Life Milk, Retail, The Netherlands - Source: Form CO.

427. A potential OOH market for long-life milk is rather different as no private label is present here. According to the notifying parties, their two brands achieved a combined share of [70-80]* % in the downstream market in 2007, followed by Inza and Inex, both Belgian with [10-20]* % each.

Table 7-22: Market Shares Downstream Market for Long-Life Milk, OOH, The Netherlands - Source: Form CO.

428. Looking at the upstream market, where retailers/OOH wholesalers source long-life milk from dairy producers, the situation looks rather different:

Table 7-23: Market Shares Upstream Market for Long-Life Milk, Retail and OOH, Belgium, Germany and The Netherlands - Source: Form CO.

429. Merged, the notifying parties would achieve a combined market share of [5-10]*% - basically unchanged since 2005 - as a result of their position in the branded segment (Campina [0-5]*%, Friesland Foods [0-5]*%) followed by several suppliers from Belgium and Germany with market shares for their respective brands of around [0-5]*%. The bulk of the market however is covered by private label products - 61.8% - where the notifying parties achieve [0-5]*% (Campina) and [0-5]*% (Friesland Foods). Here the German suppliers Nordmilch, Mueller and Milchunion Hocheifel supply roughly twothirds of the entire segment. (229)

430. Thus, based on the notifying parties own submission, their combined market shares in the upstream market would not lead to an affected market in long-life milk on the retail or OOH level as they are well below [10-20]*%.

7.2.3.2. Conclusion on the competitive assessment

431. For the reasons stated, it can be concluded that the concentration is not likely to lead to a significant impediment of effective competition as regards the market for long-life milk covering Belgium, Germany and the Netherlands regardless of whether this market should be segmented according to the distribution channel.

7.3. Organic fresh basic dairy products

7.3.1. Relevant Product Markets

432. Campina and Friesland Foods are both active in the production and sale of organic fresh basic dairy products based on organic raw milk. Their product portfolio consists of fresh milk, fresh buttermilk, plain yoghurt and fresh custard, which they sell to retailers, OOH and to organic foodstores and industrial users (Campina only).

433. Organic fresh basic dairy products are regulated by Regulation (EC) No 2092-91 (230) and are distinguished from other dairy products because they are produced from organic raw milk. (231)

434. A further subdivision could be made for biodynamic products which are produced from biodynamic raw milk. Biodynamic products are produced under even stricter rules as organic products. (232) A worldwide quality mark "Demeter" exists for biodynamic products. In the Netherlands organic products are labelled "EKO".

435. The NMa indicated in merger case (233) that biodynamic products might form a distinct product market. Following this reasoning, the notifying parties submit that biodynamic products from a distinct product market within the market for organic products. However, since Friesland Foods is not active on a potential market for biodynamic products and its value is rather small, the precise question whether biodynamic products from a separate product market can be left open, while the Decision focuses on the organic fresh basic dairy segment.

7.3.1.1. Organic fresh milk, organic buttermilk, organic plain yoghurt and organic custard each form a single product market

7.3.1.1.1. Product market definition proposed by the notifying parties

436. The notifying parties submit (234) that the production process for organic fresh diary products largely follows the production process for non-organic products as described in section 7.1.1. In particular, supply-side substitutability between the four different organic products (fresh milk, fresh buttermilk, plain yoghurt and custard) would exist, while demand-side substitutability would be limited.

437. Therefore, the notifying parties propose to define one single relevant product market covering organic fresh milk, organic fresh buttermilk, organic plain yoghurt and organic custard.

7.3.1.1.2. Assessment of the Commission

438. The result of the market investigation (235) with respect to demand-side substitutability showed, however, that there is a lack of substitution. Customers would not switch between custard, fresh milk, plain yoghurt and fresh buttermilk because of the different consumption patterns, taste and thickness of these products.

439. The replies of competitors (236) concerning supply-side substitutability have been mixed. While some competitors seem to be focused on particular products like organic fresh milk or organic yoghurt, others produce the full product range of the organic fresh basic dairy portfolio and indicated flexibility to switch among them.

440. On the basis of these replies it must be considered that the effects of supply-side substitutability would not be "equivalent to those of demand substitution in terms of effectiveness and immediacy" (237) as required by the Relevant Market Notice and therefore it is concluded that based on the lack of demand-side substitutability, organic fresh milk, organic fresh buttermilk, organic plain yoghurt and organic custard each form separate relevant product markets.

7.3.1.2. Private label and supplier brands belong to the same relevant upstream market.

7.3.1.2.1. Product market definition proposed by the notifying parties

441. Similar to fresh basic dairy products, the notifying parties agree that for organic fresh basic dairy a vertical segmentation can be made between an upstream market for the sourcing of products by retailers and OOH wholesalers serving hotels, restaurant, catering services and filling stations) and a downstream market for the sale of foodstuffs by retailers to consumers for two reasons.

442. With respect to the question if one should distinguish the procurement of private label products and the procurement of branded products by retailers (and OOH wholesalers) as two neighbouring, but separate markets, the notifying parties argue that in the context of the markets for organic fresh basic dairy products both Campina and Friesland Foods are under constant pressure from private label products. This is exemplified by the fact that the market is more and more covered by private label products. (238) Moreover, all organic fresh basic dairy products require a quality mark - the EKO label - guaranteeing the organic origin of the products, which according to the notifying parties would be more important than any brand considerations.

443. Thus, private label and branded products would belong to the same relevant product market at the upstream level according to the notifying parties.

7.3.1.2.2. Assessment of the Commission

444. As explained in section 7.1.1.2 whether private labels and supplier brands belong to the same product market upstream depends on:

(a) whether both types of brands, in general, compete closely with each other from the perspective of the end-customer; and

(b) the extent to which upstream suppliers of private label and/or brands as well as the purchasing retailers, take into account in their negotiations upstream the competitive pressure that private labels and supplier brands mutually exert on each other at consumer level.

445. For all organic fresh basic dairy products, there is significant evidence suggesting that private label and supplier brands compete in the downstream market and exert a competitive constraint on each other, sufficiently relevant for the purposes of market delineation.

446. In their responses to the market investigation, the majority of customers and competitors confirmed that private label and branded products compete at the retail level and that the quality of private label products is to a large extent similar to branded products (239).

447. In addition, on the downstream market for organic fresh basic dairy products the share of private label was 46.7% in 2007 in value terms and has increased by 4.3 percentage points over the past two years. (240)

448. Packaging for private labels and supplier brands are similar within the same market and is dominated for both types of brands by the 1 litre packaging format. In addition, there is almost no differentiation between supplier brands and private labels across product categories.

449. These points indicate that there is a sufficient degree of competitive interaction between supplier brands and private labels for the different organic fresh basic dairy products for both to be considered as belonging to the same relevant product market at retail level.

450. Looking at the upstream level the evidence has been mixed. First, customers and Competitors (241) confirmed that the procurement of branded products follows a different procedure than the one for private label products. In the case of branded products suppliers and retailers agree on the gross price, discounts, listing fees, and promotions in bilateral negotiation. For private labels, a tender procedure is usually used, followed by a selection of preferred suppliers which agree with the retailer on a net price in final negotiations.

451. This in itself would not be decisive if the same suppliers were active in both segments. The market presence of competitors and of the notifying parties hardly varies when it comes to branded or private label organic fresh basic dairy products life milk. According to the notifying parties' estimates, confirmed during the market investigation, almost all competitors like Dutch companies Vecozuivel and Weerribben, the German supplier Sobbeke or the Belgian producer MIK, supply private labels as well as branded products in the Netherlands.

452. Given this symmetric presence in both segments and the growing share of private labels in the overall market suppliers cannot be expected to ignore the competitive pressure that private labels exert on their brands.

453. Also, retailers in the Netherlands have either private labels in addition to the brands offered by Campina, Friesland Foods, Vecozuivel or MIK or consider private labels as an alternative to branded products. (242) Some customers view the EKO brand as sufficient and indicated that no additional producer brand would be necessary. (243) Thus, retailers take into account the presence of private label supply when procuring branded products.

454. Taken together, all these elements, in particular the growing importance of private labels in relation to branded products, the same set of suppliers of both private label and brands and the competitive interaction between them that affects the negotiations upstream, allow for the conclusion that private labels and branded products belong to the same product market upstream.

7.3.1.3. No separation has to be made according to distribution channel

7.3.1.3.1. Product market definition proposed by the notifying parties

455. According to the notifying parties, organic fresh basic dairy products are sold to consumers through three sales channels: retail, organic food stores and OOH. The notifying parties submit that all three segments would compete for the same customers; packaging, brands as well as logistics would be largely identical across segments.

456. Therefore the notifying parties propose that no distinction should be made according to distribution channels.

7.3.1.3.2. Assessment of the Commission

457. Compared to the replies in the fresh basic dairy section, the Commission has received less feedback during the market investigation indicating significant differences between the retail and the OOH segment for organic fresh basic dairy products.

458. Despite some differences in packaging between the OOH segment and the retail and organic food store segment, the majority of products seem to be sold in the 1 litre gable top format. Moreover, and differently from the conventional fresh basic dairy segment, the main suppliers for the three segments are the same and are therefore able to switch between the channels if prices were to diverge significantly. These companies - along with the notifying parties Vecozuivel and Weerribben - are also able to offer more than one organic fresh basic dairy product, allowing the customer to source a complete portfolio from one party.

459. Based on these elements, it is concluded that no separation according to the distribution channel is necessary.

7.3.1.4. Conclusion on relevant product market

460. On the basis of these elements, it is concluded that there are separate relevant product markets for organic fresh milk, organic fresh buttermilk, organic plain yoghurt and organic custard. No separations of private label/branded products or according to the distribution channel is necessary.

7.3.2. Relevant Geographic Market

7.3.2.1. The relevant geographic market for organic fresh basic dairy products is national in scope

7.3.2.1.1. Relevant geographic market proposed by the notifying parties

461. The notifying parties submit (244) that the geographical dimension of the upstream markets for the supply of the different organic fresh basic dairy products is national since these products are hardly traded between Member States and national markets are mainly supplied by local producers.

7.3.2.1.2. Assessment of the Commission

462. During the market investigation, customers confirmed (245) that they mainly source within the Netherlands and only occasionally source from neighbouring Member States and consider the market to be national in scope.

463. This was mirrored by the response of competitors (246) who indicated that they sell most of their produce in their own Member State.

7.3.2.1.3. Conclusion on the relevant geographic markets

464. Based on these elements it is concluded that the relevant geographic markets for organic fresh milk, organic fresh buttermilk, organic plain yoghurt and organic custard are national in scope.

7.3.3. Competitive Assessment

465. The notifying parties estimate that the size of the market for organic fresh basic dairy products on the upstream level in the Netherlands was EUR 38 027 000 in 2007 with a volume of 49 089 000 litres, of which 48.5% was sold as private labels and 51.5% as branded products. The overall market experienced a cumulative average growth rate of 11.6% between 2005 and 2007, which was significantly higher in the private label segment (16.8%).

7.3.3.1. Proposed transaction would not result in non-coordinated effects in the different markets for organic fresh basic dairy products

466. During the market investigation the Commission received few complaints from customers indicating that the merger might lead to higher prices. Most customers as and competitors that responded were not expecting a significant impact of the proposed transaction on the markets. A number of factors indicative of significant non-coordinated effects are not present in organic fresh basic dairy markets.

7.3.3.1.1. The notifying parties have lost market shares over time

467. As can be seen in Table 7-24, the notifying parties had a combined market share in the organic fresh basic dairy segment of [70-80]*% in 2007, followed by Vecozuivel ([10-20]*%), MIK ([5-10]*%) and Weerribben ([0-5]*%). (247)

468. Despite this strong position, the notifying parties have argued that they would not be in a position to increase prices post-merger because in this growing market expansion and entry would be easy. Notably, market shares are poor indicators of market power at this stage of market development because the award of one contract could change the picture significantly.

469. In this respect, the notifying parties have provided estimates and actual sales data for 2008 showing a significant decline in the overall sales for both companies and an expected market share of [50-60]*%. The main reason for this loss of almost [10-20]*- percentage points was the award of a contract by one important retailer to Vecozuivel, a competitor, at the end of 2007, boosting its market share to around [40-50]*%.

Table 7-24: Sourcing of organic basic dairy, total market, The Netherlands - Source: Form CO.

470. During the market investigation this loss of a contract and its importance was confirmed by the retailer in question, who explained (248) that "as a result of the growing presence of Vecozuivel, [...] takes the view that the proposed transaction will not significantly influence [...]. Vecozuivel is a viable alternative supplier to the merged entity for organic fresh dairy products."

471. Because the small size of the market and its significant growth, the shift of a large contract has already had significant implications on market shares. Therefore, it is concluded that the high market shares in 2007 are not a strong indication of market power, in light of the changing market conditions.

7.3.3.1.2. Customers have possibilities of switching suppliers

472. In addition, customers informed the Commission during the market investigation that they have the option to switch to other suppliers. Vecozuivel and Weerribben were mentioned as alternative suppliers to the notifying parties, who would be able to supply either brands or private label products.

473. One retailer explained (249) that "the current players might also grow proportionally. Especially Vecozuivel is growing at the moment because buyers might already anticipate the contemplated merger between Friesland Foods and Campina and move away volume from Friesland and Campina". Another one (250) mentioned that he "knows of three companies that have already entered the market".

474. This view has also been confirmed by competitors (251) who indicated that the organic fresh basic dairy markets would continue to grow and that the growth would also benefit smaller competitors.

7.3.3.1.3. Competitors could increase supply if prices increase

475. The market investigation also revealed that current suppliers would be able to increase supply in the case of a price increase as they have additional capacity at their disposal. Importantly, they do not experience problems with access to organic raw milk.

476. One competitor elaborated (252) that spare capacity is currently available and, in addition, new capacity would become available in the first quarter of 2009. Another competitor indicated (253) his intention to grow significantly in the future. Most competitors (254) also confirmed that if they were to expand, organic raw milk could be sourced from neighbouring Member States if necessary.

7.3.3.1.4. Customers and competitors expect entry to occur in the future

477. Customers and competitors do not only anticipate (255) the expansion of existing competitors, but also the entry of new competitors. This is because organic fresh basic dairy products represent a high margin product, which could also be transported over longer distances.

478. One retailer noted (256) that "the growing market is an opportunity for smaller companies to increase market share because it is an easy entry for these companies. The entry barrier is a lot lower, the quantities are a lot smaller than regular dairy products" A similar view was expressed by a second retailer explaining (257) that "Contrary to regular milk, organic milk travels many foodmiles within Europe as the impact of transport costs is much lower due to fact that the price of this product is higher. Hence it is profitable to transport organic fresh basic dairy products over longer distances."

479. One customer mentioned as a recent example that a foreign competitor participated in a tender for private label products to be supplied to the Netherlands. (258)

7.3.3.2. Conclusion on the competitive assessment

480. For these reasons - in light of the responses in the market investigation and in particular the possibility for customers to switch to existing suppliers as well as the possibility of entry -, it can be concluded that the concentration would not significantly impede effective competition as regards the markets for organic fresh milk, organic fresh buttermilk, organic plain yoghurt and organic custard in the Netherlands.

8. CHEESE

8.1. Introduction

481. Friesland Foods and Campina are both active in the production of Dutch-type cheese. Both produce Gouda and Maasdam. In addition, Friesland Foods produces Edam.

482. Gouda, Maasdam and Edam are made of cow's milk and have a fat content of 48%, 45% and 40% respectively. According to the notifying parties, Gouda has a mild and creamy taste while Maasdam is sweet and nutty and Edam has a mild taste. Gouda, Maasdam and Edam (including their varieties with low salt, low fat etc.) are typical Dutch cheeses and henceforth referred to as "Dutch-type cheese".

483. The production of cheese is milk-intensive, 1 kg of cheese requires approximately 9 kg of raw milk. As demonstrated in section 6.1.1, 50% of all raw milk procured in the Netherlands is used for the production of cheese.

484. The notifying parties submit that Dutch-type cheese belongs to the category "hard/semi-hard cheeses" and that nearly all hard/semi-hard cheese production in the Netherlands is dedicated to producing Gouda, Maasdam, Edam and varieties. (259) The notifying parties are the largest producers of Dutch-type cheese (Friesland Foods [...]* tonnes, Campina [...]* tones (260)) and account together for more than [70-80]*% of the total cheese production in the Netherlands. Other producers of Dutch-type cheese in the Netherlands and their production capacities estimated by the notifying parties are DOC ([...]* tonnes), Fromagerie Bel ([...]* tonnes), Cono ([...]* tonnes) and Rouveen ([...]* tonnes). (261) These smaller players focus on particular segments of Dutch-type cheese (as explained below in section 8.4.1.2).

485. One important feature of the cheese market in the Netherlands lies in the activities of specialised cheese wholesalers who have been active in the Netherlands for centuries and play an intermediary role in the Dutch cheese market between the production level and the downstream level: (262)

486. Many wholesalers buy Dutch-type cheese which is 15-day-old (and thus not yet ready for consumption) from the notifying parties and other cheese producers in order to further mature it within their own maturing facilities. In addition, some wholesalers are specialised in cutting, slicing and packaging services. After maturing and/or slicing and packaging the cheese, wholesalers sell it to downstream distribution channels in the Netherlands (to retail, OOH or industry customers; as explained in sections 8.2.2.7 and 8.2.2.8, the competitive strength of wholesalers differs across these channels) or export it. In doing so, they often focus on supplying smaller volumes targeted at the specific needs of their customers. (263) This distinguishes specialised cheese wholesalers from the notifying parties, who can be characterised as bulk producers that concentrate on the large-scale production of cheese. (264)

487. A characteristic of the Dutch-type cheese market in the Netherlands is that production significantly exceeds consumption, resulting in considerable exports of Dutch-type cheese. The notifying parties estimate that approximately two thirds of the Dutch production of Dutch-type cheese is exported. (265) The notifying parties themselves export [50-60]*% of their production (266), and specialised cheese wholesalers are also active in exporting matured cheese. It can reasonably be assumed that exports by specialised cheese wholesalers of cheese produced by the notifying parties amount to approximately [10-20]*% of the notifying parties' total production (see section 8.4.1.8); it follows that approximately [60-70%]* of the notifying parties' total production is directly and indirectly exported. In contrast, imports of Dutch-type cheese into the Netherlands are minimal. (267)

488. Dutch cheese producers therefore have two channels to the downstream markets in the Netherlands and abroad: (i) through direct sales and (ii) through specialised cheese wholesalers.

489. The flow of Dutch-type cheese in the Netherlands has been illustrated by the notifying parties as follows: (268)

Figure 8-1: Flow of Dutch-type cheese in the Netherlands. Source: Calculations of the Notifying parties*

490. Dutch-type cheese can be either naturally matured cheese (henceforth "nature cheese") or rindless cheese. Nature cheese has a rind and is capable of maturing for up to one year and longer. Rindless cheese is wrapped in plastic foil when it is young and does not mature any further. (269)

491. While rindless Dutch-type cheese has been characterised by the notifying parties as a "true commodity", nature cheese is often made on the basis of specific starter cultures and recipes. The notifying parties submit that most nature cheese they produced is based on the requirements of their customers. (270)

492. Despite such differences, the beginning of the production process for both nature and rindless cheese is very similar: Raw milk is pre-treated (standardised or pasteurised) and the water in the milk is partly removed to increase the proportion of dry matter in the milk. The milk is thickened with the help of rennet, and a starter culture is added to help shape the flavour of the milk. The process produces curds and a mass of solid cheese lumps with whey as by-product. After processing, the cheese is put into a brine bath. (271)

493. After the cheese is removed from the brine bath, however, the production processes of nature and rindless cheese differ considerably.

494. At the last stage of the production, rindless cheese is packed in plastic bags (foil) immediately after it has left the brine bath and it is stored in a regular cooling environment (4-6 degrees C) for about 4 weeks. (272) It is normally made available to the consumers at 28 days maturity. (273)

495. In contrast, the last stage of production of nature cheese entails coating of the rind, where it is dipped in paraffin to ripen it after it has left the brine bath. It is matured in special maturing facilities where temperature and handling are carefully monitored and the cheese needs to be turned regularly and recoated. (274)

496. All Dutch-type nature cheese produced in the Netherlands is matured for at least four weeks before it is ready for consumption; often, it is matured for significantly longer periods. A significant proportion of Dutch-type cheese purchased by Dutch consumers is matured nature cheese of different ages. (275)

497. While all Dutch-type nature cheeses have their own traditional shapes (Gouda: round wheel shape, Maasdam: round bold shape and Edam: ball shape), they can also be produced as rectangular blocks (referred to as "Euro shape cheese"). These blocks are easier to handle in the supply chain and are better suited for cutting/slicing and pre-packing equipment. Rindless cheese in the Netherlands is always produced in the form of rectangular blocks.

498. As will be explained in more detail in section 8.3.2.1, Dutch household consumers have historically had a strong preference for nature cheese produced in the Netherlands (almost all Dutch-type cheese bought in supermarkets and other modern types of retail (276) in the Netherlands is nature cheese and almost none of this nature cheese is imported). Similarly, almost all specialised cheese wholesalers require nature cheese produced in the Netherlands for their maturing activities.

8.2. Relevant product markets

8.2.1. Product market delineation proposed by the notifying parties

499. The notifying parties submit in the Form CO and later submissions that (i) the relevant product market includes Dutch-type cheese and other hard/semi-hard cheese types (such as Emmenthal, Tilsit, Cheddar, Manchego and Danbo) and that no distinctions should be made between (ii) nature and rindless cheese, (iii) variations in terms of age, fat or salt content and so forth, (iv) producer branded cheese, wholesale branded cheese, private label cheese and unbranded cheese, and (v) sales to specialised cheese wholesalers as well as to the retail, OOH and industrial channels.

8.2.2. Assessment of the Commission

8.2.2.1. Dutch-type cheese vs. other hard/semi-hard cheeses

500. The Commission has previously distinguished between different categories of cheeses such as "hard/semi hard cheese" (which includes Gouda and Edam), "soft cheese" and "fresh cheese". It is also clear from previous Commission practice that a further subdivision of these categories into different types of cheese might be necessary in Member States which have a strong tradition for a particular cheese type. (277)

501. The notifying parties claim that the relevant product market includes Dutch-type cheese and other hard/semi-hard cheese types (including Emmenthal, Tilsit, Cheddar, Manchego and Danbo). In particular, they explain that the typical consumer in most European markets is a socalled "repertoire buyer" who considers hard/semi-hard cheeses as substitutable for one another. Furthermore, the basic production processes of the different hard/semi-hard cheeses (including fat content varieties) are very similar.

502. The market investigation has not confirmed the market definition proposed by the notifying parties. A clear majority of cheese producers (278), companies active in modern types of retail (279) and specialised cheese wholesalers (280) in the Netherlands considers that Dutch-type cheese is not in competition with other hard/semi-hard cheeses in particular due to consumer preferences for Dutch-type cheeses and strong differences in taste and price. (281) This is also evidenced by the fact that the consumption of other hard/semi-hard cheeses such as Emmenthal and Cheddar in the Netherlands is very low. (282) It is therefore concluded that Dutch-type cheese is not substitutable on the demand-side with other hard/semi-hard cheeses. This conclusion is reinforced by the fact that considerable demand-side differences exist even within the category of Dutch-type cheese (Gouda, Maasdam and Edam, see section 8.2.2.2).

503. The information gathered during the market investigation also made clear that supply-side substitutability between Dutch-type cheese and other hard/semi-hard cheeses within the Netherlands is not sufficiently immediate and effective to satisfy the requirements under the Relevant Market Notice. One Dutch cheese producer estimated that such a change in production would be costly and that it would take half a year for a producer of Gouda cheese to start producing Emmenthal, Tilsit or Cheddar. (283) Also, a clear majority of specialised cheese wholesalers replied that considerable time and investment would be necessary for such a switch in production or indicated they were unable to provide an estimate on the costs and time required. (284) These responses show that switching production in the Netherlands from Dutch-type cheese to other hard/semi-hard cheeses would require significant time, investments and knowhow and would therefore not be sufficiently immediate and effective.

504. It is therefore concluded that non-Dutch-type hard/semi-hard cheeses are not in the same product market as Dutch-type cheese.

8.2.2.2. Gouda vs. Maasdam vs. Edam

505. Regarding Dutch-type cheese in particular, great importance was attached in previous decisions to traditions and consumer preferences for a particular cheese type that vary from Member State to Member State. (285) Such strong traditions and preferences exist in the Netherlands as far as Gouda, Maasdam and Edam are concerned. In particular, Dutch consumers have a strong preference for Gouda cheese, as is evidenced by the fact that most Dutch-type cheese sold in modern types of retail in the Netherlands is Gouda cheese. (286) Furthermore, even though Gouda, Maasdam and Edam are similar in price, it is clear from the market investigation that Dutch consumers would not switch easily between Gouda, Maasdam and Edam due to differences in taste and consumption patterns. (287)

506. The phase II market investigation also suggests that supply-side substitutability between the different Dutch-type cheeses is limited. It is true that some producers in the Netherlands produce Gouda and/or Maasdam and/or Edam on the same production line and/or have stated that they could switch in a timely manner (often within the regularly scheduled cleaning of production lines) and without additional costs. (288) However, not all cheese producers in the Netherlands (including Campina) produce all of the three different cheese types. (289) Furthermore, significant investments in terms of cost and time appear to be necessary to enable a cheese production line to produce a shape currently not produced, for example to enable a production line producing rectangular shape to also produce the traditional shape of a Dutch-type cheese. For instance, significant investment and time is necessary for a cheese producer currently not producing traditional Edam balls in order to switch production to such Edam balls; currently, Friesland Foods is the only producer of Edam balls in the Netherlands. (290)

507. In view of the above, supply-side substitution between Gouda, Maasdam and Edam in the Netherlands is unlikely to be sufficiently immediate and effective to satisfy the requirements under the Relevant Market Notice. Nevertheless, the precise market definition can ultimately be left open as it will not have a material impact on the competitive assessment.

8.2.2.3. Nature vs. rindless Dutch-type cheese

508. The notifying parties argue that nature and rindless Dutch-type cheese belong to the same market. In particular, they explain that nature and rindless cheese are substitutable in most applications and that switching between nature and rindless cheese is fairly easy due to the similarity of the production processes. The cost of packaging and coating equipment required for the production of nature cheese is very low. According to the notifying parties, the only really important difference is the storage and ripening facilities required for nature cheese which could be largely outsourced to specialised cheese wholesalers and subcontractors.

509. The market investigation has shown that a clear majority of specialised cheese wholesalers in the Netherlands considers rindless cheese to be of a lower quality than nature cheese, due to differences in taste, structure and colour. (291)

510. Specialised cheese wholesalers emphasised that nature cheese is suitable for ripening (and thus sold in various ages and forms) whereas rindless cheese is not (and thus normally sold at an age of 4 weeks and only in packed and sliced form). (292) Almost all specialised cheese wholesalers (293) in the Netherlands agree that consumers would not switch from nature cheese to rindless cheese if the price of nature cheese increased by 5-10%. Several cheese producers in the Netherlands who produce both nature and rindless cheese agreed with this statement. (294)

511. Similarly, most companies active in modern types of retail in the Netherlands stated that consumers would not switch from nature to rindless cheese if the price increased by 5% to 10% due to preferences of Dutch consumers for nature cheese and differences in taste ("nature cheese has the typical Dutch taste", "rindless cheese like gum"). (295)

512. The differences between nature and rindless cheese are also reflected in different consumption patterns: Dutch household customers have a clear preference for nature cheese which is evidenced by the fact that approximately 93% of all Dutch-type cheese sold in modern types of retail in 2007 was nature cheese. (296) In contrast, industrial customers mainly use rindless cheese since taste is less an issue in industrial applications (for example for pizza), rindless cheese is generally cheaper than nature cheese and no rind is left over after the cheese has been sliced or grated. (297)

513. On 18 August 2008, the notifying parties submitted price correlations between nature and rindless cheese in support of their argument that both types of cheese belong to the same market. (298) Although these prices exhibit a certain level of long run co-movement, it should be recalled that milk is the most important ingredient for cheese and therefore a significant cost component. The prices of both nature and rindless cheese therefore necessarily reflect changes in the price of milk, irrespective of whether or not these cheese types are substitutable on the demand or supply side. In addition, co-movement of variables can also be caused by changes in tastes or a series of unrelated variables which affect demand for different products in the same way but do not necessarily imply that the products are in the same market. In such cases it is generally erroneous to infer causal links or to conclude that the products in question are in the same market.

514. On 17 October 2008, the notifying parties submitted new price correlations using the notifying parties' output data which has been corrected for the raw milk input. However, these correlations display (i) prices of nature cheese in the Netherlands vs. prices of nature cheese in Germany and (ii) prices of nature cheese in the Netherlands vs. prices of rindless cheese in Germany (and not prices of nature cheese in the Netherlands vs. prices of rindless cheese in the Netherlands). These price correlations are therefore not reliable evidence for the question of substitutability between nature and rindless cheese in the Netherlands.

515. Moreover the notifying parties agree in principle that co-movement of prices cannot be considered as decisive evidence for market definition purposes. (299) Even in input-corrected price correlations co-movements can be caused by factors other than competitive constraints; further information is therefore required as regards the definitions of product categories, transactions included and the methods of aggregation. Price correlations should also be balanced against other evidence such as the qualitative evidence gathered in the market investigation which clearly suggests that demand-side substitutability between nature and rindless Dutch-type cheese in the Netherlands is limited considerably by consumer preferences.

516. Supply side substitution is taken into account by the Commission for the purpose of defining relevant markets if the supply substitution effect is sufficiently immediate and effective. (300) That is, if "suppliers are able to switch their production to the relevant products and market them in the short term without incurring significant additional costs or risks in response to small and permanent price changes", the resulting "disciplinary impact on the competitive behaviour of the companies involved" is regarded as equivalent to the restraining effects of demand substitutability. In these specific situations only can supply side substitutability constitute a basis for delineating the relevant product market.

517. The Relevant Markets Notice further explains that supply-side substitutability is not sufficiently high for the purpose of market definition if it entailed significant additional costs or risks, strategic decisions or time delays. (301)

518. The market investigation shows that a switching between nature and rindless cheese (and vice versa) can trigger significant additional costs, time delays and strategic decisions for several reasons.

519. First of all, rindless cheese has an outspoken commodity nature and is sold mainly on the basis of day prices while nature cheese produced by the notifying parties is very often based on different recipes and/or requirements of the relevant customer with contractual conditions that are fixed for longer periods. (302) Also, the notifying parties' biggest competitor - DOC - only produces nature cheese on order. (303) Furthermore, DOC does not sell directly to downstream customers (such as modern types of retail, OOH wholesalers or industrial customers) but sells exclusively to specialised cheese wholesalers. This lack of direct interaction with the downstream markets further affects the immediacy of a switch of production between nature and rindless cheese. (304)

520. In addition, a switching from rindless to nature cheese (and vice versa) will cause additional cost and time delays both in the case of plants which currently only produce rindless cheese (for example coating installation, warehousing) and plants which currently only produce nature cheese (for example foil packing equipment). Significant further costs would arise if a production line producing only rectangular rindless cheese had to be modified to produce nature cheese in traditional form. (305) It is clear from the market investigation that the notifying parties and their competitors in the Netherlands would incur at least some of these costs since not all of their production lines are able to switch between the two cheese types; Dutch cheese producers usually specialise in one of the two types. (306)

521. The production processes of nature and rindless cheese are very different once the cheese blocks have left the brine bath: while rindless cheese is packed in a plastic film and can immediately be stored in a container, nature cheese is matured in special maturing facilities where it needs to be turned regularly and recoated and the temperature needs to be carefully monitored. The notifying parties submit that cheese producers currently active in the production of rindless cheese could easily switch to the production of nature cheese by outsourcing storage and ripening to third parties. According to the notifying parties specialised cheese wholesalers in the Netherlands have available maturing capacity of approximately [10-20]*%. (307)

522. However, outsourcing is a strategic decision and it is complicated by the fact that the transport of cheese younger than 15 days can affect the consistency of the cheese which can give rise to difficulties at the stage of ripening, slicing and packing of the cheese. (308) Accordingly, the notifying parties acknowledge that transporting nature cheese younger than 15-day-old is not standard practice but rather an exceptional solution. (309)

523. It is therefore unlikely that a possible supply-side substitution effect in the Netherlands would be sufficiently immediate and effective to satisfy the requirements under the Relevant Market Notice. However, the precise market definition can ultimately be left open as it will not have a material impact on the competitive assessment.

8.2.2.4. Branded vs. private label/unbranded

524. The notifying parties submit that no distinction should be made between producer branded cheese, wholesale branded cheese, private label cheese and unbranded cheese as the overwhelming majority of Dutch-type cheese sold downstream is unbranded or sold under retail or wholesale brands.

525. Indeed, only a small part (310) of all sales of Dutch-type cheeses at retail level are producerbranded or wholesaler-branded while producers and specialised cheese wholesalers often focus both on the branded and on the private label/unbranded segment. Most companies active in modern types of retail in the Netherlands participating in the market investigation agree that branded products and private label products compete on the shelves. Most of these respondents have higher profit margins for private label cheese. (311)

526. At the upstream procurement level, several companies active in modern types of retail organise calls for tenders for private label/unbranded products while calls for tenders are not organised for branded products. (312)

527. Although such differences at procurement level indicate different competition conditions for branded products and private label products (313), supermarket chains in the Netherlands are also procuring private label/unbranded cheese on the basis of bilateral negotiations. (314)

528. Furthermore, in assessing the competitive interaction between branded and private label/unbranded cheese upstream, it must be taken into account that many of the same players are offering both branded and private label/unbranded products. (315) Specialised cheese wholesalers play a particular role in that regard as they are able to switch relatively easily between the supply of branded, private label or unbranded cheese (they can buy unbranded 15-day old cheese and sell it as their own branded matured cheese). The competitive pressure exercised by branded and private label/unbranded cheese on each other will therefore be significant for a large part of the upstream procurement market in particular due to the presence of specialised cheese wholesalers and the low volume of branded cheese sold at downstream level.

529. It is therefore concluded that branded, private label and unbranded Dutch-type cheese belong to the same product market.

8.2.2.5. Specialised cheese wholesalers

530. The notifying parties claim that no separate markets should be defined for sales to specialised cheese wholesalers, retailers, OOH wholesalers and industrial customers. The notifying parties admit that a separate product market for the sourcing of cheese by wholesalers could be defined, but argue that such an approach fails to take into account that wholesalers act as customers, suppliers and competitors of cheese producers. (316)

531. Specialised cheese wholesalers form an intermediate level between the production and downstream levels. They mature and/or slice/pack and trade Dutch-type cheese. Approximately [60-70]*% of the cheese sourced by specialised cheese wholesalers is nature cheese while approximately [30-40]*% is rindless cheese. (317) 15-day-old cheese bought by specialised cheese wholesalers is almost exclusively nature cheese which is then further ripened by them. (318)

532. Specialised cheese wholesalers do not sell cheese directly to the final consumer level. This clearly distinguishes them from the players active at the downstream level (retailers, OOH wholesalers, industrial users). The market where cheese producers meet the specialised cheese wholesalers is therefore a specific market that must also be assessed separately despite its interaction with the downstream markets.

533. Specialised cheese wholesalers often buy 15-day-old nature cheese, a semi-finished product which is not yet ready for consumption but is further ripened by them for periods up to one year or longer. (319) This business model allows specialised cheese wholesalers to control the maturing process (quality) and the overall age structure of their products. The specialised cheese wholesalers normally sell matured cheese to downstream distribution channels in the Netherlands or export it.

534. In view of this intermediary position of specialised cheese wholesalers between the production level and the downstream level, it is appropriate to define a separate market for the sale of Dutch-type cheese to specialised cheese wholesalers. The interaction between this market and the downstream markets is taken into account in the competitive assessment.

8.2.2.6. 15-day-old nature Dutch-type cheese vs. other nature Dutch-type cheese

535. There is a question as to whether the market for sales to specialised cheese wholesalers needs to be divided between 15-day-old nature cheese and nature Dutch-type cheese older than 15 days. The notifying parties submit that Dutch producers sell a significant proportion of their 15-day-old nature cheese to wholesalers but argue that no such segmentation is necessary as is evidenced by price correlations and the fact that 15-day-old cheese is not consumed as such but sold to specialised cheese wholesalers solely for the purpose of maturing. (320)

536. The notifying parties' view concerning the available evidence cannot be supported. As regards the price correlations submitted by the notifying parties, the same observations can be made as those made in the context of nature and rindless cheese. Although these prices exhibit a certain level of long run co-movement, they necessarily reflect changes in the price of milk. In addition, such a co-movement of variables can also be caused by other variables or a series of unrelated variables which affect demand for different products in the same way but do not necessarily imply that the products are in the same market. It follows that such price correlations are not in themselves evidence that 15-day-old cheese and other cheese are in the same market.

537. In addition, 15-day-old cheese constitutes a semi-finished product whose maturing is a fundamental part of the commercial activity of specialised cheese wholesalers and does therefore not satisfy the same demand as matured cheese. This is illustrated by the statement of a specialised cheese wholesaler according to whom "maturing is a form of production of better tasting cheeses". (321) Moreover, it is unclear to what extent supply-side substitutability exists between 15-day-old cheese and matured cheese; the notifying parties have not explicitly claimed that supply-side substitutability would exist despite the Commission's request that they elaborate on their view further regarding the product market definition of 15-day-old cheese. (322)

538. Furthermore, as observed in relation to the distinction between nature and rindless cheese, Dutch producers need sufficient ripening capacities to significantly increase their production of 15-day-old cheese. In the absence of sufficient maturing capacity, either a strategic decision to outsource the maturing process to specialised cheese wholesalers needs to be made or maturing facilities need to be built by the producer which will cause significant additional cost and time delays. (323)

539. However, the question whether a further segmentation into 15-day-old nature cheese and nature cheese older than 15 days is necessary can ultimately be left open as this question will not have a material impact on the competitive assessment.

8.2.2.7. Retail vs. OOH vs. industry

540. The notifying parties claim that no different procurement markets should be defined for the supply of Dutch-type cheese to the retail, OOH and industry levels in the Netherlands since specialised cheese wholesalers supply all segments in direct competition with producers and arbitrage between the different segments.

541. The Commission has previously distinguished between the sale of daily consumer goods at retail level and the sale of daily consumer goods at the wholesale level. (324)

542. As regards the procurement of daily consumer goods, the Commission has previously found that separate markets exist for different categories of products and that these categories can be further segmented according to the different downstream distribution channels based on the differences in packaging, logistics and know-how between these channels. (325)

543. The market investigation has not confirmed the market definition proposed by the notifying parties but has revealed that the sale of Dutch-type cheese at the retail level on the one hand and the sale of Dutch-type cheese at the OOH and industry levels on the other hand fall within different product markets.

544. Indeed, packaging clearly differs across these channels. Cheese destined for the retail level is typically packed in small consumer units of maximum 400g or as whole wheels which are cut at location. Cheese delivered to OOH wholesalers is typically packed in bigger units of 500g to 2kg (or very small units of 1 or 2 slices per pack). Industrial customers are supplied with bigger bulk units such as 5 kg, 10kg or 15 kg blocks. (326) It was also emphasised in the market investigation that a higher standard of packaging is required at retail level than in the other channels. (327)

545. The OOH channel has been distinguished from the retail channel on the basis of higher-prices and the fact that is more labour-intensive for suppliers. (328) This is in line with the fact that the order volumes of OOH wholesalers in the Netherlands are often considerably lower than order volumes at retail level (329) and that OOH customers thus appear to expect more flexibility from their suppliers. (330)

546. These differences are also reflected in different business models: while the notifying parties focus on the production of high volumes of cheese and are by far the biggest players at retail level, specialised cheese wholesalers typically focus on smaller volumes targeted at the specific needs of their customers.

547. Accordingly, the notifying parties explain that they would not be active in the OOH channel but that this channel is supplied only by specialised cheese wholesalers. It does not correspond to the notifying parties' business model to distribute relatively small batches, and the notifying parties do also not have the necessary storage/maturing and packaging facilities, fresh pack equipment, logistic network, sales force and contacts to serve the OOH channel. (331)

548. Similarly, the notifying parties explain that they are not active as suppliers of industrial customers in the Netherlands since the volumes required by such customers are too small and/or to diversified and/or sales patterns too varied. (332)

549. It is therefore concluded that the sale of Dutch-type cheese at the retail level needs to be distinguished from the sale of Dutch-type cheese to OOH wholesalers and industrial customers in the Netherlands. The interaction between these different channels is taken into account in the competitive assessment.

8.2.2.8. Modern types of retail vs. other retail outlets

550. The Commission has in its previous decision practice distinguished the sale of daily consumer goods through modern types of retail from the sale of such goods by other outlets at retail level, such as specialised shops. (333)

551. The notifying parties submit that with regard to procurement of cheese no distinction should be made between modern types of retail, specialised cheese shops, open markets and other small retail outlets since all formats are supplied by specialised cheese wholesalers and are in competition downstream when selling to the end consumer. (334)

552. However, it is to be recalled that differences in the packaging, logistics and know-how necessary for different distribution channels may lead to the definition of separate product markets for the procurement of a particular product. (335)

553. With regard to the distinction between modern types of retail on the one hand and other small retail outlets (such as specialised cheese shops, open markets, ambulant stands) on the other hand, the market investigation indeed revealed important differences in offerings, supply structures, know-how and logistics which indicate that the competitive conditions at procurement level in these two channels vary considerably.

554. First, as regards the notifying parties' argument that all formats are supplied by specialised cheese wholesalers, it is true that wholesalers satisfy the entire demand of small retail outlets and approximately 33% of the demand of modern types of retail (see section 8.4.2.1). However, cheese sold in specialised cheese shops and open markets is usually cut at location while approximately 75-80% of all cheese sold in Dutch supermarkets is pre-packed. (336) The largest part of what is offered by supermarkets therefore differs from specialised cheese shops and open markets, and specialised cheese wholesalers supplying small retail outlets will not be an alternative source of supply with regard to such products offered. Furthermore, for modern types of retail without their own cutting department (which is for example the case for Aldi and Lidl (337)), the products offered are completely different. (338)

555. Second, for reasons similar to those given in section 8.2.2.7, the notifying parties and other Dutch cheese producers do not have direct sales to specialised cheese shops, open markets and other small outlet. The notifying parties explain that the supply of such small diversified volumes rather falls into the exclusive domain of specialised cheese wholesalers due to the latter's focus on often fine-meshed supplies targeted at the customer's particular needs and also due to the fact that the notifying parties do not have the storage/maturing and packaging facilities, fresh pack equipment, logistic network, sales force and contacts necessary to distribute relatively small batches to smaller retailers, specialty cheese shops or market stands. (339) These different requirements as to know-how, packaging, logistics and marketing result in considerable differences of the supply structure of modern types of retail as compared to the supply structure of specialised cheese shops and therefore underline that both channels constitute different markets. (340)

556. Third, the notifying parties explain that only supermarket and discount chains have a centralised distribution centre ("DC") structure covering sufficient volumes in order to render direct supplies commercially attractive for cheese producers. (341) In contrast, cheese shops and open markets are usually delivered directly by specialised cheese wholesalers. (342) The logistics requirements of both channels therefore clearly differ from each other. (343)

557. Conversely, only cheese producers (and a small number of specialised cheese wholesalers who source from these cheese producers, see section 8.4.2.1) are able to supply modern types of retail with the larges volumes they require (while specialised cheese wholesalers typically focus on the more dispersed demand of specialised cheese shops and other small retail outlets). For the same reason, a level playing field between cheese producers and specialised cheese wholesalers as regards the supply of large volumes of cheese to modern types of retail does often not seem to exist. (344)

558. In view of these differences, it is appropriate to define a separate market for the sale of Dutch-type cheese to modern types of retail. Any competitive pressure emanating from specialised cheese shops, open markets and other small retail outlets in the Netherlands will be considered in the competitive assessment.

8.2.3. Conclusion on relevant product market

559. On the basis of the elements discussed above, it is concluded that separate product markets for the sale of Dutch-type cheese to specialised cheese wholesalers and to modern types of retail should be delineated. The need for a further distinction between Gouda, Maasdam and Edam, between nature and rindless cheese and between 15-day-old nature cheese and nature cheese older than 15 days can be left open as it will not have a material impact on the competitive assessment.

560. The following markets will be assessed hereinafter: (i) sale of Dutch-type cheese to specialised cheese wholesalers; (ii) sale of Dutch-type cheese to modern types of retail; (iii) sale of nature Dutch-type cheese to specialised cheese wholesalers; (iv) sale of 15-day-old nature Dutch-type cheese to specialised cheese wholesalers (v) sale of nature Dutch-type cheese to modern types of retail; (vi) sale of rindless Dutch-type cheese to specialised cheese wholesalers; and (vii) sale of rindless Dutch-type cheese to modern types of retail. Each of the markets referred to in (i) - (vii) could also be defined more narrowly as only comprising Gouda or Maasdam respectively. (345)

8.3. Relevant geographic market

8.3.1. Geographic market definition proposed by the notifying parties

561. The notifying parties submit that all upstream markets at issue are European-wide or in any event include those Member States in which Dutch-type cheeses are traded in significant volumes, that is the Netherlands, Germany, Belgium, France, Poland, Greece, Italy, the UK and Spain. In particular, they explain that the majority of the volume of Dutch-type cheese is manufactured outside the Netherlands and that two thirds of the Dutch production of Dutch-type cheese are exported where it competes directly with cheese of non-Dutch origin. Prices throughout Europe are highly transparent and strongly correlated, and transport costs are relatively low.

8.3.2. Assessment of the Commission

562. Cheese markets have previously been considered national in scope due to important differences on the demand-side and the impact of these differences on the level of procurement. (346)

563. As explained hereinafter, the market investigation in this case has not confirmed the geographic market definition proposed by the notifying parties, but rather, revealed that the markets for nature Dutch-type cheese (347) as well as the markets for Dutch-type cheese (comprising nature and rindless cheese) (348) are national in scope. In contrast, the markets comprising only rindless Dutch-type cheese (349) are wider than national and include at least the Netherlands and Germany.

8.3.2.1. Nature Dutch-type cheese

(i) Common features of all nature Dutch-type cheese markets

564. First, as noted in section 8.2.2.3 on the question of substitutability between nature and rindless cheese, price correlations between Dutch type nature cheese sold in the different Member States are not conclusive evidence for the question of geographic market definition. (350) Furthermore, it should be stressed that the average sales prices in the different Member States vary considerably. (351)

565. Second, the information gathered in the course of the market investigation clearly shows that the sale of nature Dutch-type cheese in the Netherlands has a national dimension. Indeed, imports of Dutch type nature cheese into the Netherlands are virtually non-existent. (352) Current sourcing patterns in the Netherlands are therefore clearly confined to the national level.

566. Furthermore, consumption patterns in the Netherlands differ considerably from those of neighbouring Member States and these differences at downstream level have repercussions at the sourcing level, in particular as far as sales to modern types of retail are concerned: while most cheese in Belgium and Germany appears to be consumed as young aged cheese (often without rind), and Germany in particular has been described as a Member State where predominately rindless cheese is consumed, the demand for older cheese is considerably higher in the Netherlands where Gouda is sold in all ages and variations. (353) Furthermore, the notifying parties themselves submit that traditionally many Dutch households slice lumps of cheese at home just prior to consumption while the majority of cheese sold to consumers in Germany is sold prepacked in slices. (354)

567. It is also interesting to note in this context the two applications for protection of geographical indications "PGI" for "Gouda Holland" and "Edam Holland" made by the Dutch Dairy Organisation NZO (the notifying parties are both members (355)) which are currently pending. (356) Both applications mention that Gouda and Edam nature cheese subject to this designation "Gouda Holland" and "Edam Holland" should henceforth be produced in the Netherlands from cows' milk obtained from Dutch dairy farms and that such cheese should be matured to a consumer-ready product in Dutch maturing rooms. For "Gouda Holland" the publication also refers to the mandatory use of a protected starter culture. Only the producers of "Gouda Holland" will be able to obtain these starter cultures which play an important role in the maturing and the development of the typical taste and aroma of "Gouda Holland". (357) This protection will also apply to "Edam Holland". (358) The PGI application further underlines the fact that Dutch milk is particularly suitable for producing high-quality cheese due to the geographical situation of the Netherlands, its climate and the composition of its grass.

568. It is not yet certain if, and to what extent, these PGI applications will be granted and it must be underlined that the conclusions do not rely on the PGI being granted. However, many observations would be reinforced should the applications be granted. Furthermore, the fact that these PGI applications have been lodged is a further indication that the markets for Dutch type nature cheese are national in scope as far as the Netherlands is concerned.

569. Attached to the PGI application is a consumer survey commissioned by the Dutch Dairy Council in 1998. According to this survey, 65% of Dutch consumers are willing to pay a 10% price premium for Dutch Gouda (61% for Dutch Edam) and 62% of Dutch consumers always prefer Dutch Gouda over foreign Gouda (57% for Dutch Edam). (359)

570. Furthermore, a protected designation of origin ("PDO") has already been obtained for a certain type of Gouda cheese that is produced in the province of North Holland. Due to this origin protection, only producers that are located in the province of North Holland are able to produce and sell cheese as "NH Gouda" cheese. North Holland Gouda cheese is being marketed as premium Gouda and cannot as a matter of course be obtained from abroad. Accordingly, the notifying parties define different categories of milk used for NH Gouda cheese and other cheese.

571. Finally, the presence of specialised cheese wholesalers (among whose core activities is the maturing of nature cheese) is a particular feature of the Dutch cheese market which clearly distinguishes the structure of the nature Dutch-type cheese market in the Netherlands from all other Member States that the notifying parties claim to belong to the same geographic market.

572. In view of importance of ingredients originating from the Netherlands, consumer preferences, consumption patterns, the absence of significant imports and the activities of specialised cheese wholesalers, and as will be shown in the subsections (ii) - (iv) below more specifically for sales to specialised cheese wholesalers and modern types of retail, it is concluded that the markets for nature Dutch-type cheese are national in scope. The market investigation does not suggest that a different conclusion should be reached in the case of a further segmentation of nature Dutch-type cheese into Gouda and Maasdam.

(ii) Sales of nature Dutch-type cheese to specialised cheese wholesalers

573. With regard to the sales of nature Dutch-type cheese to specialised cheese wholesalers, almost only negative answers were received from wholesalers in the market investigation as to their ability to source nature Dutch-type cheese from abroad in substitution of nature cheese sourced from the notifying parties. (360) Several wholesalers made clear that nature cheese produced in the Netherlands is a fundamental part of their ripening activity and that the quality of foreign nature cheese in terms of taste and crust does not meet the necessary standards. (361)

574. Specialised cheese wholesalers unanimously stated that Dutch consumers have a strong preference for Dutch-type cheese made in the Netherlands (as opposed to imported Dutch-type cheese) due to its superior quality of taste and texture which is based on the combination of Dutch milk, soil and climate with Dutch lactic starters/lactic cultures. (362) Accordingly, almost all specialised cheese wholesalers agree that a price increase of 5%-10% in Dutch type nature cheese made in the Netherlands would not be sufficient to influence Dutch consumers so as to switch to Dutch type nature cheese produced outside the Netherlands. (363)

575. It should also be recalled that specialised cheese wholesalers undertake particular activities distinguishing the Dutch cheese market from the cheese markets in all other Member States that the notifying parties claim belong to the same geographic market.

576. It is therefore concluded that the market for the sale of nature Dutch-type cheese to specialised cheese wholesalers is national in scope.

(iii) Sales of 15-day-old nature Dutch-type cheese to specialised cheese wholesalers

577. As the considerations in the preceding paragraphs apply a fortiori for 15-day-old nature Dutch-type cheese, it is concluded that also the market for the sale of 15-day-old nature Dutchtype cheese to specialised cheese wholesalers is national in scope.

(iv) Sales of nature Dutch-type cheese to modern types of retail

578. Similarly, almost all modern types of retail in the Netherlands consider that Dutch consumers strongly prefer Dutch-type cheese produced in the Netherlands ("Dutch-type cheese" should hereinafter in this section (iv) be understood as referring to nature Dutch-type cheese since 93% of all Dutch-type cheese sold to modern types of retail in 2007 is nature Dutch-type cheese). (364)

579. Indeed, respondents widely agreed that Dutch-type cheese produced in the Netherlands is different from Dutch-type cheese produced abroad due to differences in taste and quality ("Dutch cheese has a stronger taste and a higher quality"; "Dutch nature cheese is incomparable to foreign nature cheese due to Dutch milk, starter cultures and rennet"; "the use of raw milk is key"; "feed, milk quality and production process" are different; "the quality of raw milk and thus the quality and taste of the cheese are influenced by the geographical position of the Netherlands (mostly below sea level), its maritime climate and the composition of the grass growing on sandy and clay soils, as well as by cow feed in the Netherlands". (365)

580. A Dutch supermarket chain also underlined that the price of Dutch-type cheese produced in Germany has been lower than in the Netherlands in the recent past (366) but that this price difference has not caused significant substitution by either modern types of retail or end consumers between Dutch-type cheese made in the Netherlands and Dutch-type cheese made in Germany. (367)

581. These findings also correspond to the statements of specialised cheese wholesalers that Dutch consumers have a strong preference for Dutch-type cheese made in the Netherlands. One specialised cheese wholesaler even reported that sales of German-made Dutch-type cheese to modern types of retail had been unsuccessful in the past due to the different quality of that cheese. (368)

582. In view of the above, it is concluded that the market for the sale of nature Dutch-type cheese to modern types of retail is national in scope.

8.3.2.2. Rindless cheese

583. The market investigation revealed that the geographic scope is different, however, for a separate market for the sale of rindless Dutch-type cheese (including the narrower Gouda and Maasdam markets) where quality, taste and consumer preferences play a less important role and modern types of retail source almost 50% of their rindless cheese from abroad. (369) Similarly, many wholesalers indicated that they could source rindless cheese from abroad, in particular from Germany. (370) It follows that the markets for the sale of rindless Dutch-type cheese to specialised cheese wholesalers and modern types of retail are wider than national and include at least the Netherlands and Germany.

8.3.2.3. Dutch-type cheese, comprising nature and rindless cheese

584. In line with many of the observations made for nature Dutch-type cheese, also all markets for the sale of Dutch-type cheese (371) (which comprise both nature and rindless cheese) are national in scope.

585. The volumes of nature Dutch-type cheese sold in the Netherlands are significantly higher than those of rindless cheese. Indeed, 93% of all sales of Dutch-type cheese to modern types of retail and 68% of sales of Dutch-type cheese to specialised cheese wholesalers are nature Dutch-type cheese. Nature cheese therefore constitutes the predominant segment of both markets. The customer preferences and consumption patterns mentioned in section 8.3.2.1 for nature Dutchtype cheese therefore also have a decisive repercussion on the geographic dimension of Dutchtype cheese.

586. Moreover, several respondents in the market investigation representing modern types of retail in the Netherlands stated that the quality and taste differences between Dutch-type cheese produced in the Netherlands and Dutch-type cheese produced abroad also exist in the case of rindless cheese (albeit to a lower extent than in the case of nature cheese). (372)

587. Furthermore, the level of imports remains insignificant if nature and rindless cheese are viewed together as Dutch-type cheese. While the notifying parties estimate that approximately 15% of the total Dutch-type cheese consumption in the Netherlands is imported (373), the Dutch Dairy Board estimates that these imports only amount to approximately 10%. (374) The market investigation revealed that the share of imported Dutch-type cheese in the total sales of Dutchtype cheese in 2007 is even smaller as far as sales to specialised cheese wholesalers and sales to modern types of retail are concerned (approximately 5% in the case of sales to specialised cheese wholesalers and less than 5% in the case of sales to modern types of retail). (375) Current sourcing patterns in the Netherlands are therefore also clearly confined to the national level in the case of Dutch-type cheese.

588. In addition, the presence of specialised cheese wholesalers as an intermediary level between production level and downstream distribution level trading both nature and rindless Dutch-type cheese constitutes a particular feature of the Dutch market that further underlines the national scope of all markets for the sale of Dutch-type cheese.

589. In view of the above, it is concluded that the markets for sale of Dutch-type cheese are national in scope. The market investigation does not suggest that a different conclusion should be reached in the case of a further segmentation of Dutch-type cheese into Gouda/Maasdam and 15- day-old cheese/other cheese.

590. Moreover, this conclusion is confirmed by the observations made in section 8.2.2.3 which underline the fact that the competitive pressure exerted by rindless cheese upon nature cheese is limited and, in any event, insufficient so as to define a geographic scope for the Dutch-type cheese markets which would be wider than national.

8.3.2.4. Conclusion

591. On the basis of the elements discussed above, it is concluded that the markets for the sale of Dutch-type cheese (comprising both nature and rindless cheese) and the markets for the sale of nature Dutch-type cheese to specialised cheese wholesalers and modern types of retail are national in scope while the markets for the sale of rindless Dutch-type cheese to specialised cheese wholesalers and modern types of retail are wider than national and include at least the Netherlands and Germany. The same conclusion applies if each of those markets is defined more narrowly, that is, only comprising Gouda or Maasdam, and if the markets for the sale of Dutchtype cheese to specialised cheese wholesalers are in addition defined more narrowly, that is, only comprising 15-day-old cheese.

8.4. Competitive assessment

592. The concerns expressed by market players and the arguments put forward by the notifying parties have been carefully assessed and it was concluded that the proposed transaction is likely to significantly impede effective competition on the following markets:

- Dutch-type cheese sold to specialised cheese wholesalers in the Netherlands (or narrower segmentations into nature, Gouda, and 15-day-old cheese); and

- Dutch-type cheese sold to modern types of retail in the Netherlands (or narrower segmentations into nature and Gouda cheese).

593. Each of these markets covers the full territory of a Member State and thus constitutes a substantial part of the common market.

594. Although both markets (including narrower segmentations) will be assessed separately, it is important to note at the outset that these markets closely interact with each other. Indeed, whereas the notifying parties (and one other Dutch cheese producer) focus on sales to modern types of retail, specialised cheese wholesalers are active in all downstream distribution channels in the Netherlands. The Dutch-type cheese supplied by specialised cheese wholesalers to all downstream channels (including modern types of retail) will very often be produced by the notifying parties. Indeed, it can be estimated that approximately [70-80]*% of all downstream sales of Dutch-type cheese in the Netherlands in 2007 were achieved with cheese produced by the notifying parties. (376) This illustrates that the exercise of market power of the notifying parties visà- vis wholesalers is likely to have repercussions on the downstream sales to modern types of retail as well.

8.4.1. Sales of Dutch-type cheese to specialised cheese wholesalers

595. As explained in section 8.2.2.5, specialised cheese wholesalers play an intermediary role in the Dutch cheese market between the production level and the downstream level.

596. The market investigation revealed that a precise reconstruction of the sales of wholesalers to the downstream distribution channels in the Netherlands and their exports is complicated for several reasons, notably double counting resulting from wholesalers selling to each other, outsourcing contracts registered as sales and purchase back operations (the "contractual" selling back of cheese by wholesalers to producers after maturing and/or packaging in order to reduce risk). (377)

597. The notifying parties have provided the following estimate of the most important specialised cheese wholesalers in terms of domestic sales and exports in 2007: (378)

Table 8-1: Domestic sales and exports of main cheese wholesalers in 2007.

598. Although the market investigation revealed that specialised cheese wholesalers have sold considerably less Dutch-type cheese in the Netherlands (in particular to modern types of retail, see section 8.4.2.1) than estimated by the notifying parties, it can be inferred from Table 8-1 that specialised cheese wholesalers are active in all downstream channels in the Netherlands, and that exports represent a considerable part (albeit less than half) of their total sales quantity. The market investigation also showed that nature and rindless cheese have approximately equal shares in these exports. (380)

8.4.1.1. Market structure

599. Table 8-2 displays the notifying parties' most recent market share estimates (on the basis of volume) on the market for the sale of Dutch-type cheese to specialised cheese wholesalers in the Netherlands. (381)

Table 8-2: Market shares for sales of Dutch-type cheese to specialised wholesalers

600. Table 8-2 illustrates that the proposed transaction would combine the first and third largest players on the market. Together, they have a market share of [40-50]*%, followed by DOC ([20- 30]*%) and the other Dutch cheese producers with considerably lower market shares, that is CONO ([5-10]* %), Rouveen ([0-5]*%%) and Bel ([0-5]* %). (382)

8.4.1.2. Closeness of competition

601. Besides market shares, another important aspect for assessing the unilateral effects arising from the proposed merger is the degree of substitutability between the notifying parties' cheeses. The higher the degree of substitutability, the more likely it is that the notifying parties will significantly raise their sales prices to specialised cheese wholesalers.

602. As explained in sections 8.2.2.7 and 8.2.2.8, both notifying parties focus on the large scale production of nature cheese supplied to specialised cheese wholesalers and retailers.

603. In contrast, the smaller cheese producers in the Netherlands focus on particular segments, that is premium products such as "NH Gouda" and the Beemsterkaas brand (Cono), own brands such as Leerdammer (Bel), specialities and organic cheese (Rouveen). (383)

604. The market investigation also showed that the notifying parties are each other's closest competitors while DOC has a clear focus on rindless cheese at least partly destined for the industry channel, and since it is unlikely that DOC would in the near future switch a significant volume of its rindless cheese production to the production of nature cheese (see section 8.4.1.3). In contrast, approximately [60-70]*% of Friesland Food's production and approximately [90- 100]*% of Campina's production is nature cheese. (384)

605. This illustrates that the notifying parties are each other's closest competitors due to the similarity of their business models and their focus on the production of high volumes of nature cheese in the Netherlands. (385)

8.4.1.3. Possibilities of switching to other cheese producers in the Netherlands are limited

606. The notifying parties argue that in the case of price increases specialised cheese wholesalers would look for alternative sourcing from Dutch cheese producers other than the notifying parties.

607. However, the market investigation revealed that such possibilities for switching are limited in the case of nature cheese which constitutes the predominant segment of the market for Dutchtype cheese:

608. This applies in particular to 15-day-old cheese (which amount to approximately [60-70]*% of the notifying parties' sales to specialised cheese wholesalers (386)) as several respondents to the market investigation emphasised that they would be unable to source this cheese from any other producer than the notifying parties due to the focus of the other Dutch producers on particular market segments and their lower available quantities. (387)

609. The notifying parties themselves state that in the eventuality that the merged entity would raise its price of nature cheese by 5 to 10 %, it is unlikely that this would give Bel, Cono and Rouveen a strong incentive to start selling the type of commodity nature cheese sold by the notifying parties to wholesalers. In particular, the notifying parties explain that these producers sell branded and specialty cheeses that tend to earn higher profit margins. (388)

610. As regards DOC, the notifying parties argue that DOC would switch part of its production capacity from rindless cheese to nature cheese. DOC has alledgedly a low margin on rindless cheese and a price increase of 5-10% would give DOC a strong incentive to switch.

611. The market investigation revealed that DOC has the technical ability to switch at least 74 000 tonnes from rindless cheese to nature cheese and that it has additional free capacity available. (389) DOC could therefore potentially produce significantly more nature cheese, depending on the market conditions.

612. However, DOC predominantly focuses on the production of rindless cheese (rindless cheese constitutes approximately [70-80]*% of its production (390)) and it is for several reasons unlikely that DOC would produce a significant additional volume of nature cheese.

613. DOC has two plants (one older plant and one new plant which has been operational since 2003). The ratio of DOC's rindless/nature cheese production during the last three years has been relatively stable, notwithstanding considerable fluctuations of prices in the Netherlands of rindless and nature cheese. (391) Such an absence of considerable differences in the output of different products might be explained by several factors. One possible explanation is the intention of a producer to obtain optimal use of its capacity.

614. Furthermore, the notifying parties have explained that the pricing for nature cheese is determined by the pricing developments for rindless cheese. (392) This general correlation between rindless and nature cheese prices is a further reason for a producer such as DOC not to significantly change the output ratio of its nature/rindless cheese production.

615. Moreover, the notifying parties are incorrect in arguing that DOC has a low margin on rindless cheese and would thus have a strong incentive to switch in the case of a price increase of 5%-10% for nature cheese. Indeed, DOC's margins on rindless cheese are not, as such, lower than those on nature cheese but depend on the relevant price and customer. (393) It is therefore unclear if DOC has the incentive to produce a significant additional amount of nature cheese if the notifying parties were to increase prices for nature cheese by 5%-10%. Such a switch would primarily depend on market developments and the individual customer relationships of DOC.

616. Rather, general market developments point to the opposite, that is current and future market developments would not, as such, incentivise a Dutch-type cheese producer to produce more nature cheese instead of rindless cheese either by switching its existing production from nature to rindless cheese or by using additional production capacity for the production of nature cheese. Indeed, the notifying parties acknowledge that, in general, the volume of rindless cheese has shown steady growth at the expense of the volumes of nature cheese and that this trend is likely to continue ("cheese manufacturers agree that rindless cheese largely has the future, as its production costs are lower and its applications are the same"). In the Netherlands, the nature cheese market has been in decline in the past several years and further decline is expected. (394) A strategic decision of a producer to focus on a market segment which is declining does not appear likely.

617. Currently, demand is relatively low both for rindless and nature cheese. In such a climate, Friesland Foods allocates more production capacity to rindless cheese because rindless cheese can be stored longer without ripening and because the market is larger and more liquid on the demand side than that for nature cheese. (395) It can be assumed that DOC's reaction will be no different, that is to say that in such a climate DOC might not produce a significant amount of additional nature cheese to counter a potential price increase by the notifying parties of 5% to 10% unless such a price increase remained permanent at least for a fixed period.

618. A further disincentive to switch production is evident from DOC's individual customer relationships. Nature cheese is only produced by DOC on order, and DOC decides how much cheese to produce on a yearly basis, based on feedback from specialised cheese wholesalers on how much cheese these wholesalers expect to sell over the next few years. Subsequently, DOC procures the additional raw milk necessary to satisfy this demand from non-member farmers. (396) This yearly planning and the fact that nature cheese is only produced by DOC based on orders illustrates that DOC is unlikely to react in the short term to any price increases by the notifying parties of nature cheese in the Netherlands.

619. Also, DOC appears to have a traditional focus on the export market (already in 2000, when DOC's production capacity was limited to 50,000 tonnes, [80-90%]* of DOC's production was exported (397) while approximately [70-80]*% of DOC's production are reported to be exported today (398)). Any significant change of nature/rindless output might therefore require a strategic decision by DOC to change a long-lasting business model.

620. It should further be recalled that DOC does not directly sell downstream. For instance, while the notifying parties participate in tenders and supply modern types of retail directly, DOC works exclusively via wholesalers and will not even answer calls for tenders. (399) It therefore has a business strategy which differs considerably from the notifying parties' business model and this lack of direct market access makes it more unlikely that DOC would immediately address potential attempts by the notifying parties to increase prices downstream.

621. It follows that DOC's incentive to switch to the production of a significant additional amount of nature cheese is not at all evident but essentially hinges upon the incentive of wholesalers who are able to submit commercially attractive orders to DOC. Specialised cheese wholesalers may submit offers to modern types of retail in order to meet the latter's need for large-scale procurement, only if they know in advance that DOC will provide them with those respective volumes. This illustrates that the supply chain between DOC and specialised cheese wholesalers allows a less immediate and effective reaction to the large-scale procurement needs of modern types of retail than the notifying parties who can supply such large volumes directly.

622. From this perspective, it must also be borne in mind that prices would have to increase considerably in order for DOC to change its normal allocation of production capacity. Indeed, while a price increase of 5% would mean a direct profit for the notifying parties, DOC would have to share any profit from a price increase with the relevant wholesalers that wishes to undercut the notifying parties downstream. It is reasonable to assume that a price increase of 5% would therefore not increase DOC's profit by more than 2.5%. This further underlines the fact that a price increases in nature cheese needs to be relatively high and of a rather significant duration so as to incentivise DOC to switch a significant part of its rindless cheese production to nature cheese.

623. It follows that it is unlikely that DOC would produce a significant additional amount of nature cheese unless prices for nature cheese increased significantly and more permanently by 5%-10%.

624. In addition, even if DOC produced more nature cheese as a result of a production switch or use of its spare capacity, possibilities for wholesalers to switch to nature cheese produced by DOC are limited for reasons of [...]*. Indeed, several market participants have reported that the [...]*. (400) Also at the level of modern types of retail, it was emphasised that [...]*. (401)

625. It follows that the possibilities for wholesalers to switch to other cheese producers in the Netherlands are limited.

8.4.1.4. Possibilities of switching to other cheese producers in Germany are limited

626. The notifying parties further argue that in the case of a price increase specialised cheese wholesalers would look for alternative sourcing from German producers.

627. However, such alternatives for sourcing nature cheese abroad are limited. Many wholesalers participating in the market investigation find it difficult to mature imported nature cheese and the majority of them do not consider nature cheese from abroad as a viable option for their business. (402) Several wholesalers made it clear that cheese produced in the Netherlands is a fundamental part of their ripening activity and that the quality of foreign cheese in terms of taste and crust does not meet the required standards. (403) Specialised cheese wholesalers unanimously stated that a strong consumer preference exists in the Netherlands for Dutch-type cheese made in the Netherlands which is perceived as being of superior quality and taste. (404) Accordingly, almost all specialised cheese wholesalers in the Netherlands expect that a price increase of 5%-10% in Dutch type nature cheese made in the Netherlands would not be sufficient to influence Dutch consumers so as to switch to Dutch-type cheese produced for example in Germany where mostly rindless cheese is produced. (405)

628. Although the notifying parties explain that German producers are increasingly improving their cheese production know-how, only one specialised cheese wholesaler has so far been able to substitute the notifying parties' nature cheese with nature cheese produced in Germany. According to that wholesaler, it required several years of co-operation with a German supplier until that supplier was able to produce the desired nature cheese quality for some recipes of that wholesaler. For other recipes, that is cheese destined to ripen for a longer period, the co-operation was terminated after four years because the quality required by that wholesaler could not be achieved. (406) These difficulties in switching are also acknowledged by the notifying parties. (407)

629. Similarly, several specialised cheese wholesalers have attempted to obtain 15-day-old cheese from abroad in the past but were not satisfied with the quality of the imported Dutch type nature cheese when maturing it over a longer period of time and underlined that Dutch type nature cheese used for maturing needs to be of Dutch origin. (408) The importance of Dutch origin has also been emphasised by a further respondent to the market investigation. (409)

630. In addition, the market investigation revealed that the additional capacity of German competitors is limited, in particular due to the availability of maturing facilities. The total additional capacity in Germany thus amounts to only approximately 10% of the total downstream market volume of nature cheese in the Netherlands. (410).

631. In line with the above, a specialised cheese wholesaler stated: "The newly formed company [...] has major market shares (above 75%) in Gouda cheeses that can be matured over 8 weeks. Dutch Gouda cheeses are the only Gouda cheeses that can be well matured in the sense that they improve in taste, fragrance, flavour, develop mineral cristals etc. Gouda from other Member States like Germany is not good for ripening. They need to be consumed before 6/8 weeks of age." (411)

632. It appears that cheese matured for a longer period of time is one of the segments of the downstream market where specialised cheese wholesalers can best compete with producers. In this vein, a specialised cheese wholesaler stated: "Maturing is a form of production of better tasting cheeses. There is direct competition. If we still can compete there is a role for us (and others) in the fragmented market." (412)

633. All the evidence therefore suggests that the alternatives available to specialised cheese wholesalers to source Dutch type nature cheese from abroad are significantly limited and in the short-term (with limited exceptions) virtually non-existent.

634. Sourcing of rindless cheese from abroad appears to be easier but it should be borne in mind that rindless cheese is not a sufficient substitute for nature cheese on the demand-side and only accounts for approximately 31% of the overall Dutch-type cheese sales to specialised cheese wholesalers.

635. Specialised cheese wholesalers therefore consider the notifying parties as their main source of Supply (413) and are dependent on them. This dependency is even heavier in the case of nature cheese and in particular in the case of 15-day-old cheese which constitutes [60-70]*% of the notifying parties' sales to wholesalers and where most wholesalers have experienced shortages in the previous year (414) but did not switch to suppliers abroad.

636. It can be concluded that none of the notifying parties' competitors in the Netherlands or abroad (or all of them together) would be able to adequately replicate within a short time frame the constraining effect which Friesland Foods and Campina exert on each other pre-merger. They will therefore be unable to restrain the notifying parties post-merger and to prevent them from increasing prices to specialised cheese wholesalers.

8.4.1.5. Entry unlikely to occur

637. The market investigation also revealed that entry into the Dutch market is unlikely to occur.

638. No respondent was aware of any recent entry into the market for the sale of Dutch-type cheese to specialised cheese wholesalers. The clear majority of respondents does not expect such an entry in the future. (415) Only one wholesaler mentioned that there might be a new German entrant if the price for nature cheese stayed higher than the price for rindless cheese. However, this wholesaler considers it difficult to enter a new market where most farmers are members of cooperatives as are the notifying parties. (416) Similarly, another wholesaler underlined that the supply of milk required for the production of cheese is locked into the notifying parties. It would therefore be impossible for smaller cheese producers to increase their production of Dutch-type cheese. (417)

639. The market investigation has indicated that all Dutch cheese producers are planning to expand their current production capacity as a result of the expected increase of the existing milk quotas and possible abolition of it in 2015. However, it is unlikely that any of these expansion plans will lead to additional output of Dutch-type cheese within the next two years, (418) the time frame normally taken into account for the purpose of merger control.

8.4.1.6. Competitive pressure resulting from the possibility of re-importing exported cheese back into the Netherlands in case of price increases

640. The notifying parties claim that any attempt to increase prices by 5% to 10% within the Netherlands would lead to re-imports of cheese produced by the notifying parties and their Dutch competitors into the Netherlands, that is to say, any attempt to increase prices would thereby be defeated.

641. However, it is to be recalled that most of the notifying parties' sales to specialised cheese wholesalers is 15-day-old cheese. (419) The notifying parties export only [5-10]% of their 15-dayold cheese, that is to say any re-exports of 15-day-old cheese produced by the notifying parties will be very small in volume. (420)

642. All other cheese (nature cheese older than 15 days, rindless cheese) is exported and can therefore in principle also be re-imported. However, it does not appear that such re-imports would have a sufficiently restraining effect so as to prevent the notifying parties from increasing prices for Dutch-type cheese in the Netherlands, for several reasons.

643. First, the price of re-channeled Dutch-type cheese will necessarily reflect that it has been subject to at least two cross-border transports. This constitutes a competitive disadvantage compared to Dutch-type cheese destined for a first sale in the Netherlands. It is therefore unlikely that re-imports from areas other than neighbouring Member States (in particular Belgium and Germany) would be economically viable.

644. Second, the notifying parties themselves account for [50-60]*% of total exports of Dutch-type cheese. (421) A large part of these exports is sold directly to retailers, and not to traders or other parties that would be likely to re-import it into the Netherlands and sell it to specialised cheese wholesalers. (422)

645. Third, it has been demonstrated in section 8.2.2.3 that rindless cheese is only a limited substitute for nature cheese on the demand-side. Accordingly, any re-import of rindless cheese will only have a limited constraining effect on prices for the sale of nature cheese to specialised cheese wholesalers in the Netherlands.

646. Fourth, the market investigation revealed that the notifying parties are able to charge different prices in the Netherlands and abroad: one specialised cheese wholesaler reported that the notifying parties engage in intra-Community price dumping by keeping the price of private label cheese below the cost price outside the Netherlands in a successful attempt to eliminate specialised cheese wholesalers from the common market. (423) Another wholesaler noted that prices in the Netherlands are higher than in Belgium or Germany (424), and a third pointed to the ability of the notifying parties to raise prices in the Netherlands: "The prices of nature cheese are intentionally kept much higher by cheese producers than the prices of foil cheese". (425) These statements support the argument that the notifying parties are to a certain extent able to pricediscriminate between domestic and foreign customers. Furthermore, this ability does not appear to be halted by re-imports of Dutch-type cheese into the Netherlands. This ability to engage in price discrimination appears to already exist pre-merger and will undoubtedly be reinforced as a result of the transaction as customers will no longer have the possibility of switching between the notifying parties - each other's closest competitors.

647. The notifying parties' argument that any attempt to increase prices by 5% to 10% within the Netherlands would lead to re-imports of cheese, produced by the notifying parties and their Dutch competitors, can therefore not be supported.

8.4.1.7. Competitive pressure resulting from the possibility of selling cheese in the Netherlands that was originally destined for exports

648. The notifying parties also claim that any attempt to increase prices by 5% to 10% within the Netherlands would lead to increased sales of cheese in that Member state, originally destined for exports and that any attempts by the notifying parties to increase prices would therefore be defeated.

649. However, it must be recalled that 15-day-old cheese accounts for more than [40-50]*% (426) of all sales of Dutch-type cheese to wholesalers. 15-day-old cheese is hardly exported outside the Netherlands. It follows that any sales in the Netherlands of cheese originally destined for exports will not be sufficient to prevent the notifying parties from increasing the prices of 15-day-old cheese which constitutes approximately [60-70]*% of their overall sales of Dutch-type cheese to wholesalers. In other words, the notifying parties will not be prevented from increasing prices for the majority of their sales.

650. Furthermore, together Cono and Bel export a significant amount of Dutch cheese but they focus on different and higher market segments (branded segment and "NH Gouda"). Consequently, any opportunity to sell this cheese domestically (instead of exporting it) would not be sufficient to deter the notifying parties from increasing prices. DOC does not export at all but sells its cheese to wholesalers who subsequently export it.

651. Wholesalers who might wish to sell cheese in the Netherlands originally destined for exports need to buy that cheese from cheese producers in the first place. Any subsequent sale of this cheese in the Netherlands to other wholesalers will necessarily include a mark-up which will make this cheese more expensive. It is therefore equally unlikely that the possibility of wholesalers selling cheese in the Netherlands originally destined for exports will prevent the notifying parties from increasing prices to wholesalers.

652. In conclusion, the notifying parties' view that increased sales in the Netherlands of cheese originally destined for exports would prevent the notifying parties from increasing prices for Dutch-type cheese in the Netherlands cannot be accepted.

8.4.1.8. Competitive pressure resulting from potential demand reduction due to a price Increase

653. The notifying parties claim that a price increase of Dutch-type cheese sold to specialised cheese wholesalers would have negative repercussions on the volume of Dutch-type cheese bought by specialised cheese wholesalers, particularly their exports. Even if wholesalers could pass on price increases to their customers in the Netherlands, this would not be possible in export markets and specialised cheese wholesalers would therefore export less Dutch-type cheese produced by the notifying parties.

(i) Exports

654. The notifying parties export a considerable amount of Dutch-type cheese themselves, more than [50-60%]* of their production volumes. The majority of these direct exports are supplied to retail chains in [...]*. (427) The notifying parties have not explained why they could not increase these direct exports. In view of the fact that [60-70%]* of all cheese exported by the notifying parties is non-branded (428), it is reasonable to assume that the notifying parties could increase their direct exports if necessary.

655. Furthermore, as discussed above in section 8.4.1, it can reasonably be assumed that [50- 60%]* of the cheese exported by specialised cheese wholesalers ([...]* tonnes) was produced by the notifying parties. (429) This means that the notifying parties export approximately [5-10]* times more cheese themselves than wholesalers. (430)

656. Furthermore, even if it were accepted that wholesalers' demand would fall by 50% following a price increase (they would then only export [...]* tonnes of the notifying parties' cheese) and if the notifying parties could not substitute half of these lost exports, the final "lost exports" would still be low, that is, [...]* tonnes - and therefore only a small part of the total market for sales of Dutch-type cheese to wholesalers.

657. Furthermore [...]* tonnes only constitute approximately [10-20]* % of the notifying parties' production volumes, while [30-40]* % of the notifying parties' production is sold and remains in the Netherlands. This relatively small proportion of exports of the notifying parties' cheese by wholesalers means that it is unlikely that the notifying parties would abstain from increasing prices to wholesalers in order to prevent a slight decrease of their cheese's exports by wholesalers.

(ii) Domestic sales

658. As regards domestic sales of specialised cheese wholesalers, a price increase of Dutch-type cheese will not necessarily lead to a decreased demand to the extent that wholesalers have the option to pass on price increases to final consumers. As discussed in section 8.4.2, the notifying parties are indeed able to raise prices vis-à-vis modern types of retail which constitute more than half of the downstream market.

659. Furthermore, the notifying parties appear to be able to replace specialised cheese wholesalers in sales to modern types of retail (correspondingly, wholesalers complain about being driven out of the market, as discussed in section 8.4.2.3).

(ii) Conclusion

660. The notifying parties' argument that they would have no incentive to increase the price of Dutch-type cheese sold to specialised cheese wholesalers in the Netherlands as demand of wholesalers would decrease cannot, therefore, be accepted.

661. In addition, even a decrease of overall demand for cheese resulting from decreased demand by wholesalers which could not be substituted by other sales would not necessarily make it unattractive for the notifying parties to increase prices if they are able to divert their milk into other products such as milk powder. Indeed, Friesland Foods has capacity to produce milk powder and is able to vary its total cheese output within a [5-10]*% bandwidth. (431) It reduced the output of Dutch-type cheese in 2007 by diverting [...]* raw milk to milk powder production when prices for milk powder had increased and it was thus more profitable to produce milk powder instead of cheese. (432) Furthermore, there is no reason to assume that the notifying parties would not be able to expand their milk powder facilities.

8.4.1.9. Alleged inability to increase prices in "supply-driven" cheese market

662. The notifying parties finally submit that producers of Dutch-type cheese in the Netherlands have a clear economic rationale to always produce at maximum capacity. Cheese production is highly milk intensive (more than any other valorisation application for milk) and no alternative valorisation routes are available for these milk volumes in the short run.

663. However, for the reasons set out in section 8.4.1.8, it is unlikely that demand of wholesalers would decrease significantly following a price increase.

664. Furthermore, it is to be recalled that the notifying parties have high market shares in the sale of Dutch-type cheese to specialised cheese wholesalers and are each other's closest competitors. Wholesalers have only limited possibilities of switching, and entry and expansion are unlikely to occur in the Dutch market within the next two years. Sourcing possibilities from abroad are even more limited and will further decrease in the event the PGI applications for "Gouda Holland" and "Edam Holland" are granted.

665. This market power will also allow the notifying parties to increase prices while maintaining the current level of milk valorisation by way of cheese production. Market power on both the procurement and downstream cheese levels reciprocally reinforce each other.

666. Furthermore, the notifying parties have valorisation possibilities other than sales of Dutchtype cheese in the Netherlands, such as exports or increased production of milk powder.

667. Finally, as discussed in section 8.4.2, the notifying parties have the ability and the incentive to increase prices to modern types of retail in the Netherlands. This will allow specialised cheese wholesalers to pass on any price increase in Dutch-type cheese.

8.4.1.10. Alleged dependency on wholesalers' storage and maturing capacity

668. The notifying parties claim that they are not in a position to limit output or increase prices vis-à-vis specialised cheese wholesalers as they would incur storage and maturing capacity problems as a result of reduced purchases by wholesalers.

669. First, storage problems do not appear to be an issue in the case of rindless cheese as rindless cheese can easily be stored in containers in cooled warehouses after it has left the production line.

670. In the case of nature cheese, the notifying parties' dependency on specialised cheese wholesalers in the Netherlands for storage and ripening of their own cheese appears to be limited. Most of the matured nature cheese sold by the notifying parties is already stored and matured by the notifying parties themselves and the notifying parties would only need a few wholesalers for the outsourced part. (433) The notifying parties have also argued in a different context that cheese wholesalers in the Netherlands would have an additional [10-20]*% maturing capacity available. (434) The market investigation confirmed that the Dutch maturing industry is indeed not working at capacity limits (435).

671. The notifying parties' view that storage and maturing capacity problems due to reduced purchases by wholesalers would prevent the notifying parties from increasing prices cannot, therefore, be accepted.

8.4.1.11. "Gouda Holland", "Edam Holland"

672. One specialised cheese wholesaler already anticipates that the aforementioned PGI applications for "Gouda Holland" and "Edam Holland" will be granted (436). Both applications state that Gouda and Edam nature cheese should henceforth be produced in the Netherlands from cows' milk obtained from Dutch dairy farms and should be matured into a consumer-ready product in Dutch maturing rooms. Furthermore, only the producers of "Gouda Holland" and "Edam Holland" will have access to protected starter cultures.

673. Another specialised cheese wholesaler expects further price increases once the applications are granted: "Friesland Foods and Campina, amongst others, would certainly want to increase the public awareness of PGIs in the Netherlands and the rest of the EU through marketing campaigns. The necessary marketing budget would probably be created through a levy on the 15- day-old nature cheese of Gouda and Edam. The cheese producers might also want to obtain a premium price for the product that fulfils the criteria of the PGI. Moreover, purchasing 15-dayold cheese, milk for Gouda and Edam or even starters for Gouda and Edam abroad would be impossible if a specialised cheese wholesaler wants to fulfil the criteria of the PGI." Similarly, another specialised cheese wholesaler stated: "We think that it would lead to an increase in pricing" (437).

674. As mentioned in section 8.3.2.1, this Decision does not rely on the PGIs being granted. However, if the PGIs for "Gouda Holland" and "Edam Holland" were granted, the dominant position of the notifying parties post-merger would be further reinforced due to the fact that Dutch milk would be necessary for the production of "Gouda Holland" and "Edam Holland" and that the notifying parties control approximately [70-80]*% of the raw milk production in the Netherlands.

8.4.1.12. Conclusion

675. It can therefore be concluded that the concentration will lead to a significant impediment of effective competition on the market for the sale of Dutch-type cheese to specialised cheese wholesalers in the Netherlands. This market constitutes a substantial part of the common market. As the notifying parties have also underlined in their reply to the Statement of Objections and subsequent submissions, sales to specialised cheese wholesalers are linked to downstream sales (in particular to modern types of retail) and this link will be discussed in section 8.4.2. (438)

8.4.1.13. Sales of nature Dutch-type cheese to specialised cheese wholesalers

676. The conclusion reached in section 8.4.1.12 would be no different if the market was defined more narrowly - as nature Dutch-type cheese. In such a market, the notifying parties would have high combined market shares of approximately [50-60]*% (increment of [20-30]*%, competitors DOC [10-20]*%, Cono [5-10]* %, Rouveen [5-10]*% and Bel [5-10]*%) (439).

677. Furthermore, as nature cheese constitutes 68% (440) of all Dutch-type cheese sold to specialised cheese wholesalers, the considerations in sections 8.4.1.2 - 8.4.1.5 apply a fortiori in the case of nature Dutch-type cheese. In particular, the notifying parties are each other's closest competitors as they both focus on the production of large volumes of nature cheese. Their competitors are either specialised in branded products and/or specialities or in rindless cheese which considerably restricts the ability of specialised cheese wholesalers in the Netherlands to switch suppliers. It is therefore clear that the notifying parties' competitors would not be able to adequately replicate within a short time frame the constraining effect which Friesland Foods and Campina exert on each other pre-merger. Also any competitive pressure resulting from the possibility of reimporting nature cheese or selling nature cheese in the Netherlands originally destined for exports, or the alleged dependency of the notifying parties on wholesalers' demand or wholesalers' storage and maturing capacity, will not prevent the notifying parties from increasing prices in the Netherlands (see sections 8.4.1.6 - 8.4.1.10 which apply mutatis mutandis to nature Dutch-type cheese).

8.4.1.14. Sales of Gouda to specialised cheese wholesalers

678. The transaction will also lead to a significant impediment of effective competition if the market was defined as only comprising Gouda. Similar to the market for Dutch-type cheese, the notifying parties have a high combined market share of approximately [40-50]*% (increment of [10-20]*%, competitors DOC [20-30]*%, Cono [5-10]*%, Rouveen [5-10]*% and Bel [0- 5]*%) (441).

679. Furthermore, the arguments set forth in section 8.4.1 apply mutatis mutandis for Gouda as 80% of Dutch-type cheese sold to wholesalers in the Netherlands is Gouda cheese (442). Gouda cheese therefore constitutes by far the biggest segment of the Dutch-type cheese market.

8.4.1.15. Sales of nature Gouda to specialised cheese wholesalers

680. The transaction will lead to a significant impediment of effective competition if the market was defined as only comprising nature Gouda. In that market, the proposed transaction would combine the two largest players on the market with a combined market share of [50-60]*% (increment of [20-30]*%), followed by the other Dutch cheese producers with considerably lower market shares, namely DOC ([10-20]*%), CONO ([10-20]*%), Rouveen ([5-10]*%) and Bel ([5- 10]*%) (443).

681. Furthermore, as nature Gouda cheese represents 90% (444) of the total Dutch type nature cheese market, the considerations in sections 8.4.1.2 - 8.4.1.5 also apply in the case of nature Dutch-type cheese. In particular, the notifying parties are each other's closest competitors due to their focus on the production of large volumes of nature Gouda cheese. Their competitors are either specialised in branded products and/or specialities or in rindless Gouda cheese which considerably restricts the ability of specialised cheese wholesalers in the Netherlands to switch suppliers. It is therefore clear that the notifying parties' competitors would not be able to adequately replicate the constraining effect which Friesland Foods and Campina exert on each other within a short time frame. Also any competitive pressure resulting from the possibility of re-importing nature Gouda cheese or selling nature Gouda cheese in the Netherlands that was originally destined for exports or the alleged dependency of the notifying parties on wholesalers' demand or wholesalers' storage and maturing capacity, will not prevent the notifying parties from increasing prices in the Netherlands (see sections 8.4.1.6 - 8.4.1.10 which apply mutatis mutandis to nature Gouda cheese).

8.4.1.16. Sales of 15-day-old nature Dutch-type cheese to specialised cheese wholesalers

682. The transaction will also lead to a significant impediment of effective competition if the market was defined as 15-day-old nature cheese. In such a market, the notifying parties would have high combined market shares of approximately [60-70]*% (increment of [20-30]*%, competitors DOC [20-30]*%, Cono [5-10]*%, Bel [0-5]*% and Rouveen [0-5]*%) (445).

683. Furthermore, as 15-day-old nature cheese constitutes [60-70]*% of the notifying parties' sales to specialised cheese wholesalers, the considerations in sections 8.4.1.2 - 8.4.1.5 apply a fortiori for 15-day-old nature Dutch-type cheese. In particular, specialised cheese wholesalers consider the notifying parties to be their main source of supply and they would not be able to adequately replicate the constraining effect which Friesland Foods and Campina exert on each other within a short time frame (446). Any competitive pressure from decreased demand from wholesalers as a result of a price increase is therefore even more limited, and exports that could be diverted into the Netherlands are virtually non-existent in the case of 15-day-old nature Dutch-type cheese.

8.4.1.17. Sales of 15-day-old nature Gouda to specialised cheese wholesalers

684. The transaction will also lead to a significant impediment of effective competition if the market was defined as 15-day-old nature Gouda cheese where the notifying parties would have a market share of [60-70]*% (increment of [20-30]*%), followed by DOC ([20-30]*%), CONO ([5-10]*%), Bel ([0-5]*%) and Rouveen ([0-5]*%). Furthermore, as 15-day-old nature Gouda cheese represents 93% of the total market for 15-day-old nature Dutch-type cheese, all considerations mentioned above in section 8.4.1.16 apply mutatis mutandis (447).

8.4.1.18. Conclusion

685. It is therefore concluded that the concentration would lead to a significant impediment of effective competition on the markets for the sale of Dutch-type cheese to specialised cheese wholesalers (including narrower segmentations into nature, Gouda and 15-day-old cheese).

8.4.2. Sales of Dutch-type cheese to modern types of retail

686. For the reasons set out in section 8.4.2, it is concluded that the proposed transaction would also significantly impede effective competition on the market for the sale of Dutch-type cheese to modern types of retail in the Netherlands.

8.4.2.1. Market structure

687. According to the notifying parties' most recent market share estimates (on the basis of volume), the notifying parties would have a combined market share of [40-50]*% on the market for the sale of Dutch-type cheese to modern types of retail in the Netherlands (448) .

688. During the procedure there have been numerous exchanges between the notifying parties and the Commission concerning the total market volume and hence the notifying parties' market shares at retail level.

689. The notifying parties have based their market share estimates on the assumption that the total consumption of Dutch-type cheese (nature and rindless cheese including varieties) in the Netherlands was 280 000 tonnes in 2007 and that the retail level accounts for 70% of this consumption (196 000 tonnes) while the OOH and industry channel each account for around 15% (449). The retail level was further segmented by the notifying parties into modern types of retail [80-90]*%, that is [...]* tonnes), specialised cheese shops ([0-5]*%), open markets ([5-10]* %) and other small retail channels ([0-5]*%) (450). The notifying parties explain that their own market volume estimates would be more reliable than data compiled by the market research institutes AC Nielsen and GfK on the total market volume at retail level (451).

690. However, the notifying parties' approach to determine the total market volume at the retail level has various shortcomings which also affect their estimation of the total volume of the market for sales of Dutch-type cheese modern types of retail.

691. The notifying parties base themselves on the total cheese production in the Netherlands (730 000 tonnes, 717 000 tonnes after evaporation) increased by imports of Dutch-type cheese into the Netherlands (42 000 tonnes) and decreased by exports (478 000 tonnes). However, the data published by the Dutch Dairy Board on production, imports and exports which the notifying parties have used for their calculation do not specifically relate to Dutch-type cheese but rather to all cheese produced, imported and exported in the Netherlands. The notifying parties have therefore adjusted these Dutch Dairy Board data by excluding non-Dutch-type cheese but the categorisation of cheese as Dutch-type cheese or non-Dutch-type cheese is unclear in several instances (452) .

692. Furthermore, the notifying parties assume that the OOH and industry channel each account for 15% of the downstream market in the Netherlands and that the remaining part of the downstream market - 70% - accrues to retail. The main basis for these estimates is an internal study by Friesland Foods on the market size of the OOH and industry channels in the Benelux; the Netherlands alone was not analysed (453).

693. This approach taken by the notifying parties to determine the total market volume at supermarket level as residual volume after estimating the total Dutch-type cheese production, imports, exports as well as the size of the OOH and industry segments (454) is not only imprecise but also at odds with the fact that the notifying parties' downstream sales are to modern types of retail only. It can be assumed that the notifying parties have the necessary expertise to undertake a more direct approach of assessment of the total volume sold at the level of modern types of retail.

694. Considerable resources have been devoted to clarify the contested issue of the correct market size at the level of modern types of retail based on the most reliable sources available.

695. The information gathered during the market investigation strongly suggests that the notifying parties have overstated the total market volume at retail level (and thus also the volume of the modern types of retail which excludes specialised cheese shops, open markets and others.).

696. According to the questionnaire replies of companies active in modern types of retail in the Netherlands, approximately 120 000 tonnes of Dutch-type cheese have been sold by them in the Netherlands in 2007. (455) The notifying parties have acknowledged during the oral hearing that they do not contest this direct approach to calculate the total volume at the level of the modern types of retail.

697. AC Nielsen estimates that 130 400 tonnes of cheese have been sold by modern types of retail in the Netherlands in 2006 and that 18.7% of this cheese is non-Dutch-type cheese. (456) This results in a market volume for Dutch-type cheese of 106 000 tonnes at the level of modern types of retail.

698. The total market volume at that level has been estimated to be even lower by a respondent to the market investigation, that is, approximately 98 000 - 106 000 tonnes. (457)

699. Taking these factors into consideration, the notifying parties' estimate of the market volume of [...]* tonnes at the level of modern types of retail is too high. (458)

700. For the purpose of market share calculation, a market volume of 122 000 tonnes will be used. This total market volume has been relied upon in the Statement of Objections and finds authority both in the Commission's market investigation as well as data gathered by AC Nielsen. (459)

701. The accuracy of the present approach is further confirmed by comparing the notifying parties' sales to five main modern types of retail in the Netherlands with the quantities reported by these modern types of retail in their questionnaire replies; it should be noted that these figures vary by less than [0-5]*%. (460) Similarly, the notifying parties provided an overview of sales of Dutch-type cheese in modern types of retail in the Netherlands in their reply to the Statement of Objections which shows that the total sales of Dutch-type cheese in the Netherlands in 2007 ranges from a [...]*- [...]* tonnes. (461)

702. The notifying parties sell approximately [...]* tones (462) of Dutch-type cheese to modern types of retail and thus have a market share of [60-70]*% on a market for the sale of Dutch-type cheese to modern types of retail.

703. The market structure can be illustrated by Table 8-3. (463)

Table 8-3: Market shares for sales of Dutch-type cheese to modern types of retail

Year 2007 Sales of Dutch-type cheese to modern types of retail (in 1,000 tons) Market share Campina [...]* [30-40]*% Friesland Foods [...]* [20-30]*% Combined [...]* [60-70]*% Cono [...]* [0-5]*% Westland [...]* [5-10]*% Uniekaas [...]* [0-5]*% Vergeer [...]* [0-5]*% German competitors [...]* [0-5]*% Other [...]* [20-30]*% Total 122,000 100% Table 8-3 contains (i) all sales of the notifying parties to the modern types of retail (464), (ii) all branded sales of Dutch competitors and (iii) in the category "other" all private label/unbranded sales of Dutch competitors. It follows that also private label sales of Westland, Uniekaas, Vergeer and Cono might be contained in the category of "others". The category of "others" further contains sales of other specialised cheese wholesalers such as Cheese Partners. (465)

704. Table 8-3 illustrates that the proposed transaction would combine the two largest players on the market. Together, they would have a market share of [60-70]*% followed by several specialised cheese wholesalers, a Dutch cheese producer and German cheese producers which all have considerably lower market shares.

705. The market investigation revealed that the main specialised cheese wholesalers selling to the modern types of retail are Westland, Uniekaas, Vergeer, Zijerveld & Veldhuyzen and Cheese Partners. While Westland and Cheese Partners are not supplied by the notifying parties, Uniekaas, Vergeer and Zijerveld & Veldhuyzen source more than half of their cheese from the notifying parties. (466)

8.4.2.2. Closeness of competition

706. Table 8-3 confirms that the notifying parties account for approximately [90-100]*% of all direct sales of cheese producers to modern types of retail. (467) Their business models are similar in that they focus on the large scale production of nature Dutch-type cheese and private label/unbranded cheese. In fact, the notifying parties are the only cheese producers that supply each of the six main modern types of retail in the Netherlands. (468)

707. DOC, the biggest Dutch cheese producer after Friesland Foods and Campina, does not compete with the notifying parties at the level of modern types of retail but sells its entire production in the Netherlands to specialised cheese wholesalers. (469) The predominant part of these sales is rindless Dutch-type cheese, and it has been explained in section 8.4.1.3 that DOC is unlikely to be a significant alternative source of supply for nature cheese in the near future. Cono, the other Dutch producer with direct sales to modern types of retail, focuses on premium products such as NH Gouda.

708. The notifying parties' other competitors at the level of modern types of retail are essentially specialised cheese wholesalers who often have different business models with smaller volumes and an emphasis on branded products.

8.4.2.3. Competition between the notifying parties and specialised cheese wholesalers is Limited

709. Traditionally, Dutch cheese producers mainly sold 15-day-old cheese to specialised cheese wholesalers while the maturing and downstream sales of this cheese fell into the business of specialised cheese wholesalers. However, due to a growth of retailers, purchase organisations and product flows, cheese producers gradually started to mature and sell cheese themselves, thus entering into competition with specialised cheese wholesalers. (470) At the same time, specialised cheese wholesalers sell a considerable amount of cheese which they do not mature themselves.

710. The notifying parties submit that the relationship between suppliers and wholesalers is complementary and that wholesalers have a competitive advantage over cheese producers due to their portfolio of cheeses from different producers. However, a vast majority of wholesalers does not agree with this statement. (471) Competition is only possible on service (472) but not on price, in particular vis-à-vis big modern types of retail who obtain significant discounts when sourcing directly from the producer. (473)

711. The information gathered in the course of the market investigation indicates that specialised cheese wholesalers are to such an extent dependent on the notifying parties for their sourcing of Dutch-type cheese that their competitiveness in the downstream market for the sale of Dutch-type cheese to modern types of retail is significantly impaired.

712. In this regard, one wholesaler stated: "Specialised cheese wholesalers are in direct competition with cheese producers especially for the large contracts. If cheese producers keep their internal transfer price low than cheese wholesalers will not be able to keep these contracts and will easily by pushed out of the market. Only a few wholesalers will be left which have to do the laborious work." (474)

713. Another wholesaler stated: "[We] have to pay a considerable mark-up to the supplier for the 15-day-old cheese. The margin on matured cheese is not big enough for [us] to absorb the markup for the supplier in its own margin on the matured cheese. [We] are thus obliged to pass on the mark-up on 15-day-old cheese to the consumers of matured cheese. The suppliers of the 15-dayold cheese do not, of course, have to pay this mark-up." (475)

714. Similarly, many specialised cheese wholesalers stated that if their sales price of matured cheese was compared to the sales price of matured cheese of their supplier of 15-day-old cheese, their sales price would be higher. (476)

715. One of the same wholesalers also stated: "Cheese producers are at present very actively underbidding wholesalers who mature cheese themselves for supply to retailers for private label products. Most Dutch cheese wholesalers who mature cheese themselves are very much dependent upon sales of private label products [...]. [We] cannot afford to loose the private label sales as [we] need the extra scale. [...] If [the merged entity] would decide to stop supplying wholesalers that mature cheese, the wholesalers would have nowhere to go, as the production capacity of other producers is limited - not just by their facilities, but especially by the limited supply of milk. [...] It can be expected that the combined firm will be the cooperative that can offer the best milk price as it will have selling power. It will thus be able to continue to lock in the milk supply. The cheese market would thus clearly be dominated by the combined company [...] and would probably have the ability to raise prices for nature cheese". (477)

716. Another wholesaler stated: "There [is competition], but it's never a fair battle. If producers want they can always be cheaper. [Specialised cheese wholesalers] are not going to exist, the risk is too high". (478)

717. Another wholesaler stated: "[Specialised cheese wholesalers] will have to focus on the niches because the producers can only handle the big flows/volumes and/or they have to specialise on added values for the producers." (479)

718. Similarly, another wholesaler stated: "[Dutch cheese wholesalers] will lose all the large volume to prepack for large retailers. So they have to specialize, merge and do all other things to survive." (480)

719. Yet another wholesaler stated: "A too dominant party in the Dutch market can lead to monopolistic behaviour and unfair competition that can put us out of the market. Customers will in the beginning benefit from lower prices. But will later on also suffer monopolistic behaviour. The consumers ask for cheese from Holland. If the new company wants to take our big clients, they will be able to do so by increasing our price and sell directly to our customers for lower prices." (481)

720. Further wholesalers expressed concerns about the market power of the merged entity and their increasing dependence on the merged entity. (482)

721. These statements support the argument that the notifying parties will have a stronger position on the market for sales of Dutch-type cheese to specialised cheese wholesalers post-merger. The transaction will therefore both increase the dependency of specialised cheese wholesalers upon the notifying parties and impair their competitiveness in the downstream market for the sale of Dutch-type cheese to modern types of retail.

8.4.2.4. Limited possibility of switching to Dutch producers/wholesalers

722. Some modern types of retail voiced strong concerns in the market investigation about the lack of alternatives to the notifying parties. These concerns mostly relate to nature cheese but it should be recalled that nature cheese accounts for approximately 93% of all Dutch-type cheese sold by modern types of retail in the Netherlands. The question at issue is therefore whether modern types of retail are able to switch suppliers as far as their procurement of nature Dutch-type cheese is concerned.

723. In this regard, one major supermarket expressed concerns that the merged entity would become an unavoidable and monopolistic trading party, in particular for Gouda type cheese. No other suppliers in the Netherlands would be able to meet that supermarket's volume requirements. In addition, it would not be realistic to expect that other cheese suppliers could expand their capacity if they did not get access to sufficient quantities of raw milk. It could also not be excluded that smaller producers would follow the price setting of the merged entity and that this price raising strategy may turn out to be more advantageous for them than a capacity enhancing strategy. (483)

724. Another respondent expects less competition on all markets, less brands and a lack of competition on pricing matters for its customers. (484) Another respondent is concerned about possible price increases. (485)

725. It should be recalled that DOC is so far not supplying the level of modern types of retail but rather sells its cheese exclusively to specialised cheese wholesalers, and that [...]*. In addition, as explained in section 8.4.1.3, it is not certain that DOC will produce a significant amount of additional nature cheese in the near future. The comments made by specialised cheese wholesalers in section 8.4.2.3 further reinforce this conclusion.

726. As far as specialised cheese wholesalers as alternative source of supply are concerned, the notifying parties submit that the wholesalers Vergeer, Uniekaas, Zijerveld en Veldhuizen and Cheese Partners compete fiercely with them in tenders and that these wholesalers had won several calls for tenders in the last years. (486)

727. However, it is to be recalled that approximately [70-80]*% of all downstream sales of Dutchtype cheese in the Netherlands was achieved in 2007 with cheese produced by the notifying parties. That a considerable amount of cheese sold by specialised cheese wholesalers to modern types of retail was originally produced by the notifying parties is illustrated by the fact that in 2007 Vergeer, Uniekaas and Zijerveld en Veldhuizen sourced a majority of their Dutch-type cheese from the notifying parties. It follows that a lost tender is not necessarily a lost sale for the notifying parties.

728. Furthermore, the notifying parties submit that the bulk of their sales of [...]* tonnes to modern types of retail come from tenders. (487) It is illustrative that the tenders won by specialised cheese wholesalers and mentioned in the notifying parties' tender analysis in 2007 are much smaller in volume ([...]* tonnes in the case of Cheese Partners, [...]* tonnes in the case of Uniekaas and [...]* tonnes in the case of Zijerveld en Veldhuizen). (488)

729. In this respect, it must also be stressed that wholesalers' procurement of 15-day old cheese constitutes a major cost component of their sales price of Dutch-type cheese. (489)

730. The above underlines that the competitive pressure emanating from specialised cheese wholesalers upon the notifying parties, and hence the possibilities for modern types of retail to switch suppliers, are considerably limited.

731. It is thus not surprising that a respondent explained that the presence of specialised wholesalers would not reduce the market power of the notifying parties on the markets for Dutchtype cheese since wholesalers would be equally dependent on the notifying parties for the supply of Dutch-type cheese as modern types of retail. (490)

732. It follows that none of the notifying parties' Dutch competitors (or even all of them together) would be able to adequately replicate the constraining effect which Friesland Foods and Campina currently exert on each other. They will therefore be unable to prevent the notifying parties from increasing prices to modern types of retail post-merger.

8.4.2.5. Limited possibility of switching to German producers

733. The notifying parties argue that in case of price increases modern types of retail would look for alternative sourcing from German producers. Although German producers predominantly produced rindless cheese, they have state-of-the-art cheese producing facilities as well as Dutch technology and know-how for the production of Dutch-type cheese.

734. According to the notifying parties, there is no evidence that the quality of younger-aged German nature cheese (mild and mild mature) is inferior to the cheese produced in the Netherlands and this foreign-made cheese therefore competes with the notifying parties' sales in the younger segment.

735. It is true that young and mild mature cheeses represent more than half of all sales of Dutch type cheese at the level of modern types of retail.

736. However, the market investigation clearly showed that the ability of Dutch modern types of retail to switch to German suppliers is limited.

737. Modern types of retail active in the Netherlands who participated in the market investigation hardly sold any imported Dutch type nature cheese in 2007. (491) Almost all respondents, including German discounters active in the Netherlands, consider that Dutch consumers much prefer Dutch-type cheese produced in the Netherlands. (492) Furthermore, almost all modern types of retail in the Netherlands agree that Dutch-type cheese produced in the Netherlands is different from Dutch-type cheese produced abroad in both taste and quality. (493)

738. For these reasons, modern types of retail and Dutch suppliers indicated that only some consumers at the lower end of the market might switch to Dutch-type cheese produced abroad. (494) In their latest submission, the notifying parties have quantified this lower (that is, young) segment as [20-30]*% of the total retail market volume which is line with the market investigation. (495) In 2007, the notifying parties sold [...]* tonnes of young cheese to modern types of retail in the Netherlands which represents approximately [20-30]*% of their overall sales to modern types of retail in the Netherlands. (496)

739. It is clear that most modern types of retail in the Netherlands would have an incentive to source from abroad in the event of a price increase of 5% to 10% for Dutch-type cheese, but most of these modern types of retail also stated in their reply that such a switching would only be possible if cheese of comparable quality is found abroad. (497) As explained in section 8.3.2.1, almost all respondents currently consider that there is a difference in quality between Dutch-type cheese made in the Netherlands and Dutch-type cheese made abroad. This is confirmed by the fact that no respondent directly stated that a 5% to 10% price increase of nature Dutch-type cheese produced in the Netherlands would be enough for consumers to switch from nature Dutchtype cheese produced in the Netherlands to nature Dutch-type cheese produced abroad. (498)

740. The unlikelihood of such a switching is further underlined by the fact that no respondent (apart from two discounters active in the Netherlands) mentioned any foreign cheese producer as current or alternative source of supply for Dutch-type cheese whereas respondents did state that they could only be supplied by the notifying parties or specialised cheese wholesalers in the Netherlands. (499) As regards the discounters who could source from abroad, they have also made several statements which suggest that it would be unlikely that they would source cheese to be sold in the Netherlands from abroad ("the majority of our customers are heavy users and notice the slightest differences"; "consumer preferences are very strong"; "foreign producers are not equipped to sell 500-1000gr pre-packed lumps"). (500)

741. Furthermore, most respondents explained that either the packaging of their private label cheese informs that it is produced in the Netherlands and in any event, Dutch consumers would expect Dutch-type cheese to be produced in the Netherlands.(501)

742. This is confirmed by the experience of a Dutch supermarket chain who started to import Gouda cheese from Germany, mainly for the purpose of better understanding the product and its ripening capabilities. This experience showed that German Gouda, besides not being available in sufficient quantities, cannot serve as a substitute for Dutch Gouda due to differences in ripening. (502)

743. This respondent considers that in the near future (the next few years), the problems relating to quality, taste and capacity would remain a decisive hurdle in sourcing Dutch cheese from abroad. From a more long-term perspective, greater possibilities to source from abroad would probably arise but the development of these possibilities would remain uncertain and unpredictable and would in any event not solve the sourcing problems in the coming years (503), the time perspective relevant to the Commission's decisional practice.

744. These considerations apply even more so if the applications for "Gouda Holland" and "Edam Holland" are granted. The same respondent is concerned that if the applications are granted, the possibility of purchasing Gouda type cheese from abroad would be further complicated and the dependency on the notifying parties would increase further. Indeed, while most of the Dutch customers currently assume that Gouda cheese is always made in the Netherlands (which is at present not necessarily the case), they would become more attentive to the origin of cheese. (504)

745. It can therefore be concluded that the ability of Dutch modern types of retail to switch to German suppliers is considerably limited.

746. Moreover, this conclusion is reinforced by the evidence gathered during the market investigation which suggests that the notifying parties are not correct in arguing in their reply to the Statement of Objections that a small price increase in the Netherlands would give German cheese producers a strong incentive to increase their exports to the Dutch market:

747. Indeed, the notifying parties' most important German competitors for nature cheese (Nordmilch, Humana and Müller, which are the only German cheese producers with maturing Capability (505)) consider that specific barriers exist to enter the Dutch-type cheese market in the Netherlands. One German producer stated: "The Dutch market consists most of all of nature ripened cheese of all ages, i.e. cheese which we can deliver only to a limited extent due to restricted ripening capacities"; another German producer stated: "Sometimes the [Dutch] customers we approached in the past did not accept German cheese as equivalent to Dutch cheese. For the same product German cheese must be much lower in price"; yet another German producer stated: "High investment necessary to build up enough capacity in maturing room and specific equipment for maturing the cheese; existing plan, pushed by Friesland and Campina, to protect Gouda and Edam as Dutch Edam and Gouda". (506)

748. No German producer participating in the market investigation made a distinction between the different age segments of Dutch-type cheese in the Netherlands when replying to questions about barriers to entry and market developments. The possibility for German suppliers to enter the lowend segment of the Dutch market appears rather remote as is evidenced by the fact that most of them were unable to answer the question whether a price increase of 5%-10% of Dutch-type cheese made in the Netherlands would be sufficient to influence Dutch customers so as to switch to Dutch-type cheese produced outside the Netherlands. Only one German producer of rindless cheese considers that Dutch customers would switch and only one German producer of nature cheese indicated that switching is possible, provided the products at issue are of the same quality. (507)

749. Correspondingly, the most important German competitors are rather sceptical in their expectations concerning the evolution of the Dutch cheese market: One German competitor stated: "Dutch consumers will maintain existing preferences for natural ripened older products"; another German competitor stated: "The Dutch market is interesting due to high demand [...] We await that Dutch suppliers will try to protect their markets with low prices and special agreements"; another German competitor stated: "Foil ripened cheese will grow in the Netherlands. Quality issues (especially better taste) play big role in segment of natural matured cheese". (508)

750. All German producers participating in the market investigation stated unanimously that Dutch consumers perceive Dutch-type cheese made in the Netherlands as superior in quality than Dutch-type cheese produced outside the Netherlands and that Dutch consumers therefore have a strong preference for Dutch-type cheese made in the Netherlands, in particular for nature Dutchtype cheese. Reasons given were "long lasting experience of Dutch suppliers"; "hardly any production and maturing capacity outside the Netherlands that would allow to produce cheese older than 5 weeks"; and "based on long experience of producing such type of cheeses. Today some producers outside the Netherlands have more or less similar quality level by foil ripened cheese". (509)

751. All these replies are evidence of the unlikelihood that a small price increase in the Netherlands for nature Dutch-type cheese would give German cheese producers a strong incentive to increase their exports to the Dutch market.

752. The notifying parties further argue in their reply to the Statement of Objections that instead of relying on the majority of consumers, the Commission should have identified the set of marginal customers, that is those customers who discipline the pricing of sales of nature Dutch-type cheese made in the Netherlands. The survey attached to the PGI application suggests that 38% of Dutch consumers could potentially switch to Gouda of foreign origin (and 43% for Edam).

753. However, this study reports that only 15% of the consumers would certainly switch to non- Dutch origin Gouda if there were a 10% price difference increase. It is not clear that the substitutability of nature cheese produced in the Netherlands with nature cheese produced abroad at the lower end of the nature Dutch-type cheese market would sufficiently constrain the notifying parties and defeat the incentives for a price increase post-merger. Moreover, given the relatively limited experience of Dutch consumers with non-Dutch origin Gouda, it is not clear whether they are aware of the issues of differences of quality and taste between Dutch and non- Dutch origin Gouda which has been pointed out by respondents active in modern types of retail. The likelihood of such a substitution is even lower if "Gouda Holland" will be approved which will only allow nature cheese produced in the Netherlands to bear the designation "Gouda Holland".

754. In addition, Campina has devoted almost [...]* to the production of nature cheese for [...]*. This cheese is meant for sale in Germany (specifications that are in line with demand from German retailers in particular). (510) [...]* has, according to the market investigation, the largest capacity of all German producers to produce nature cheese. (511) At least one other main German producer bought nature cheese from the notifying parties in 2007. The existence of these supply agreements for nature cheese further underlines the unlikelihood that German producers would immediately enter into the Netherlands following a price increase of nature Dutch-type cheese.

755. This is further confirmed by the statement of a respondent at the level of modern types of retail in the Netherlands according to which the price of Dutch-type cheese produced in Germany had been lower in Germany than in the Netherlands in the recent past but that this price difference had not caused any significant substitution. (512)

756. Any ability of German producers to offer Dutch type nature cheese (which is limited, as shown in this section) will therefore not have a significant impact on the competitive behaviour of the notifying parties in the market for the sale of Dutch-type cheese to modern types of retail in the Netherlands.

8.4.2.6. Alleged dependency of notifying parties on retailers

757. The notifying parties submit that modern types of retail in the Netherlands have considerable buyer power which would prevent the notifying parties from increasing their sales prices. Specifically, two purchasers represent more than 60% of the whole Dutch consumption of Dutchtype cheese sold through supermarkets in the Netherlands. The five largest modern types of retail represent approximately 90% of the entire consumption of Dutch-type cheese sold through modern types of retail in the Netherlands.

758. The notifying parties have also argued that their sales to the discounters Aldi and Lidl account for over [80-90]*% of all private label sales of Friesland Foods and between 40 and 50% of all private label sales of Campina. Lidl negotiates centrally supply contracts for rindless and nature cheese annually on a European-wide scale. In the case of Aldi, there is not an overall European contract but Aldi Netherlands negotiate collectively for the Netherlands, Belgium, France, Portugal and Spain for both nature and rindless cheese. In the negotiations with Lidl and Aldi, the notifying parties compete head-on with German suppliers.

759. The notifying parties submitted further evidence in which in one instance one of their brands was delisted by a modern type of retail.

760. The notifying parties admit, however, that although the bulk of the notifying parties' sales are concentrated with a small number of modern types of retail, such a high level of concentration may not be in and of itself be sufficient evidence of buyer power. (513)

761. Furthermore, modern types of retail often threat to delist products for several reasons (lower than expected turnover, price negotiations and so forth). It is easier for modern types of retail to delist products if alternative suppliers are available.

762. The latter situation may change after the merger. Premerger, modern types of retail have two reliable sources of supply for their private label/unbranded Dutch type nature cheese. Post-merger they will have the choice to buy from the merged entity or to look for alternatives.

763. The likelihood that buyer power would act as a countervailing factor to an increase in market power is thus limited where buyers lack alternatives to the notifying parties. The information gathered during the market investigation and explained in section 8.4.2 indeed strongly suggests that alternative supply options of modern types of retail in the Netherlands are not sufficient enough to allow them to exercise buyer power post merger.

764. This applies where modern types of retail want to pursue multi-sourcing strategies in particular. (514)

765. Similarly, tenders are effective for competition if the players participating in tenders have the capacity to fulfil the needs or the buyer can split the order effectively. However, as explained in section 8.4.2.3, the competitiveness of specialised cheese wholesalers in tenders is limited.

8.4.2.7. Entry and expansion unlikely to occur

766. As mentioned in section 8.4.1.5, entry is unlikely to occur within the next two years and expansion by Dutch producers within the next two years is equally limited.

767. In particular, DOC can be considered the most significant potential entrant into the Dutch market for sale to modern types of retail. However, it is uncertain if DOC would alter its business model which entails the exclusive supply of specialised cheese wholesalers. In addition, as explained in section 8.4.1.3, it is unlikely that DOC would produce a significant additional amount of nature cheese in the near future.

8.4.2.8. Competitive constraints resulting from possible increased use of rindless cheese

768. As the transaction does not give rise to competition concerns with regard to rindless cheese (see section 8.4.3), the notifying parties submit that any attempts to increase prices for nature Dutch-type cheese at the level of modern types of retail would be defeated by an increased use of rindless cheese. The notifying parties also explain that an increased need for sliced finished products in modern types of retail would promote the production of rindless cheese. (515)

769. However, it is to be recalled that at the level of modern types of retail no significant substitution has yet taken place between nature and rindless Dutch-type cheese, as is evidenced by the low ratio of sales of rindless cheese in 2007 (only 7% of all sales of Dutch-type cheese at modern types of retail were rindless cheese, see section 8.2.2.3). Indeed, according to the market investigation demand side substitutability between nature and rindless cheese is very limited and most modern types of retail in the Netherlands consider that consumers would not switch from nature to rindless cheese if there were a price increase of 5% to 10%. (516) This illustrates that any competitive constraints emanating from rindless cheese on the nature cheese segment (which constitutes 93% of all sales at modern types of retail and where the notifying parties are each other's closest competitors) are limited.

770. The notifying parties submit that such a substitution between nature and rindless cheese would nonetheless be possible between younger nature cheese and rindless cheese, as both have a similar age and are therefore close substitutes at retail level.

771. However, the notifying parties explained in the Form CO that quality differences exist between rindless cheese and "mild mature" nature cheese. (517) Indeed, while "mild" nature cheese and rindless cheese have the same age, "mild mature" nature cheese is between two and five weeks older than rindless cheese which leads to improved taste and quality. (518) The notifying parties expect that the quality differences between mild mature nature cheese and rindless cheese will only be removed in the future. (519)

772. The segment of the nature Dutch type market cheese where substitutability with rindless cheese appears potentially significant is therefore rather confined to the "mild" segment.

773. According to the market investigation, modern types of retail achieve approximately 25% of their turnover in the segment of young nature cheese. The notifying parties achieve approximately [20-30]*% of their sales of Dutch-type cheese to modern types of retail in this segment. (520)

774. Furthermore, the notifying parties themselves submit that no substitution would be possible between mild nature cheese and rindless cheese as regards OTC sales (which account for [20- 30]*% of all sales of modern types of retail according to the notifying parties) and as regards sales of pre-packed wedges. (521) Only in the remaining part, that is pre-packed sliced cheese (which according to the notifying parties' submission of 8 October 2008 accounts for [30-40]*% of all sales of modern types of retail (522) while the notifying parties have quantified the size of this segment with [20-30]*% at the oral hearing (523)) are rindless and nature cheese theoretically substitutable to the extent that this sliced cheese is of young age. However, a significant part (if not most) of all young nature cheese sold in sliced form still contains a rind and is therefore clearly distinguishable from sliced rindless cheese. (524) Furthermore, as the notifying parties have submitted in a different context that they would only supply prepacked lumps and as Dutch consumers have a preference for lumps (see section 8.3.2.1), an increased use of rindless cheese would have, at most, an indirect effect on the notifying parties. (525)

775. Of the notifying parties' competitors in the Netherlands, only DOC has a focus on rindless cheese but DOC does not sell directly to modern types of retail and an estimated [70-80]*% of DOC's production is exported. (526) On the other hand, imports of rindless cheese at the level of modern types of retail amount to less than 5% of the total market volume at modern types of retail level. This further illustrates that the use of rindless cheese at the level of modern types of retail is unlikely to significantly increase in the near future.

776. In view of the above, it is unlikely that any price increase of nature Dutch-type cheese will be prevented by a substitution of nature Dutch-type cheese with rindless Dutch-type cheese. This conclusion will be reinforced in the event that the PGI applications are granted.

8.4.2.9. Possibility of re-importing Dutch-type cheese in case of price increases

777. The notifying parties have also argued that any attempt to increase prices by 5% to 10% within the Netherlands would lead to export trade flows being rechanneled into the Netherlands. However, similar observations can be made as those made in section 8.4.1.6 when discussing specialised cheese wholesalers.

778. In particular, Dutch-type cheese exported by wholesalers is typically packaged for the relevant export market, with different languages, logos and so on. In Germany, the main export Member State, the share of pre-packed cheese sales is higher than in the Netherlands. (527) A Dutch consumer would therefore perceive such a rechanneled cheese more as a German product and not as a substitute for Dutch-type cheese made in the Netherlands.

779. The market investigation also revealed that the notifying parties are able to charge different prices in the Netherlands and abroad (see section 8.4.1.6). This ability to price discriminate appears to already exist pre-merger and will only be reinforced by the transaction as customers will no longer have the possibility of switching between the notifying parties which are each other's closest competitors.

8.4.2.10. Possibility of selling cheese in the Netherlands that was originally destined for Exports

780. The notifying parties argue that any price increases in the Netherlands would be prevented by the possibility of specialised cheese wholesalers selling cheese originally destined for exports within the Netherlands.

781. It should be noted that the notifying parties were unable to estimate how much of their cheese is exported by specialised cheese wholesalers (the notifying parties were even unable to estimate the minimum amount of cheese produced by them and sold by specialised cheese wholesalers in the Netherlands). (528)

782. However, it can reasonably be assumed that [50-60%]* of all Dutch-type cheese destined for export by specialised cheese wholesalers is produced by the notifying parties themselves (see section 8.4.1.8). As shown in section 8.4.1, the notifying parties have both the ability and the incentive to increases prices to wholesalers and can therefore do so for [50-60%]* of all Dutchtype cheese exported by wholesalers. Unless wholesalers assume internally such a price increase, that cheese could only be sold by them in the Netherlands with a mark-up reflecting this price increase. The possibility of selling cheese originally destined for exports in the Netherlands will therefore not prevent the notifying parties from increasing prices vis-à-vis modern types of retail.

783. It should further be noted that approximately half of all Dutch-type cheese exported by specialised cheese wholesalers is rindless (see section 8.4.1) and that several wholesalers also focus on branded cheese (see section 8.2.2.4). In both instances, the substitutability between rindless and/or branded cheese destined for exports and the private label/unbranded nature cheese sold by the notifying parties is very limited and will not prevent the notifying parties from increasing prices vis-à-vis modern types of retail.

784. An important question at issue is therefore what quantities of exports are (i) rindless cheese, (ii) nature cheese produced by the notifying parties, and (iii) branded cheese.

785. If this question is answered on the basis of the market investigation which revealed that specialised cheese wholesalers have exported a total of 118 500 tonnes in 2007, it follows that (i) 61 500 tonnes of exported cheese are rindless cheese; that (ii) out of the remaining 57 000 tonnes of exported (nature) cheese [...]* tonnes is produced by the notifying parties (529) ; and that (iii) of the remaining [...]* tonnes of (nature) cheese not produced by the notifying parties approximately [...]* tonnes is private label/unbranded.

786. Alternatively, if this question is answered on the basis of the Dutch Dairy Board data according to which all exports of Dutch-type cheese by specialised cheese wholesalers amount to 98 000 tonnes in 2007 and if the general trends revealed in the market investigation were applied to this total of 98 000 tonnes, it follows that (i) 51 000 tonnes (52%) of exported cheese is rindless cheese, that (ii) out of the remaining 47 000 tonnes of exported (nature) cheese [...]* tonnes ([50-60]*%) is produced by the notifying parties; and that (iii) of the remaining [...]* tonnes of (nature) cheese not produced by the notifying parties approximately [...]* tonnes ([70- 80]*%) is private label/unbranded cheese.

787. Both calculation methods illustrate that the volume of private label/unbranded cheese that could prevent the notifying parties from increasing prices vis-à-vis modern types of retail is limited.

788. In addition, even if this quantity was greater, any rechanneling of exports will be complicated by several factors. Most importantly, it is unlikely that wholesalers will abandon all supply relations with customers abroad at short notice, as significant efforts are required to establish such supply relations. Furthermore, wholesalers are normally bound to their export customers by 1 year-contracts. (530)

789. It should also be noted that nature cheese is often produced on the basis of the relevant customer's requirements. (531) It is therefore highly doubtful that a modern type of retail (that has so far ordered nature cheese to be produced on the basis of its specific requirements) would turn to cheese not produced on the basis of its specific requirements if the notifying parties increased prices by 5%-10%. It follows that any competitive pressure emanating from the possibility of specialised cheese wholesalers to divert Dutch-type cheese to the Netherlands instead of exporting it, will be insufficient so as to prevent a price increase in the Netherlands.

8.4.2.11. Competitive constraints emanating from small retail outlets and other downstream distribution channels

790. The notifying parties have argued in their replies to the Statement of Objections that competitive constraints emanating from specialised cheese shops should be taken into account and would prevent price increases.

791. However, the total volume of Dutch-type cheese sold in specialised cheese shops, open markets and other small retail outlets amounts to approximately 24 000 tonnes and is therefore limited. Volumes sold to the OOH and industry channels are higher but it is to be recalled that packaging, forms of delivery are very different across the different channels (see sections 8.2.2.7 and 8.2.2.8). Furthermore, a major part (approximately [70-80]*% in 2007) of all cheese sold in retail, OOH and industry in the Netherlands is produced by the notifying parties. It is therefore unlikely that downstream competition between the different channels will be sufficient enough to prevent the notifying parties from increasing prices for Dutch-type cheese to modern types of retail. 8.4.2.12. Supply shock analysis

792. On 12 September 2008, the notifying parties submitted a supply shock analysis in support of their argument that they will be unable to raise prices post-merger. One of Campina's plants (Lutjewinkel) caught fire during weeks 34 and 51 of 2004 and were closed down for the whole of 2005. When compared to years 2003 and 2006 the level of production of Campina at its lowest point in mid 2005 was about [...]* tonnes lower while the sales price for cheese did not respond drastically but was rather flat throughout the period.

793. The per-month production figures for the Lutjewinkel plant, provided by the notifying parties on 24 September 2008, indicate that the production at the Lutjewinkel plant was indeed close to zero between September 2004 and January 2006. The Lutjewinkel plant at normal capacity produced on average around [20-30]*% of Campina's total cheese production. The notifying parties further argue that the negative shock in quantity corresponded to about [5-10]* % of total domestic consumption of cheese and that a negative temporary shock of such magnitude should have a negative effect on the domestic price level of cheese. The notifying parties interpret the absence of significant domestic price fluctuation as evidence that the merged entity would not be able to profitably reduce its cheese output due to its limited ability to increase the price in so doing.

794. Natural events, like disruptions of output due to a fire, can be useful in evaluating the potential of a market player to raise prices. After persistent large disruptions in output, one would expect that prices would rise as this represents a shift in the market supply curve. However, the outcomes of such types of natural event cannot be directly translated into predictions about post merger incentives for a number of reasons.

795. Firstly, a merger may change the structure of the industry permanently, whereas supply shocks are by their very nature temporary.

796. Secondly, when contracts are negotiated for longer periods of time, a temporary shock in supply will not be reflected in the price to the extent it would be in a different market structure where notifying parties enter into short term relations, such as local daily grocery markets.

797. Thirdly, when firms are active in different markets where competitive conditions differ, analysing the effects of supply shock in production with reference to only one of the markets may be misleading. In their argument, only price sensitivity and consumption in the Netherlands is referred to. Yet it is known that exports represent more than half of total cheese output in the Netherlands and the output to which the disruption should be compared to is therefore not that of cheese consumption in the Netherlands, but of total cheese production. In this comparison the lost output is at less than [0-5]*% of the relevant output, instead of [5-10]* % as the notifying parties contend. A lower share of disrupted output would result in a smaller price change than a larger disruption.

798. Fourthly, a given firm can outsource a part of the lost production to other plants and suppliers with spare capacity, so as to be able to honour to long term contracts with its customers, reducing the actual loss of total output in the market, due to a natural disaster. It is not clear to what extent the total output of cheese in the Netherlands was indeed affected by the temporary shutdown of Campina's plant. (532)

799. For all the stated reasons, the supply shock analysis provided by the notifying parties is of limited informative value regarding the incentives of the merged entity to raise prices.

8.4.2.13. Conclusion

800. It can therefore be concluded the concentration will lead to a significant impediment of effective competition on the market for the sale of Dutch-type cheese to modern types of retail in the Netherlands.

8.4.2.14. Sale of nature Dutch-type cheese to modern types of retail

801. The conclusion would not be different if the market was defined more narrowly as only comprising nature Dutch-type cheese. The notifying parties have a market share of [60-70]*% for nature Dutch-type cheese (increment of [20-30]*%), and the market structure is similar to the market structure described in section 8.4.2.1, that is to say the notifying parties would be followed by several cheese wholesalers and a Dutch producer with considerably smaller market shares. (533)

802. Furthermore, as nature Dutch-type cheese represents 93% of all sales of Dutch-type cheese at modern types of retail, the considerations in sections 8.4.2.2 - 8.4.2.11 apply a fortiori in the case of nature Dutch-type cheese.

8.4.2.15. Sale of Gouda to modern types of retail

803. The transaction would also lead to a significant impediment of effective competition if the market was defined more narrowly as only comprising Gouda. Similar to the market for Dutchtype cheese, the notifying parties have a high combined market share of [60-70]*% (increment of [20-30]*%). (534) Furthermore, as Gouda represents 87% (535) of all sales of Dutch-type cheese at supermarkets, the considerations in sections 8.4.2.2 - 8.4.2.11 apply accordingly in the case of Gouda cheese.

8.4.2.16. Sale of nature Gouda to modern types of retail

804. The transaction would also lead to a significant impediment of effective competition if the market was defined more narrowly as only comprising nature Gouda. In particular, the notifying parties have a market share of [60-70]*% (increment of [10-20]*%) for nature Gouda cheese (536) and the considerations mentioned in sections 8.4.2.14 and 8.4.2.15 apply correspondingly.

8.4.2.17. Conclusion

805. It is concluded that the concentration would lead to a significant impediment of effective competition on the markets for the sale of Dutch-type cheese to modern types of retail (including narrower segmentations into nature and Gouda cheese) in the Netherlands.

8.4.3. Markets where no competition concerns exist

806. On a market for the sale of Maasdam to specialised cheese wholesalers, the notifying parties have a market share of approximately [40-50]*% (increment of [5-10]*%), followed by Bel ([30-40%]* and DOC [10-20%]*). (537) In a narrower market for the sale of nature Maasdam sold to specialised cheese wholesalers, the notifying parties have a market share of [40-50]*% (increment of [10-20]*%) while Bel has a market share of [50-60%]*. (538) In addition to the presence of several players in the Maasdam market, Maasdam constitutes only approximately 4% of the total Dutch-type cheese market, that is, a small switch of Dutch producers from Gouda to Maasdam would have a significant impact on this market. In view of these reasons, and since no particular concerns have been voiced in the course of the market investigation in relation to Maasdam, it is concluded that the transaction does not rise competition concerns as regards the Maasdam markets in the Netherlands.

807. Similarly, no competition concerns exist in all markets for the sale of rindless Dutch-type cheese (equally if defined more narrowly as rindless Gouda and rindless Maasdam) as is already illustrated by the fact that the notifying parties' combined share in the German and Dutch production of rindless Dutch-type cheese is significantly less than [20-30]*%. (539) On all markets, the notifying parties face important competitive constraints, in particular from DOC and German cheese producers. Accordingly, no complaints were received in the market investigation as far as rindless cheese is concerned.

8.4.4. Conclusion on competitive assessment

808. For these reasons, it is concluded that the concentration will lead to a significant impediment of effective competition on the following markets: Dutch-type cheese sold to specialised cheese wholesalers in the Netherlands (or narrower segmentations into nature, Gouda, and 15-day-old cheese); and Dutch-type cheese sold to modern types of retail in the Netherlands (or narrower segmentations into nature and Gouda). Each of these markets constitutes a substantial part of the common market.

9. BUTTER

9.1. Introduction

809. Butter is roughly what is left of milk when all other non-fat components have been removed; economically, it is a by-product. To produce butter, cooled down cream is churned using a continuous butter machine. The sweet buttermilk is drained and collected in a tank after pasteurisation (unlike the buttermilk drunk by consumers, this sweet buttermilk is used merely as a source of protein). The butter is worked in a kneading section. During this process, cultures of lactic acid bacteria are added to the butter. The butter is packed in 25 kilogram cartons (bulk butter) or in 250 gram wrappers (packet butter).

810. The standard way of supplying bulk butter is in 5 kg and 25 kg cartons. These are delivered frozen or refrigerated. Bulk butter is made up of 82-84% fat (basic bulk butter) or 99.8% fat (butter oil). The latter is simply a purer form of butter than the former with less water and fewer minerals.

811. By fractionating butter, it is possible to "sort" the butter particles in accordance with their melting range; for example, butter with a melting range starting at 10, or 20, or 45 C°. Likewise, it is possible to produce extra white butter or (with the addition of carotene) extra yellow butter. Fractions (fractionated butter) may be sold pure or in addition to butter which has not been fractionated (to increase the ease of spreading it on bread at room temperature).

812. Bulk butter is used in the bakery, ice-cream and chocolate industries and various other segments of the food processing industry.

813. Packet butter is the same basic product as bulk butter and therefore the production process is entirely the same (except that packet butter is packaged in consumer size packaging). Packet butter is packaged in wraps, rolls, cups, and tubs. Wraps are the common 250 gram brick-shaped paper wrapped packages. Rolls are the traditional (now almost out-dated) round-shaped 500 gram paper wrapped packages. Cups are the small 10 or 15 gram plastic packages which are used in restaurants and canteens. Tubs are mostly 250 gram plastic oval tubs. Wraps, rolls, cups and tubs are all offered to retailers and rolls and cups are offered to the OOH sector. Also 2.5, 5 and 10 kg (and some 25 kg) packages are sold as packet butter to the small industrial bakeries via the OOH segment.

814. Within the packet butter market, traders/packaging companies, such as Kaptein and Van der Pol, play an important role. The packaging companies do not produce butter themselves but buy bulk butter which they then process into the required volumes and composition and package into consumer packages which are sold under private labels via retail and OOH sales channels.

815. With respect to the distinction between bulk butter and packet butter, the notifying parties submitted that in case COMP-M.3130 - Arla Foods/Express Dairies, (540) the supply of packet yellow fats and the supply of bulk yellow fats were distinguished as two separate product markets. The notifying parties agree with this distinction.

816. The market investigation has confirmed the basic distinction between bulk butter and packet butter, and thus the two markets will be assessed separately.

9.2. Bulk Butter

817. Both Campina and Friesland Foods produce and sell bulk butter (including butter oil and butter with special fractions) throughout Europe and outside Europe. On the other hand, neither Campina nor Friesland Foods are active in manufacturing or selling margarine or vegetable oils in Europe (the blends Campina sells to South-East Asian customers contain vegetable oils as well as butter). Any possible sub-segments for the supply of margarine or vegetable oils are therefore not affected by the proposed concentration.

818. The activities of Campina on the European bulk butter market are in the field of supplies to the food processing industry, for bakery, ice-cream and chocolate applications. Campina also sells blends: a combination of butter and vegetable oil. Those products are sold to the food processing industry in South Korea, Singapore and Japan. Small volumes of the blends are sold in Belgium, to traders for South-East Asian customers.

819. The activities of Friesland Foods on the European bulk butter market are comparable to those of Campina.

9.2.1. Relevant Product Market

9.2.1.1. Bulk butter and vegetable fats do not belong to the same product market

9.2.1.1.1. Product market proposed by the notifying parties

820. The notifying parties noted that, as regards the market for bulk yellow fats, the notifying parties in Arla Foods/Express Dairies contended that the market for bulk yellow fats encompasses butter, margarine and liquid vegetable oil. In that case, however, it was finally concluded that the market for bulk yellow fats could be subdivided into at least a separate market for bulk butter. The notifying parties submitted that they will use this market definition as their reference.

821. The notifying parties, however, also indicated that higher butter prices have caused food processing companies to look for substitutes for butter. Several food processing companies have transferred their butter requirements to vegetable fats within a couple of months.

822. The notifying parties expect that customers who switched to the cheaper vegetable fats due to an increase in the butter price would not return to butter if the price of butter fell back to the level before the switch, as it takes a while before all packaging materials that are in stock to be used up and sold. Before that, it would not make sense to switch.

823. Therefore, the notifying parties submit that, although vegetable fats are not in the same market as butter, significant competitive constraints are exerted on the butter market by vegetable fats.

9.2.1.1.2. Assessment of the Commission

824. During the market investigation, it was verified whether customers consider vegetable fats as a substitute. The result of the inquiry was that customers consistently rejected a potential switch to vegetable fats, if prices of bulk butter should rise by 5-10%. (541) In addition, a number of producers noted that their customers would not switch to vegetable fats and a price increase of bulk butter in excess of 10% (about 40%) would be necessary to persuade them. (542)

825. It is, therefore, agreed that bulk butter and vegetable fats belong to separate product markets.

9.2.1.2. Basic butter, non-fractionated butter oil and fractionated butter oil belong to separate product markets

9.2.1.2.1. Product market proposed by the notifying parties

826. The notifying parties submit that bulk butter is used in the bakery, ice-cream and chocolate industries and various other segments of the food processing industry. Food processors that prefer a more consistent quality (less varying constitution) may prefer the 99.8% purity (butter oil). Generally speaking, however, 82% fat butter and 99.8% fat butter are used for similar purposes. Many applications also involve water, making the difference in applications small. It is not more difficult for the food processing industry to switch from 82% fat composition butter to 99.8% fat composition butter once they choose for butter or butter oil. According to the notifying parties, the production lines in the food processing industry can process both butter and butter oil without any additional switching costs.

827. As to supply side substitutability, the notifying parties argue that the largest producers of butter in Europe produce both 82% and 99.8% fat butter (butter oil and other bulk butter). Although these kinds of butter are produced on different production lines, they are generally produced in the same factory. Such producers typically shift production towards 82% fat levels or 99.8% fat levels, depending on demand and the potential to charge a higher price for one or the other. This supply side substitutability would cause the prices of both kinds of butter (and fractioned butter) to move in parallel.

828. With respect to fractionated butter, the notifying parties argue that it is priced slightly higher than the sum of the butter oil and the cost of fractionating it. It is not a particularly technologically advanced market and butter oil producers can easily start producing butter fractions without any specific knowledge of the process, be it at the extra cost of the fractionating equipment.

829. Therefore, the notifying parties submit that supply side substitutability means that the 82% fat butter, the 99.8% fat butter and fractionated butter form a single market.

9.2.1.2.2. Assessment of the Commission

830. With respect to the question whether, within bulk butter, basic butter (with a 82% fat content), butter oil (with a 99.8% fat content) and fractioned butter oil all belong to the same market, the results of the market investigation contradict the submission by the notifying parties.

Demand-side substitutability

831. The market investigation indicated that there is limited substitutability between basic butter and butter oil. (543) In particular, a majority of customers (for example, chocolate and ice- cream manufacturers) are currently using only butter oil and could not switch to basic butter or fractionated butter oil, without devoting substantial resources to re-engineering their recipes. This is due to the fact that basic butter and butter oil differ for taste, functionality and performance. (544) In some cases, switching may also be prohibited by law. (545)

832. Secondly, the market investigation indicated that most customers consider fractionated butter oil non-substitutable with non-fractionated butter oil or other butter varieties. One customer, with respect to fractionated butter, explained that "applications and functionalities are completely different; products of a low or high melting point cannot be replaced by a standard product". (546)

Supply-side substitutability

833. With respect to supply-side substitutability, it has been investigated whether it is possible, for butter producers which are not producing butter oil or fractionated butter, to swiftly switch to production of those butter specialties. On this point, respondents to the market investigation have clearly and consistently indicated that there are high switching costs to bear and switching cannot, therefore, easily occur. (547) In particular, production lines used to produce non-fractionated butter oil cannot be used to produce fractionated butter oil and the setting up of fractionation equipment is an expensive investment. (548) One of the producers responding to the market investigation estimated that the investment necessary to set up a fractionation line is approximately EUR 1 million -1.5 million and entails a 6-12 month timeframe. (549)

9.2.1.3. Conclusion on the relevant product market

834. Therefore, based on the above considerations, it is concluded that dairy bulk butter belongs to a separate market to that of bulk vegetable fats. In addition, dairy bulk butter can be divided into basic butter, non-fractionated butter oil (or, simply, "butter oil") and fractionated butter oil (or, simply, "fractionated butter").

9.2.2. Relevant Geographic Market

9.2.2.1. Geographic market proposed by the notifying parties

835. With respect to the geographic scope of the market for bulk butter, the notifying parties argue that, contrary to what it has been considered in Arla Foods/Express Dairies, the geographic market definition is more likely to be worldwide or at least EEA-wide than national. The notifying parties emphasise that bulk butter is considered to be a high value product in relation to volume, which means its transport costs are relatively low. Imports from outside the EEA are common. In addition, the notifying parties note that, due to the fact that the Community decreased subsidies for butter, the market for bulk butter has become more worldwide in scope.

9.2.2.2. Assessment of the Commission

836. The market investigation confirmed that all suppliers with substantial volumes (such as Uelzena, Lactalis, Flechard, Corman) serve customers on an EEA-wide level. (550) Almost no supplier participating in the market investigation considers that sourcing patterns would differ across the EEA. Suppliers also widely agree that transport costs do not to play a significant role. (551) Similarly, approximately half of the customers participating in the market investigation source from EEA Member States other than Belgium, the Netherlands and Germany, or have indicated that they consider the market to be EEA-wide. (552) One supplier explained that customers and producers would increasingly act on a Community-wide scale or even globally, and that regional purchases are less and less frequent. Furthermore, it should be borne in mind that brands do not play role in the sale of bulk butter and no particular national preferences have been articulated as to the origin of bulk butter. The market investigation also indicated that sourcing patterns are the same for basic butter, butter oil and fractionated butter oil. (553)

837. Therefore, it is concluded that the market for bulk butter, fractioned butter oil and nonfractioned butter oil is EEA-wide and the effects of the transaction will be assessed accordingly.

9.2.3. Competitive assessment

9.2.3.1. Market position of the notifying parties in the bulk butter market

838. With respect to bulk butter, the notifying parties estimate their EEA-wide market shares to be [10-20]*% in basic butter (with 82% fat content), [30-40]*% in butter oil fractioned and [20-30]*% in non-fractioned butter oil.

839. These market shares indicate that the EEA markets for fractionated and non-fractionated butter oil are affected markets. However, such market shares remain well below [40-50]*%. In addition, on each of these markets, the notifying parties are under competitive pressure by two strong competitors with double-digit market shares and several smaller competitors with market shares above [0-5]*%.

840. Indeed, on the fractionated butter oil market, the notifying parties have provided the following market shares for their competitors: Corman [10-20]*%, Lactalis [10-20]*%, Beuralia [5-10]*%, Ulzena [5-10]*%, Flechard [5-10]*%, Ilas [5-10]*%, VIV [0-5]*% and Glanbia [5- 10]*%.

841. With respect to the non-fractioned butter oil market, the notifying parties provided the following estimates: Corman [20-30]*%, Lactalis [10-20]*%, Beurralia [5-10]*%, Ulzena [5- 10]*%, Flechard [0-5]*%, Ilas [0-5]*%, Meadow Foods [0-5]*%, VIV [0-5]*%, Bayernland [0- 5]*%, Arla [0-5]*%, Glanbia [0-5]*%, Roil [0-5]*% and Saumweber [0-5]*%.

9.2.3.2. Ability of customers to switch to alternative suppliers

842. In order to assess the profitability of a price increase post-merger, it has also been investigated the ability/willingness of bulk butter customers to switch to alternative suppliers and to rely on multiple sourcing.

843. In this respect, while some customers indicated that switching suppliers would be difficult, the majority of customers appear indeed able to switch suppliers in case of price increases and without any obstacles and/or have already done so in the past, mainly due to prices. Customers have switched to alternative suppliers, such as Corman, Uelzena, Bayernland, Meggle and Twente Foods. (554) In addition, the market investigation indicated that the majority of respondents rely on multiple sourcing for their bulk butter supplies. (555) Indeed, some respondents have replied that they do not expect anticompetitive effects from the merger, due to the availability of "several potential suppliers". (556)

844. It is, therefore, considered that there are alternative sources available to customers.

9.2.3.3. Spare capacity available

845. In addition, producers appear to have significant spare capacity available. Spare capacity is a characteristic of the butter market, since demand is seasonal. Some butter producers responding to the market investigation have indicated that there is indeed spare production capacity, although little milk allocated to butter production. This is the result of a market choice to use the majority of raw milk to produce cheese and milk powder. (557) Customers also have indicated that they do not expect an impact on the merger on the bulk butter price, since there is "enough backup capacity" on the market. (558)

9.2.3.4. Entry barriers

846. Although customers responding to the market investigation have provided mixed replies, several envisage possible new entries (First Milk from the UK, Brand Dairy, Omira) on the market after the merger. The estimated timeframe for such entry is estimated to be between a few months and 1 year. (559)

9.2.3.5. Conclusion

847. In view of the market shares of the notifying parties, the structure of the bulk butter market with a variety of big and small competitors, and the possibilities for entry and expansion, the notifying parties would not be able to exercise any market power post-merger on the bulk butter markets at issue.

848. For these reasons, the concentration would not lead to a significant impediment of effective competition in the field for basic bulk butter, for fractionated and non-fractionated butter oil in the EEA.

9.3. PACKET BUTTER

849. Both Campina and Friesland Foods produce and sell packet butter. However, their presence in the various submarkets and market segments differs.

850. Friesland Foods only sells small volumes of packet butter via the retail channel (only branded) in the Netherlands (less than EUR [...]* per year) and the UK and a substantial volume of branded packet butter in Germany (via Campina as distributor). Friesland Foods also sells branded packet butter via the OOH channel in the Netherlands. It sells no PL packet butter anywhere in the Community.

851. Campina sells branded packet butter: "Botergoud" and "Elke Dag" in the Netherlands (and small quantities in Belgium) and "Beste Butter Frau Antje" (produced by Friesland Foods), "Landliebe", "Tuffi", "Buttergold", "Leichtgold" and "Mark Brandenburg" in Germany. Campina also sells packet butter for private label use. Campina sells packet butter through the retail sales channel as well as the OOH segment.

852. Friesland Foods only produces branded packet butter under the "Beste Butter Frau Antje" brand (which is distributed by Campina in Germany) and the "Nedgold" brand in the Netherlands and the "Wheel barrow" brand in the UK, but does not sell any packet butter in the Community for PL use. Friesland Foods sells packet butter via OOH channels under the brand name "Hollandia" (the Netherlands) and "Debic" (Belgium). Friesland Foods also sells butter oil via OOH channels under the brand name "Congres", mostly to the small industrial users such as bakeries. Friesland Foods has minor sales through the retail channel.

9.3.1. Relevant Product Market

9.3.1.1. Packet butter and vegetable fats each belong to single separate product markets

9.3.1.1.1. Relevant product market proposed by the notifying parties

853. The notifying parties acknowledged that in Arla Foods/Express Dairies, packet butter and margarine were not considered to be part of the same product market. Their prices do not show a significant correlation. However, the notifying parties believe that there are serious competitive constraints from margarine being exerted on packet butter.

9.3.1.1.2. Assessment of the Commission

854. During the market investigation, customers were asked whether they would switch to vegetable fats in the case of a price increase of packet butter. The outcome of the investigation revealed that switching would not occur (due to taste and other product characteristics) or would only occur in the case of a price increase well above 10%. (560) 855. Therefore, it is concluded that vegetable fats (namely, margarine) are not in the same market as packet butter, given the extent to which prices would need to increase before inducing consumers to switch.

9.3.1.2. It can be left open whether branded and private label packet butter belong to the same market

9.3.1.2.1. Relevant product market proposed by the notifying parties

856. With respect to packet butter, the notifying parties acknowledged the Commission's position that a possible further subdivision could be made between branded and PL packet butter.

857. The notifying parties do not agree with this subdivision for packet butter and submit that private label and branded packet butter both belong to the overall packet butter market although the price and quality levels differ. Consumers are brand loyal but more focused on price developments. The price difference is largely due to differences in functionality (easier to spread) and taste. That they belong to the same market is evident from parallel developments in price and from the effect on relative sales volumes caused by changes in relative prices.

9.3.1.2.2. Assessment of the Commission

858. In the packet butter market in the Netherlands, Belgium and Germany, private labels represent slightly less than 50% of the retail market (EUR 518 million of a total market of EUR 1 076 million) and approximately 38% of the OOH market (EUR 76 million of a total market of EUR 200 million).

859. Whether private labels and supplier brands belong to the same product market upstream depends on multiple factors. In particular:

(a) Whether both types of brands, in general, compete closely with each other from the perspective of the end-customer; and (b) The extent to which upstream suppliers of private label and/or brands as well as the purchasing retailers, take into account in their negotiations upstream the competitive pressure that private labels and supplier brands mutually exert on each other at consumer level.

860. The market investigation revealed that, according to competitors, branded and private label butter compete downstream. (561) Retailers indicated that it is difficult to differentiate branded and private label butter in any respect other than price which, therefore, is the main competition factor for private labels. Upstream, the procurement process is in principle different for brands, which are negotiated bilaterally, and for private label products, which are tendered. (562)

861. As to the upstream competition, the market investigation has not indicated that private labels exercise any constraint on branded products, nor provided any information as to the competitive interaction between the two brands upstream.

862. However, it was concluded that, for the purpose of assessing the competitive effects of the notified merger, the issue of whether branded and private label packet butter belong to the same market can indeed be left open, as it does not have any determinant impact on the outcome of the Commission's analysis.

9.3.1.3. The retail sales channel and the OOH sales channel each represent a separate product market

9.3.1.3.1. Relevant product market proposed by the notifying parties

863. Both notifying parties sell packet butter via the retail channel (Friesland Foods mostly in Germany via Campina). Both Notifying parties also sell packet butter via the OOH channel, where they sell cups and "kluiten" which are 2.5, 5 and 10 kilogram packages of butter. Friesland Foods also sells some 25 kg cartons to the OOH segment.

864. The notifying parties recall that, in the Friesland Coberco-Nutricia Decision, (563) a distinction was made between the retail and food service (OOH) distribution channels, in reference to the Unilever-Bestfoods Decision. (564) According to the Commission, "these distribution channels are considered as different markets, because of distinguishing features including a service dimension, separate sales forces, different price structures, different packaging sizes and different health and safety regimes." In the subsequent Arla Foods-Express Dairy Decision, however, no distinction between the various distribution channels was made. The notifying parties note that accordingly, no separate product markets for retail and food service were established and that that Decision explicitly addressed the butter market.

865. The notifying parties further argue that, in the packet butter market, the packages sold via the retail and the OOH channels are different: mainly 250 and 500 grams packages for retail, portion cups and 2.5, 5 and 10 kg for OOH. However, according to the parties, this does not necessarily mean that sales via the retail and OOH belong to separate product markets. The notifying parties point out that packet butter sold via the OOH channels amounts to only 7% of the total sales of packet butter. Moreover, according to the notifying parties, the prices used in the different segments show a high correlation.

866. Therefore the notifying parties submit that no distinction should be drawn between the retail and OOH markets.

9.3.1.3.2. Assessment of the Commission

867. With respect to the distinction between the retail and OOH sales channels, the market investigation sought to verify whether there are difference between them in terms of brands, services provided to customers, packaging and prices.

868. The market investigation showed that there are specific brands for OOH customers (565) and also smaller "one use" packages, while retailers mostly sell 200 gram and 500 gram wraps. Such difference in packaging also reflects on the price, as smaller packages are proportionally more expensive than large formats. (566) OOH wholesalers, in particular, have consistently indicated that the difference between the retail channel and the OOH channel relate to brands, to the fact that in the OOH butter is sold with additional services, to packaging and to prices. (567) In addition, butter producers have explained that the additional services provided to the OOH customers consist of technical support and that also the type of contracts made with OOH customers are different. (568)

In particular, OOH customers seem to rely more on mid- and long-term supply contracts (covering the selling season at least). (569)

869. During the market investigation, it was asked whether OOH customers are able to cross-source their butter supplies from retailers. One OOH customer explained that, while this is possible in theory, in practice the most important packaging formats - 10 gr. and 5kg - will have to be sourced from third parties. Therefore, for logistic reasons, OOH customers tend to source the whole of their supply from suppliers which have the whole packaging range. (570)

9.3.1.4. Conclusion on the relevant product market

870. In light of these considerations, it is concluded that the market for packet butter must be further divided into packet butter sold to retailers and packet butter sold to OOH customers. It can be left open whether branded and private label butter belong to the same market, as the distinction will not have an impact on the competitive assessment.

9.3.2. Relevant Geographic Market

9.3.2.1. Relevant geographic market proposed by the notifying parties

871. The notifying parties submit that the market for packet butter is at least EEA-wide, or even worldwide as suggested by some respondents to the market investigation conducted by the Commission in Arla Foods-Express Dairies. The geographic market definition may be worldwide due to butter being a relatively high value product in relation to volume which reduces transport costs in relation to its value and therefore imports from outside the EEA would be possible.

9.3.2.2. Assessment of the Commission

872. The investigation focused on the territorial scope of the packet butter business, inquiring on the location of butter customers and suppliers and on cross-border butter sales. The investigation shows that currently both retail and OOH customers currently source packet butter from their own Member States. (571) However, almost all respondents have noted that they would source from neighbouring Member States, should the price increase by 5-10% in their home market. (572) This statement is valid for both branded and private labels. (573) The respondents to the market investigation have also pointed out that imports from outside the EEA do not play a role in these markets. (574)

873. Some butter suppliers have indicated that they supply retail and OOH customers throughout the EEA, which indicates that the market for packet butter might even be EEA-wide in scope.

874. In light of these considerations, it is concluded that the relevant geographic market includes at least Belgium, Germany and the Netherlands. The question whether the relevant geographic market for packet butter is EEA-wide can be left open as this conclusion will not have a determinant effect on the Commission's competitive assessment.

9.3.3. Competitive Assessment

9.3.3.1. Market shares of the notifying parties in the packet butter market

875. The market shares of the notifying parties in the market for packet butter in Belgium, Germany and the Netherlands (taken together) and in the EEA respectively are detailed in the Tables 9.1 to 9.4.

Table 9-1: Market shares in retail, Belgium, Germany and the Netherlands - Source: Form CO.

Table 9-2: Market shares in retail, EEA - Source: Form CO.

Table 9-3: Market shares OOH, Belgium, Germany and the Netherlands - Source: Form CO.

Table 9-4: Market shares OOH, EEA - Source: Form CO.

876. The figures reported in Tables 9-1 to 9-4 show that, regardless of the market definition, there is no overlap between the notifying parties in the retail market.

877. In the OOH market for packet butter, there are affected markets, regardless of the geographic definition of the market. The effects of the merger, in terms of market shares, appear slightly higher, although still not high enough to raise concerns, if the market for branded packet butter is considered as a separate market. However, there is still a large contestable portion of the market and competitors with relevant market shares exert competitive pressure on the notifying parties. In addition, it must be taken into account that private label are also very well established in the OOH market and currently amount to approximately 39% of the total OOH packet butter market in volume in the market including Belgium, Germany and the Netherlands and 50% in the EEA market.

9.3.3.2. Ability of customers to switch to alternative suppliers

878. The market investigation explored switching possibilities for OOH customers after the merger. Firstly, Campina and Friesland Foods are generally not seen as each other's closest competitors by their other competitors on the market for packet butter. (576) A few OOH customers indeed consider the notifying parties as closest alternatives to each other. Their replies, however, also indicated that there are no long term supply contracts in this segment and often, there are no formal contracts at all. (577) Since prices are often indexed to the quotation of the Dutch butter quotation committee, (578) supplies are negotiated at very short intervals, thus providing the opportunity for customers to switch.

879. The replies of OOH wholesalers also indicated that in the past they have either already switched supplier for price-related reasons (579) or could switch in the future (580). Therefore, no particular hurdles to switching exist. 880. In addition, OOH customers indicated during the investigation that they rely on multiple sourcing, in order to get the best quality products at the best possible price. (581)

881. It is, therefore, considered that, in case of a post-merger price increase by the notifying parties, customers would have sufficient sourcing alternatives.

9.3.3.3. Spare capacity available and new entry on the market

882. During the investigation, spare production capacity on the market for packet butter was also assessed. In this respect, respondents indicated that they have unused capacity available for the production of packet butter. (582) It was inferred that the aggregate average amount of spare capacity is estimated at approximately 40%-50% of each producer's total capacity.

883. The investigation revealed that OOH customers foresee that, in the case of a price increase, the notifying parties' competitors are able to promptly expand production (583) and meet the demand for packet butter at a lower price. In particular, viable alternatives mentioned include the packaging company Van der Pol and Irish butter producers, but also niche players which could expand locally.

884. In addition, the notifying parties have also indicated that new entrants can access the packet butter market by simply sourcing bulk butter and outsourcing the packaging to packaging companies, such as Kaptein and Van de Pol. During the market investigation, this possibility was explored and the two packaging companies concerned have confirmed that they have enough spare packaging availability to meet the demand of new entrants. (584)

9.3.3.4. Conclusion

885. For these reasons, the concentration is not likely to lead to a significant impediment of effective competition in the EEA-wide market of packet butter sold to retail and OOH customers, nor on a narrower geographic market including only Belgium, Germany and the Netherlands.

10. VALUE ADDED YOGHURT AND QUARK

886. Friesland Foods and Campina are both active in the production and sale of value added yoghurts. Campina produces and sells quark, while Friesland Foods is only active in the sale (but not the production) of quark.

887. In the Netherlands, Campina (585) sells value added yoghurt and quark mainly under the Campina and Mona brands as well as under private labels. In Germany, Campina is also active in various brands and as a private label supplier, while in Belgium only the Campina brand is present. Finally, Campina also sells value added yoghurt and quark in other Member States such as Spain and United Kingdom.

888. Friesland Foods sells value added yoghurt and quark under the Friesche Vlag brand in the Netherlands only. Friesland Foods ended its private label production in 2007. (586)

10.1. Relevant Product Markets

889. According to the notifying parties (587) value added yoghurts and quark comprise a heterogeneous group of products, including products with health claims and products with a health appeal ("health yoghurts") and products which do not have such claim or appeal and are primarily intended for indulgence ("indulgence yoghurts"). Both kinds of yoghurts are sold in various flavours, colours and with various additives (most notably fruits).

890. Unflavoured yoghurt and unflavoured quark, both in non-gable top packaging, is considered by the notifying parties as value added yoghurt and quark (and since all quark is packaged in nongable top packaging, all quark is included in this category).

891. Value added yoghurts are made from basic plain yoghurt, to which flavours, colouring, sugar, fruits and/or other additives are added. They are normally sold in portion packs of 125 g or 150 g up to 500 ml beakers, and some are sold in 1 litre gable tops.

892. The process of making quark begins with skimmed milk. This milk is heated to a maximum of 30°C. Then the milk is stored and starter cultures are added in the form of milkacid bacteria and rennet causing the coagulation process. Once this process is finalised, the quark is separated from the leftover. After cooling and adding cream, ingredients may be added to the quark, such as sugar, colorants and spices. Fresh fruit is added during the filling process and the portion packages are then packaged and palletised.

10.1.1. The question whether value added yoghurt and quark form one product market or should be separated can be left open

10.1.1.1. Product market definition proposed by the notifying parties

893. The notifying parties submit that because of substantial demand-side substitutability between value added yoghurts and quark, they form a single relevant product market to be distinguished from basic plain yoghurt packed in gable tops. (588)

10.1.1.2. Assessment of the Commission

894. While customers as well as competitors broadly agreed that plain yoghurt in gable top forms a separate market (589), the replies from customers in particular, with respect to whether value added yoghurt and quark form a single product market or should be separated, were mixed. (590) Some respondents indicated that both products target the same customer groups and that the perception of the products as well as their time of consumption is the same. Others, however, pointed to the different tastes, texture and usage between value added yoghurt and quark, concluding them to be separate markets.

895. Based on the results of the market investigation, the Decision leaves the question open as to whether value added yoghurt and quark form a single relevant product market as, in any event, it would not affect the competitive assessment.

10.1.2. A distinction between health related and non-health related value added yoghurt and quark can be left open

10.1.2.1. Position of the notifying parties

896. The notifying parties argue (591) that no distinction should be made between the health segment and the indulgence segment for value added yoghurts and quark, because retailers do not divide these segments and no meaningful difference can be made between health yoghurts and quarks and indulgence yoghurts and quarks from a demand-side perspective. In addition, Friesland Foods is not active in a potential health market for value added yoghurt and quark.

10.1.2.2. Assessment of the Commission

897. The results of the market investigation have been mixed. Half of the respondents (592) considered the market to be a single market given that a delineation is impossible as healthy yoghurts must also be tasty, while the other half were in favour of a separation since both customers' needs and marketing would differ.

898. The question whether value added yoghurt and quark should be further divided into health and indulgence segments can be left open as, in any event, it would not affect the competitive assessment since Friesland Foods is only active in the indulgence segment of the market.

10.1.3. A distinction between private label and branded value added yoghurt and quark can be left open

10.1.3.1. Position of the notifying parties

899. As set out in section 7.1.1.2, the notifying parties agree with a vertical segmentation for the sourcing of products by retailers (and OOH wholesalers) and a downstream market for the sale of dairy products like value added yoghurt and quark by retailers/OOH wholesalers to consumers.

900. With respect to the question whether one should distinguish between the procurement of private label products and the procurement of branded products by retailers (and OOH wholesalers) as two neighbouring, but separate markets, the notifying parties argue that in the context of the markets for value added yoghurt and quark, both Campina and Friesland Foods are under constant pressure from private label products. This is exemplified by the fact that the market is more and more covered by private label products. (593) In addition, the notifying parties argue that despite the presence of strong brands, in particular in the health segment, retailers are in a position to threaten with delisting and have significant bargaining power as the retail segment is heavily concentrated.

901. Thus, private label and branded products belong to the same product market at the upstream level according to the notifying parties.

10.1.3.2. Assessment of the Commission

902. Value added yoghurt and quark are available in two broad categories: brands owned by the dairy manufacturer and private label products which are marketed by the retailer (594). While Campina produces both branded and private label, Friesland Foods decided to exit the private label production of value added yoghurt and quark in 2007.

903. As explained in detail in section 7.1.1.2, the question whether private labels and supplier brands belong to the same product market upstream depends on several factors, in particular: (a) whether both types of brands, in general, compete closely with each other from the perspective of the end-customer; and (b) the extent to which upstream suppliers of private label and/or brands, as well as the purchasing retailers, take into account in their upstream negotiations the competitive pressure that private labels and supplier brands mutually exert on each other at consumer level.

904. For value added yoghurt and quark, there is some evidence to suggest that private label and supplier brands compete in the downstream market and exert a competitive constraint on each other relevant for the purposes of market delineation.

905. In their responses to the market investigation, the majority of customers and competitors confirmed that private label and branded products compete at the retail level and that the quality of private label products is to a large extent similar to branded products (595).

906. On the downstream market for value added yoghurt and quark the share of private label was 26.5% in value terms in 2007 and has declined by 3.9 percentage points over the past two years. (596)

907. There is, however, some differentiation between supplier brands and private labels across the health and indulgence segments. In the health segment there are hardly any private label present - 1.7% in 2007 - although retailers have started to introduce them, whereas in the indulgence segment, the market share has been stable at 36% since 2005.

908. This broad analysis suggests that there is a degree of competitive interaction between supplier brands and private labels at least for the indulgence segment of the value added yoghurt and quark market, while in the health segment, the competitive interaction currently seems to be limited and has not changed at all since 2005.

909. Looking at the upstream level the evidence has been mixed. First, customers and Competitors (597) confirmed that the procurement of branded products follows a different procedure than the one for private label products. In the case of branded products suppliers and retailers agree on the gross price, discounts, listing fees, and promotions in bilateral negotiation. For private labels, a tender procedure is usually used, followed by a selection of preferred suppliers which agree with the retailer on a net price in final negotiations.

910. This in itself is not decisive if the same suppliers are active in both segments. The market presence of both competitors and the notifying parties varies to some extent when it comes to branded or private label value added yoghurt and quark. According to the notifying parties' estimates (598), German competitors like Dr. Oetker, Humana, Mueller, Ehrmann or Bauer supply private label as well as branded products, while Danone and Friesland Foods (since 2008) focus on branded products only. Campina is active in both segments; no supplier seems to focus on private label exclusively.

911. Consequently, there is symmetry in the conditions of competition - with the exception of Friesland Foods and Danone. Given the relative importance of private label sales at least in the indulgence segment of the markets, suppliers cannot be expected to ignore the competitive pressure that private labels exert on their brands.

912. Retailers have private labels in addition to the brands offered by for example Campina, Friesland Foods, Danone, Mueller, or Dr. Oetker, to take a few examples. The existence of private label indicates that retailers consider and condition their private labels in relation to supplier brands.

913. Taken together, these elements, in particular the importance of private label indulgence value added yoghurt and quark to branded products and the competitive interaction between the two brands that affects the negotiations upstream, allow for the conclusion that in that segment, private label and branded products might belong to the same product market upstream. However, for the health segment this does not seem to be valid. In any event, since Friesland Foods is not active in the health segment of the market and does not supply private labels anymore, the question whether private label and branded products belong to the same relevant product market can be left open as it would not affect the competitive assessment of the case.

10.1.4. The market for value added yoghurt and quark should be further separated according to distribution channel into retail and OOH

10.1.4.1. The position of the notifying parties

914. Friesland Foods and Campina both sell value added yoghurt and quark to the retail and the foodservice channel OOH. The OOH segment serves restaurants, cafes, hotels, catering, services, hospitals but also small businesses like bakers or other food processing entities. According to the notifying parties, the OOH customer is typically served by foodservice wholesalers who deliver at the doorstep of the customer and by cash-and-carry business. In the Netherlands, direct sales to OOH users would constitute only a very small proportion of the entire market segment. (599)

915. In previous decisions (600) a distinction was made between these sales channels because of differences in services, sales force, price structure, packaging sizes and health and safety regimes. The notifying parties in the present case argue that such a distinction is no longer appropriate, for several reasons: most of the products sold through the OOH segment are more or less identical to those in the retail segment; both channels partially supply the same customers who cross over depending on the circumstances; in part, players in the segments overlap; prices in both segments show similar developments and the absence of private label in OOH would not be determinative; the logistics of supply and additional services are largely the same.

10.1.4.2. Assessment of the Commission

916. Several respondents during the market investigation indicated that some differences between the retail and the OOH segment still exist.

917. Firstly, it has been confirmed that, contrary to the notification of the notifying parties, distribution in the OOH segment is different from the retail segment in the sense that dairy producers try to target end customers directly and use OOH wholesalers only as logistic providers. (601) Friesland Foods and Campina confirmed (602) that they indeed target end-customers directly and these direct contractual relationships represent [60-70]*% of Friesland Foods annual turnover and about [70-80]*-[80-90]*% of Campina's annual turnover. These contracts cover bonus fees and promotional budgets, but also the price. OOH wholesalers are in charge of delivery. Thus, contrary to the retail segment where the end customer is approached through the retailer, a significant sub-set of final consumers is targeted directly by the dairy manufacturers in the OOH segment.

918. In addition, the market investigation confirmed that packaging sizes and services differ in the OOH segment compared to the retail segment. Clients in OOH demand daily deliveries including the weekend, and special packages which are not available in the retail segment, like small portions consumable immediately after purchasing or larger sizes to be used in the catering or hotel business. At least for these products which for some OOH wholesalers represent up-to 25% of their turnover, alternatives and therefore a cross-over to the retail segment does not seem possible. (603)

919. The service requested from an OOH customer has direct implications for the wholesaler. Since OOH customers expect a timely and flexible delivery from their wholesaler (604), they therefore have to look for a supplier able to mirror the customers' demand - which implies that unless products can be stored over longer periods like long-life dairy or butter such a need favours suppliers close by. (605)

920. Moreover, order volumes in the OOH segment are significantly lower than in the retail segment - notifying parties estimate that 6% of the overall market is covered by the OOH segment in value added yoghurt and quark. Orders by wholesalers are usually small and shipments over longer distances rarely occur as transport costs per unit are too high. This fact seems to limit sourcing possibilities from abroad. (606)

921. Contrary to the argument of the notifying parties, prices in the retail segment seem not to constrain the prices in the OOH segment since OOH clients are not willing to cross-over and buy from a retailer because the additional services will not be offered by the retailer. Demand seems therefore to be inelastic as customers have few alternatives.

922. Several competitors as well as customers have argued that the notifying parties have the possibility to offer "package deals", which means that entering the market with only one dairy product is difficult as OOH wholesalers like to get as much as possible from one supplier for logistical reasons - achieving economies of scale in deliveries, in particular. (607)

923. Contrary to the submission of the notifying parties (608) that a separate sales force would not be necessary to supply OOH - although at least Campina employs 31 persons to target the OOH customers - competitors already present in the Netherlands in the retail segment indicated that additional investments into a distribution network and logistics would be necessary to supply the OOH segment. (609)

924. Finally, average prices for identical products sold to retailers and OOH wholesalers differ. According to internal data covering the best-selling products of Campina and Friesland Foods in value added yoghurt and quark, the price difference between OOH and retail is significant with a [...]*% difference for the Campina product and a difference of [...]*% for the Friesland Foods' product. The difference in prices is also reflected in the difference in gross margins achieved by the notifying parties for identical products sold in retail and OOH. For Campina, these margins are on average [...]*% higher when selling to OOH compared to retail. Friesland Foods' gross margin in OOH is [...]*% above the one achieved in retail. (610)

Table 10-1: Price comparison for identical products in retail and OOH. Source: Campina and Friesland Foods.

925. The notifying parties have argued (611) that the gross margins do not include the fixed costs which are higher for OOH-customers since the sales force and the logistic costs are higher due to the larger number of customers and their smaller size. Taking this into account, comparable margins are evident for both distribution channels. First, the notifying parties have not submitted these corrected margins to support their claim. Second, the notifying parties themselves point towards important distinctions on the demand side between OOH and retail: the OOH segment is less concentrated than the retail side and the order volume per customer is smaller, impacting logistics.

926. On the basis of the replies and the difference in prices the markets for value added yoghurt and quark should be separated according to distribution channels in OOH and retail.

10.1.5. Conclusion on relevant product market

927. On the basis of these elements, it is concluded that there are separate relevant product markets for value added yoghurt and quark according to distribution channels. A separation into value added yoghurt on the one hand and quark on the other, and health on the one hand and indulgence on the other and private label and branded products, can be left open as it would not affect the competitive assessment. As Friesland Foods is not active in the health segment, health value added yoghurt and quark will not be further discussed.

10.2. Relevant Geographic Market

10.2.1. Relevant geographic market proposed by the notifying parties

928. The notifying parties submit (612) that the relevant geographic market for value added yoghurt and quark at the downstream level, where retailers/OOH wholesalers sell to their customers, is national.

929. At the upstream level, the markets for value added yoghurt and quark is wider than the Netherlands and includes Belgium and Germany in any event. This is due to several factors including (i) advances in logistics due to centralised distribution and extended shelf life allowing for transportation over longer distances, (ii) procurement on an international scale and (iii) the availability of Community-wide brands. All these elements would prevent suppliers in the Netherlands from increasing prices by 5-10% as they would lose significant volumes to foreign competitors.

10.2.2. Assessment of the Commission

930. Since a distinction must be made according to the distribution channels within the relevant product market, the analysis of the relevant geographic market also distinguishes the retail market from the OOH market. (613)

10.2.2.1. The retail market

931. The market investigation in the present case allows for the conclusion that the relevant geographic market is wider than the Netherlands for the following reasons. The market investigation confirmed that several foreign producers such as Danone, Novandie, Ehrmann, and Mueller already supply Dutch retailers with value added yoghurt and quark. In total, these imports represent - according to the notifying parties and confirmed in the market investigation - a significant volume of products sold to consumers in the Netherlands (between 30-40%). (614)

932. As value added yoghurt and quark have a longer shelf-life (around 20 days) and are higher value products than, for example, the low margin fresh basic dairy products transportation over longer distances is facilitated. Indeed, almost all competitors indicated (615) that they ship value added yoghurt and quark over at least 500 km within the Community.

933. Moreover, the majority of customers and competitors informed the Commission that unlike in fresh basic dairy where Dutch raw milk has been considered a key input, the origin of raw milk is not important in the case of value added yoghurt and quark. (616)

934. Retailers explained that they usually "work with sales representatives from different foreign suppliers or trading companies" (617) and also use international tenders for the sourcing of private label products. (618) In addition, when asked about potential suppliers, the great majority listed at least four additional producers in addition to the notifying parties. (619)

935. Based on these arguments, it can be concluded that the retail market for value added yoghurt and quark at the sourcing level is wider than the Netherlands and also includes Belgium and Germany.

10.2.2.2. The OOH market

936. Contrary to the wider geographic market definition for the retail market, the conclusion is different for the upstream level (where OOH wholesalers procure their products) because of several market characteristics described in section 10.1.

937. First, foreign competitors - with the exception of Danone - currently supplying the retail market are not active in the supply of the OOH market. In the market investigation, several competitors argued that supplying the OOH market would only be possible after setting up a sales force and logistics. (620)

938. Sourcing from abroad means longer transport distances which lead to higher transport costs. This has a limited effect on the costs per unit sourced if sufficiently high volumes are ordered. However, the volumes usually ordered by OOH wholesalers are small compared to the ones ordered by retailers. The largest OOH wholesaler has a market share of roughly 20% according to the notifying parties, while 42% of the OOH market is covered by wholesalers with a market share below 4%. (621) Based on these elements, it is unlikely that OOH wholesalers would be able to achieve sufficient order volumes not to be affected by the additional transport costs. (622)

939. Alternatively, OOH wholesalers could try to source more than one product from a foreign supplier to achieve sufficient volume. However, not all foreign suppliers offer a complete product range and in addition, they often do not supply branded products that can be considered as alternatives to the brands of the notifying parties, for example in fresh milk or fresh flavoured dairy drinks.

940. Finally, the service requested from an OOH customer has direct implications for the wholesaler. Since OOH customers expect timely and flexible delivery from their wholesaler (623), these wholesalers must therefore look for suppliers able to mirror the customers' demand - which implies that the supplier should be close by. (624)

10.2.3. Conclusion on the relevant geographic market

941. In view of these elements, it is concluded that the relevant geographic market is national for the upstream market of value added yoghurt and quark to OOH wholesalers and wider than national for the upstream market of value added yoghurt and quark to retailers.

10.3. Competitive Assessment

942. In light of the distinction according to distribution channels in retail and OOH, the competitive assessment first analyses the likely effect on the retail market before looking at the OOH market. Within each section, the assessment is based on an overall market for value added yoghurt and quark as the market share and the competitive interaction on both segments is similar and would not lead to different results. In addition, since Friesland Foods is not active in the health segment and therefore no overlap would arise in this potential market, the competitive assessment focuses on the indulgence segment and the overall market (health plus indulgence), each split further into private label and branded products.

10.3.1. The retail market

10.3.1.1. Market structure and market shares

943. According to the notifying parties the downstream retail market for value added yoghurt and quark covering the Netherlands had a total value of EUR 223 267 000 in 2007 with a private label share of 26.5%. The health segment achieved sales of EUR 61 808 000 (private label 1.7%), while the indulgence segment had revenues of EUR 161 459 000 (private label 36.1%). The share of value added yoghurt products was roughly 80%.

944. The notifying parties estimated their combined market share in the indulgence segment (for details see Table 10-2) at [20-30]*%, followed by Danone ([10-20]*%), Albert Heijn's private label brand ([10-20]*%) and the German producer Mueller ([5-10]*%). On the overall downstream market, the combined market share was [30-40]*%, Danone achieved [20-30]*%, while Albert Heijn had [10-20]*% and Mueller [5-10]*%. The increase for the notifying parties and Danone results from the fact that both Campina and Danone are strong in the health segment of the market.

Table 10-2: Downstream market for value added yoghurt and quark, indulgence segment, the Netherlands - Source: Form CO.

945. Looking at the upstream market, where retailers source value added yoghurt and quark from dairy producers, the situation looks rather different:

Table 10-3: Sourcing market for value added yoghurt and quark, based on value, Belgium, German and the Netherlands - Source: Form CO.

946. The merged entity would achieve a combined market share below [20-30]*% with an overlap of less than [0-5]*%: [10-20]*% in the overall market including private label/branded as well as health and indulgence. Within the branded segment the share would result in [10-20]*% and [10- 20]*% in the private label supply. The notifying parties would face several competitors - in particular Danone, the market leader with [20-30]*%, but also among others such as Mueller, Ehrmann and Bauer.

947. In the indulgence segment, the notifying parties would achieve a combined market share of [10-20]*% (private label plus branded) in a geographic market comprising Belgium, Germany and the Netherlands followed by Danone ([10-20]*%), Mueller ([10-20]*%), Ehrmann ([5- 10]*%) and Bauer ([5-10]*%). The combined market shares would only be slightly different if one were to distinguish between private label and branded products - [10-20]*% for branded and [10-20]*% for private label. In all cases the overlap would be limited to [0-5]*% (for details see Table 10-4).

Table 10-4: Sourcing market for value added yoghurt and quark, based on value, indulgence segment, Belgium, Germany and the Netherlands - Source: Form CO.

948. Thus, based on combined market shares in the upstream market, the proposed transaction would lead to an affected market in value added yoghurt and quark in a potential market for indulgence private label products as well as for the branded plus private label indulgence products. On the overall market comprising health and indulgence, only the potential private label market would be technically affected. Given the limited overlap and the presence of several strong competitors for branded as well as private label products, the market shares give no indication of a significant impediment of competition.

949. Moreover, on the potential markets affected by the merger a significant impediment of effective competition is unlikely because of the several additional aspects.

10.3.1.2. Customers have possibilities of switching suppliers

950. Firstly, retailers (625) informed the Commission during the market investigation that they have the possibility to switch to other suppliers. Pre-merger retailers often source value added yoghurt and quark from several suppliers. (626) In addition to their current suppliers - often four different dairy manufacturers - retailers also indicated the set of potential suppliers for these products active in the private label as well as branded segment, which included at least four producers other than Friesland Foods and Campina. (627)

951. Based on these elements, the possibility to switch to other branded or private label suppliers would exist after the merger.

10.3.1.3. Competitors could increase supply if prices increase

952. In addition, the market investigation showed that several main suppliers of value added yoghurt and quark have additional capacity available to be able to supply the market. (628)

953. Some competitors also expect that retailers might further diversify their procurement asa result of the merger and therefore see opportunities to expand their supply. One competitor stated (629), for example, that "this transaction could be an opportunity for us to be an alternative [...] for distributers in both areas yoghurt and quark." Similar a second one who wrote (630) that it would have "more chances to enter NL markets since retailers want to be more independent from the new giant".

10.3.1.4. Conclusion of the competitive assessment

954. For these reasons, it can be concluded that the proposed transaction is compatible with the common market as regards the retail market for value added yoghurt and quark covering Belgium, Germany and the Netherlands regardless of whether this market should be segmented into private label/branded and health/indulgence.

10.3.2. The OOH market

10.3.2.1. Market structure and market shares

955. According to the notifying parties the downstream OOH market for value added yoghurt and quark covering the Netherlands had a total value of EUR 12 228 000 in 2007, with 20% coming from quark products and the remaining 80% from value added yoghurt. (631) The market has slightly lost in value over time.

956. Campina's brand portfolio achieved a market share of [70-80]*% ([5-10]*% Campina brand, Mona [10-20]*%, Optimel [0-5]*%, others [40-50]*%), followed by Friesland Foods' Breaker [10-20]*% and other Friesland Foods brands ([5-10]*%). The Dutch competitor Zuivelhoeve achieved [0-5]*% and Danone [0-5]*%.

957. Since no private label products are sold in the OOH segment, the market shares at the upstream level are identical to those for the downstream one.

10.3.2.2. Non-coordinated effects in the OOH market

958. During the market investigations the Commission received complaints, in particular from customers who believe that the merger will ultimately lead to higher prices because the two leading suppliers of brands would merge and no alternative supplier would be available. Given the strong position of the notifying parties, they would have an incentive to increase prices for their products.

10.3.2.2.1. Notifying Parties have large market shares

959. The larger the market share, the more likely a firm is to possess market power. Moreover, the larger the addition of market share, the more likely it is that the merger will lead to a significant increase in market power. The proposed transaction would combine the current market leader - Campina - with the second strongest player in the market - Friesland Foods - with a combined market share of [90-100]*% with an additional [10-20]*% in the OOH market.

960. Clearly, this significant increase in the sales base on which higher margins can be enjoyed after a price increase makes it more likely that the notifying parties will find such a price increase profitable despite a potential reduction of demand as a consequence of higher prices.

10.3.2.2.2. Customers have limited possibilities of switching suppliers

961. Customers have not only informed the Commission that they view the notifying parties as closest substitutes, but also that they would have difficulties in switching to other suppliers. Since there are only a few alternative suppliers (such as Danone, Zuivelhoeve, Heida, or Boermarke - all of them as suppliers of value added yoghurt and quark only), the OOH wholesalers - mostly sourcing more than one product from the same supplier - would have limited alternatives to the notifying parties. (632)

962. Switching to foreign suppliers has not been perceived by customers as a credible alternative because there would only one be supplier of value added yoghurt and quark active in the Netherlands - Danone. Even Danone is not perceived as a potential alternative as it would only be able to supply one product and not the whole range necessary to achieve economies of scale in deliveries and logistics. (633)

963. In conclusion, switching possibilities would be limited for OOH wholesalers after the merger.

10.3.2.2.3. Lack of countervailing buyer power

964. The notifying parties argue that countervailing buyer power from the large OOH wholesaler restrain them from raising prices, because of the wholesalers' ability to source individual products from abroad. (634)

965. According to the Commissions' horizontal merger guidelines, countervailing buyer power should be understood as the bargaining strength that the buyer has vis-à-vis the seller in commercial negotiations due to its size, commercial significance to the seller and ability to switch to alternative suppliers.

966. However, since the notifying parties are perceived as their own closest competitors (with respect to flexibility and product portfolio) and customers would have only limited switching possibilities post-merger as the notifying parties would hold a quasi-monopolistic position an immediate switch to alternative sources of supply can be excluded. In addition, even if large OOH wholesalers were able to off-set potential price increase, due to their size, this would only cover a segment of the market and not the majority of customers who are rather small in size. Therefore, such hypothetical countervailing buyer power cannot be found to sufficiently off-set potential adverse effects of the merger.

967. Furthermore, even if buyer power existed prior to the merger, it must also exist and remain effective following the merger. This is because the merger of two suppliers may reduce buyer power if it thereby removes a credible alternative. In the present case, the merger will remove one of the two main suppliers in the market, since alternative suppliers with a similar size could not be identified in the course of the market investigation.

968. Therefore, it is concluded that countervailing buyer power post-merger would not be sufficient to off-set potential adverse effects of the merger.

10.3.2.2.4. Entry unlikely to occur

969. The notifying parties argue that in the case of a price increase, other competitors would enter the market, in effect constraining the notifying parties.

970. The market investigation did not confirm whether competitors currently inactive in the OOH market for value added yoghurt and quark in the Netherlands have any intention to enter the Dutch market. Almost all competitors that replied to the market investigation stated no intention to enter the Dutch market and highlighted that entry would be difficult if one were not be able to supply the whole range of products (635) or be flexible in terms of delivery(636).

971. As previously discussed in the relevant geographic market (637), OOH wholesalers argued that transport costs would be higher if they sourced from abroad. The longer supply distance would increase costs for the wholesaler and put the foreign suppliers at a disadvantage concerning flexibility of delivery because Dutch suppliers would always be able to supply on short notice as they are located closer to the OOH wholesalers. Since Dutch OOH customers consider flexibility in deliveries as a key variable in their purchasing decision, Dutch incumbents have a clear competitive advantage.

10.3.2.3. Conclusion on the competitive assessment

972. For these reasons, it is concluded that the proposed transaction would significantly impede effective competition as a result of the creation of a dominant position on the market for value added yoghurt and quark in the Netherlands supplied to the OOH segment.

10.3.3. Overall conclusion on the competitive assessment

973. In light of these considerations, it can be concluded that the concentration is not likely to lead to a significant impediment of effective competition as regards the retail market for value added yoghurt and quark covering Belgium, Germany and the Netherlands, regardless of whether this market is segmented into private label/branded and health/indulgence or not.

974. The proposed transaction would, however, significantly impede effective competition as a result of the creation of a dominant position on the market for value added yoghurt and quark in the Netherlands supplied to the OOH market.

11. FLAVOURED DAIRY DRINKS

11.1. Fresh flavoured and Long-Life flavoured dairy drinks belong to separate product markets

975. The notifying parties submit that fresh-flavoured dairy drinks belong to a separate product market from long-life flavoured dairy drinks. This is because, among other things, fresh dairy drinks have to be transported and stored refrigerated and their shelf life is shorter (generally between 24 and 35 days) than long-life flavoured dairy drinks. In contrast, long-life dairy drinks can be transported and stored at ambient temperatures (that is, non-refrigerated) and their shelf life is longer (six to twelve months).

976. The markets for dairy drinks have been investigated in a number of previous cases. In the case Friesland Coberco-Nutricia, two main categories of dairy drinks were distinguished- health oriented and indulgence oriented drinks- without determining whether these categories constituted distinct product markets (638). The decision stated that health oriented drinks are consumed as a variation to plain milk, often have a short shelf life and are more aimed at adults. The decision also mentions that Friesland Foods' products "Milk and Fruit", consisting of 20% fruit juice and 80% milk falls within this category. The decision also stated that indulgenceoriented dairy drinks are regarded as less healthy because they are not fresh but long life, contain artificial fruit flavouring and/or sugar and are largely aimed at children. Friesland Foods' "Fristi" brand falls within this category, as does chocolate milk. The decision did not explicitly refer to segmentation between fresh and long-life dairy drinks but the arguments mentioned in this decision that distinguish between health-oriented and indulgence-oriented dairy drinks correspond to a segmentation between fresh dairy drinks and long-life dairy drinks.

977. The distinction between fresh and long-life dairy drinks has also been envisaged by the Commission in the Arla Foods/Express Dairies decision (639). In that decision, a distinction was made between a market for fresh and a market for long-life dairy drinks (referred to as flavoured milk). It stated that, from a demand perspective, fresh and long-life flavoured milks have different characteristics as to flavour and pricing, fresh flavoured milks being more expensive than long-life flavoured milks. The decision also identified differences in handling/display of fresh and long-life flavoured milks: fresh flavoured milk having a short shelf life code, whereas long-life flavoured milk are ambient (non-refrigerated) and have a longer shelf life code, making their distribution less time critical. On the supply side, the decision also mentioned different production methods with respect to the main consistent ingredients of flavoured milks (either fresh processed milk or long-life - UHT and sterilized milk).

978. This segmentation has been widely confirmed by competitors and customers questioned during the market investigation (640). Therefore, a distinction is made between fresh flavoured dairy drinks and long-life flavoured dairy drinks.

11.2. Fresh flavoured dairy drinks

11.2.1. Relevant Product Market

11.2.1.1. Fresh Flavoured Dairy Drinks can be segmented further into health and nonhealth product markets

979. Friesland Foods and Campina are both active in the production of fresh flavoured dairy drinks. According to the notifying parties Friesland Foods sells fresh fruit flavoured dairy drinks under the Milk & Fruit brand. In addition, Friesland Foods offers a very small volume of products with a more explicit health appeal under the "TOPontbijt" and the "Fristi Xtra" brands which are sourced from the Austrian dairy company Nöm. All fresh flavoured dairy drinks sold by Friesland Foods are fermented.

980. Campina sells fruit flavoured fresh dairy drinks under the "Campina", the "Mona", the "Vifit", the "Optimel" and "Optiwell" brands, and some other minor brands. According to the notifying parties the Vifit products carry the most outspoken health image, whereas Optimel also have a clear health connotation. Optimel Control is a new range of products aimed at active weight management. Most products sold by Campina are fermented, a small part of the portfolio is not (chocolate and coffee flavoured drinks and some flavoured milk drinks). Of Campina's total turnover in fresh dairy drinks in the Netherlands, non-fermented drinks represent only 7%.

11.2.1.1.1. Product market definition proposed by the notifying parties

981. The notifying parties submit that although the market for fresh flavoured dairy drinks is heterogeneous, it constitutes a single product market. The differences between fermented and non-fermented dairy drinks - in terms of taste, consistency but also price - are hardly noticeable if at all. Most producers supply a product portfolio including products with a more and less pronounced health image. Variations in fat and sugar contents contribute to the differentiation into products with an explicit health claim, other appealing to a health concern and products with no health appeal.

982. According to the notifying parties, a further segmentation into health on the one hand and indulgence on the other is not necessary within the fresh flavoured dairy drinks market.

11.2.1.1.2. Assessment of the Commission

983. The market investigation confirmed that fresh flavoured dairy drinks are offered in a wide range of variety and with different characteristics (641). Some drinks have an explicit health claim such as Danone's Activia (improving digestion), Campina's Vifit (aims to reduce cholesterol) or Yakult drinks (improving digestion, other drinks appeal to a reduction in weight (Optimel from Campina or Vitalinea from Danone) and some drinks have no explicit health claim.

984. The market investigation has found strong indications for a separation of the fresh flavoured dairy drink market into two distinct markets: one with a functional or health related claim and another non-health oriented market. The separation is confirmed by a detailed analysis of scanner data by IRI Nielsen covering all fresh flavoured dairy drinks transactions at the retail level in the Netherlands - excluding the discounters Aldi, Lidl and Koopconsult - during the period March 2004 - May 2008.

985. IRI Nielsen distinguishes four segments within the fresh flavoured dairy drink category:

(a) "Functional" - implying that these products might be able to improve the health of the consumers, for example drinks with additional vitamins, calcium or bifidus cultures. Danone's Actimel and Activia, Yakult, Campina's Vifit and Optimel Control. Some private label products of Albert Heijn and Superunie have also been allocated to this segment.

(b) "Kind" - covers all drinks specially made for children excluding baby drinks which are not dairy-based (642). Some of Campina's Mona products are in this segment, some produced by Danone, products under the Campina brand as well as private label products (Albert Heijn and Superunie).

(c) "Weight management" - covers all drinks with less or no sugar added, no fat or indications that their consumption would allow the consumer to keep his/her weight. Brand in this segment include Danone's Vitalinea, Campina's Optimel, Friesland Foods and private label (Albert Heijn and Superunie).

(d) "Regular" - includes all fresh flavoured dairy drinks not covered by the three specific categories. Some Campina brand products, like the Mona brand, and Friesland Foods' Milk&Fruit can be found here.

986. Comparing that segmentation with the classification submitted by the notifying parties shows that their "explicit health claim" seems to coincide with IRI Nielsen's functional segment. In contrast, the products the notifying parties identify with a "health appeal" are classified by IRI in the weight management category. IRI's categories kind and regular together address what the notifying parties call "no explicit claim".

987. A descriptive analysis of the IRI scanner data provides evidence that the "explicit health claim" segment (corresponding to IRI's category "Functional") forms a product market separate from all the other fresh flavoured dairy drinks belonging to the other three IRI's segments.

988. First, as can be seen in Table 11-1, the brands (and the producers) differ in each category . Producers seem to focus and concentrate on particular segments. Danone and Yakult mainly target the functional segment; Friesland Foods is hardly active in the functional segment but rather concentrates on the regular one and has some activities in weight management. Only Campina uses several brands to cover all four categories, but no brand cuts across all segments at the same time.

Table 11-1: Distribution of value sales per segment for different fresh flavoured dairy drink brands, retail, Netherlands - Source: IRI.

989. Second, prices also differ significantly per unit of volume between functional products and non-health oriented ones. As can be seen in Figure 11-1, functional products are able to achieve an average price per unit of volume which is at least twice as high as the average price charged in the other categories. Prices of regular, weight management and kind fresh flavoured drinks are within the same price range (643).

Figure 11-1: Mean average volume price per segment fresh flavoured dairy drinks, retail, Netherlands - Source: IRI.

990. Third, not only do brands and prices differ significantly, the packaging also varies in particular between the health and non-health segment. While in the health segment small packaging sizes of 65ml and 100ml have experienced an increase in their market shares since 2005, and the largest volume available is 500ml only, the other segments, especially the kind and weight management segments, are dominated by the 1 litre gable top packaging.

Figure 11-2: Distribution of packaging formats for fresh flavoured dairy drink categories. Source: IRI

991. Fourth, the kind, weight management and regular segments are dominated by flavoured offers (as opposed to plain or natural). In fact, the majority of the functional or health segment corresponds to sales of dairy products with no flavour at all, as shown in Figure 11-3.

Figure 11-3: Distribution of flavours across segments, retail, Netherlands - Source:IRI.

992. Finally, the notifying parties argue that advertisement and marketing expenses for fresh flavoured dairy drinks focus entirely on the functional and weight management segment (644). However, at least with respect to marketing at the retail level all four segments are heavily promoted (relative to fresh basic dairy) as shown in Figure 11-4. The graph in Figure 11-4 shows the percentage of total volume on sale subject to either price promotion or some kind of feature and display promotion (for example end-of-aisle positioning or advertisement in the store magazine - weekly or bi-monthly). The only point worth highlighting is that the weight management segment relies relatively more on feature and display promotions than average.

Figure 11-4: Percentage of total volume on sale subject to either price promotion or some kind of feature and display promotion - Source: IRI.

993. Based on the significant difference in price, packaging, brands and advertisement costs between the functional segment on the one hand and the non-functional segment on the other hand, it is concluded that fresh flavoured dairy drinks should be separated into two markets: a market for functional (or health) fresh flavoured dairy drinks and another one for non-functional fresh flavoured dairy drinks including the categories weight management, regular and kind used by IRI. Contrary to the submission of the notifying parties who categorise Campina's Optimel in 1 litre gable top into the health segment, this particular product is viewed as a fresh flavoured dairy drink with less sugar/no fat and without an explicit health claim. Since its price and the packaging are also clearly distinct from the health related fresh flavoured dairy drinks, it is considered to belong in the non-functional market.

994. In conclusion, two relevant product markets in fresh flavoured dairy drinks are distinguished - one for functional or health-related products and another one for non-health related products.

11.2.1.2. Private label and supplier brands of fresh flavoured dairy drinks belong to separate upstream product markets

11.2.1.2.1. Product market definition proposed by the notifying parties

995. The notifying parties agree in principle to a vertical segmentation of a particular product market for consumer products between an upstream market for the sourcing of products by retailers (and OOH wholesalers serving hotels, restaurant, catering services and filling stations) and a downstream market for the sale by retailers to consumers for two reasons. (a) Firstly, the approach would recognise both that on the retail-to-consumer level private label products compete with (producer) branded products, and that the market share of private label products should be attributed to the retailers that own the private label brands. (b) Secondly, the approach would recognise that the competitive conditions on the markets on which retailers and OOH wholesalers source their products are fundamentally different from the competitive conditions on the retailer-to-consumer markets. (645)

996. The notifying parties explain (646) that the procurement of private label products and the procurement of branded products by retailers (and OOH wholesalers) might constitute two neighbouring, but separate markets, because the competitive conditions under which retailers source private label products and branded products would be fundamentally different. This may be the case in situations where specific branded products are perceived as "must carry" products by retailers.

997. Although the notifying parties have well known brands in the fresh flavoured dairy drink market this would not be sufficient for a separation at the upstream level according to the notifying parties as both Campina and Friesland Foods would be under constant pressure from private label products. According to the notifying parties this would be exemplified by increase of private label products over the years. Moreover, the notifying parties argue that the Netherlands would lag behind in the penetrating private labels since private labels account for 29.9% and 14.7% in Germany and Belgium respectively, a development which could also be expected for the Netherlands. (647)

998. More generally, the notifying parties put forward that a differentiation between a sourcing market for supplier brands and for private label is artificial in markets where the retail segment is so heavily concentrated as is the case in the Netherlands in particular. It would be unrealistic, according to the notifying parties, to assume that suppliers of branded products have strong bargaining power against giants like Albert Heijn (approximately 30% market share) and Superunie regrouping over 35% of retail demand.

999. Thus, private label and supplier brands would, according to the notifying parties belong to the same relevant product market at the sourcing level.

11.2.1.2.2. Assessment of the Commission

1000. In a recent consumer goods case (648) the upstream level - where retailers source their products - was distinguished from a downstream level, where the products are sold on to the final customer. In the present case, the same approach is followed for fresh flavoured dairy drinks.

1001. All fresh flavoured dairy drink producers sell their products to retailers or OOH wholesalers which, in turn, sell these products to consumers. Therefore, there are two stages in the supply chain: the upstream level, where the drinks are produced and supplied to retailers/OOH wholesalers and the downstream level with the supply to consumers. Campina and Friesland Foods are only active on the production and supply level.

1002. Fresh flavoured dairy drinks are available in two broad categories: brands owned by the dairy manufacturer and private label products of the retailer. (649) The two categories are sourced separately, but are displayed next to each other on the shelves. Both Campina and Friesland Foods currently produce only branded fresh flavoured dairy drinks.

1003. As explained in section 7.2.1.2, whether private labels and supplier brands belong to the same product market upstream depends on: (a) whether both types of brands, on the aggregate, compete closely with each other from the perspective of the end-customer and (b) the extent to which upstream suppliers of private label and the purchasing retailers, take into account the competitive pressure that private labels and supplier brands mutually exert on each other, in their negotiations upstream.

1004. With respect to the downstream interaction between private label and branded products, the majority of customers and competitors indicated in their responses to the market investigation that they compete to some extent at the retail level and that the quality of private label products is similar to branded products (650). Almost all retailers in the Netherlands have at least one private label in addition to the premium brands offered by Campina, Danone, Yakult and Friesland Foods and position them in relation to these brands, which have a price premium over private labels of around 15-20%. (651) Retailers are promoting their own retailer brands (652).

1005. A descriptive analysis of IRI data reveals that the limited quantities of private label sold in each of the four standard segments are priced only slightly below that of supplier brands (and sometimes above). Packaging is also similar with the large majority of sales in packages of 1 litre (for the non-health market) and 100cl or less for the functional or health market.

1006. However there are slight differences in total promotion activity, with private labels producers increasing their efforts in the last years to around 15% of volume on sale promoted as opposed to 10% for supplier brands. This difference is largely explained by the fact that private labels are subject to more price promotions than supplier brands.

Figure 11-5: Total promotion activities, private label and branded fresh flavoured dairy drinks, retail, Netherlands - Source: IRI.

1007. With respect to interaction at the upstream level it does not appear that upstream suppliers of private labels or retailers take the impact of private labels on sales of supplier brands of fresh flavoured dairy drinks seriously into consideration. This is mostly because private labels account only for a small size of the market and, despite greater promotional efforts the penetration of private labels in the market remains virtually unchanged since 2005. According to the notifying parties (653) the overall share - in value- of private label was 5.1% in 2007 compared to 5.4% in 2005. IRI data shows a small increase in 2007 relative to 2006 but the shares are still below 5% (these small discrepancies can be due to the exclusion of discounters in the IRI data set).

Table 11-2: Value sales private label and branded fresh flavoured dairy drinks, retail, Netherlands - Source: IRI.

1008. Other qualitative evidence obtained during the market investigation also suggests that at the upstream level retailers and suppliers do not take the competitive interaction between private label and supplier brands at the retail level into account.

1009. One aspect, as confirmed by customers and competitors (654), is that the procurement of branded products, as in the case of fresh milk, follows a different procedure than the one for private label products. Suppliers and retailers agree about the gross price, discounts, listing fees, and promotions in the case of branded products in bilateral negotiation. For private labels a tender procedure is usually undertaken, followed by a selection of preferred suppliers which agree on a net price in final negotiations. This in itself is not decisive as virtually the same suppliers and retailers participate in both segments. This is not the case however. Indeed, in contrast to fresh basic dairy products, the notifying parties (and other suppliers of brand products such as Danone) supply branded fresh flavoured dairy drinks almost exclusively and do not supply any private labels. Conversely, the few existing suppliers of private label do not sell supplier brands.

1010. This is confirmed by the notifying parties' submissions. According to notifying parties' Estimates (655), none of the branded suppliers such as Danone, Yakult, Campina or Friesland Foods is active in the supply of private label products in the Netherlands. Noem (Austria), Katshaar (Netherlands) and German competitors Gropper and TMA GmbH, supply only private label products in the Netherlands. One important reason for this limited presence in the branded segment for these producers is probably the relatively high entry cost into the branded segment of the market, with considerable investment and time needed to "build" a brand and raise consumers' awareness thereof.

1011. According to the notifying parties, the overall market is growing strongly and the growth is mainly driven by branded products. Based on the notifying parties submissions, during 2005- 2007, branded products like Danone increased by 135% in value, Yakult, 68.1%, or Campina's Optimel [140-150]*% (656). All the three brands invested into marketing and advertisement: Optimel [...]*% of sales, Yakult 19.3% and Danone 52.1% in 2007 (657). The IRI data confirms that the bulk of the growth in this segment was driven by increase in sales of supplier brands. Branded sales increased in absolute terms by EUR 83 988 000 whereas private label sales increased only EUR 3 851 000. Further while the slight increase in private label sales is driven by the overall growth of the weight management segment, in the other segments, in particular in the kind and regular segments, private labels are virtually absent.

Figure 11-6: Evolution of private label and branded fresh flavoured dairy drinks across segments, retail, Netherlands - Source: IRI.

1012. These elements taken together show that the fresh flavoured dairy drinks market is a brand dominated market. It follows that, in the upstream market, suppliers of branded products are unlikely to take into account, the competitive pressure downstream from private labels on supplier brands in their supplies to retailers, given that the former can be expected to exert only negligible competitive pressure on the latter.

1013. It is concluded that sourcing of private label and branded products should be viewed as separate product markets. Since the proposed merger would not lead to a significant impediment of effective competition on the market for private label fresh flavoured dairy drinks, as Friesland Foods is not active on that market and Campina has a market share well below [10-20]*%, this market is not discussed further.

11.2.1.3. The market for branded non-health related fresh flavoured dairy drinks has to be further separated according to distribution channels.

11.2.1.3.1. Delineation proposed by the notifying parties

1014. Friesland Foods and Campina both sell fresh flavoured dairy drinks to the retail and the foodservice channel (OOH). The OOH segment serves restaurants, cafes, hotels, catering services, hospitals but also small businesses like bakers or other food processing entities. According to the notifying parties, the OOH segment is typically served by foodservice wholesalers who deliver at the doorstep of the customer and by cash-and-carry businesses. In the Netherlands direct sales to OOH users would constitute only a very small proportion of the entire market segment (658).

1015. In previous decisions (659) a distinction was made between these sales channels because of differences in services, sales force, price structure, packaging sizes and health and safety regimes. The notifying parties case argue that such a distinction is not appropriate for several reasons: most of the products sold through the OOH segment are more or less identical to those in the retail segment; both channels partially supply the same customers who cross over depending on the circumstances; in part, players in the segments overlap; prices in both segments show similar developments and the absence of private label in OOH would not be determinative; the logistics of supply and additional services are largely the same.

11.2.1.3.2. Assessment of the Commission

1016. As explained in the Section 7.2.1, the market investigation reveals that differences between the retail and the OOH segment exist. These elements are: (i) different form of distribution, (ii) some differences in packaging size (iii) differences in the services demanded by OOH customers, (iv) logistics requested by OOH wholesaler, and (v) smaller order volumes in OOH compared to retail. Furthermore, different prices charged to retailers and OOH wholesalers also apply for the fresh flavoured dairy drink markets - in particular, according to the notifying parties, prices are [...]*% higher in the OOH segment compared to the retail segment (660). The notifying parties explain this by virtue of the smaller volumes purchased in the OOH segment and the smaller size of OOH wholesalers (661). These differences, however, are also an indication that competitive conditions in the OOH segment indeed differ from the general retail segment. Precisely because OOH wholesalers lack the size and the volume of retailers, they are not able to achieve similar prices. This also contradicts the claim made by the notifying parties that OOH wholesalers have purchasing power vis-à-vis the dairy manufacturer (662).

1017. On the basis of these elements, it can be concluded that the OOH segment forms a separate market for the supply of fresh flavoured dairy drinks.

11.2.1.4. Conclusion on relevant product market

1018. On the basis of the elements discussed in sections 11.2.1.1 to 11.2.1.3, it is concluded that there are separate relevant product markets for health related fresh flavoured dairy drinks and non-health related fresh flavoured dairy drinks, which can be further separated into the supply of branded and private label products and according to the distribution channel (retail/OOH).

1019. The proposed merger would not lead to an impediment of effective competition in the market for health related fresh flavoured dairy drinks and the private label market would not be affected. As a result, the competitive assessment focuses only on the branded, non-health related market for fresh flavoured dairy drinks.

11.2.2. Relevant Geographic Market

11.2.2.1. The relevant geographic market for non-health related fresh flavoured dairy drinks is national in scope.

11.2.2.1.1. Relevant geographic market proposed by the notifying parties

1020. The notifying parties submit (663) that the relevant geographic market for fresh flavoured dairy drinks at the downstream level is not wider than national.

1021. According to the notifying parties, the market for fresh flavoured dairy drinks at the upstream level is wider than the Netherlands and includes in any event Germany and Belgium due to several factors: (i) no significant differences in consumer taste between Member States, (ii) Community-wide and international brands (iii) supply chain, sufficiently long shelf life and transport cost considerations (iv) growing share of private label. All these elements would not allow suppliers in the Netherlands to increase prices profitably by 5-10% as they would lose significant volumes to foreign competitors.

11.2.2.1.2. Assessment of the Commission

1022. The market investigation in the present case has yielded strong indications that the relevant geographic market for non-health related fresh flavoured dairy drinks is narrower than submitted by the notifying parties.

1023. The conditions of competition in the different Member States - Belgium, Germany and the Netherlands - are not homogeneous. Brands vary depending on the Member State; Friesland Foods for example is not active outside the Netherlands (664), while Campina is present in Germany not with the brands used in the Dutch market like Mona or Yogho Yogho, but with German brands like Landliebe, Tuffi or Mark Brandenburg. In Belgium only the Campina brand is offered. Suppliers of brands active in Germany (Mueller, Zott, Gropper, Emmi or Nöm) or Belgium (Eurofit) cannot be found in the other two Member States according to the market share tables submitted by the notifying parties. In addition, the presence of private labels (at the retail level) differs significantly across the three Member States: private labels covered 42% of value sales in 2007 in Germany, 31% in Belgium and only around 5% in the Netherlands.

1024. Revealed preferences of consumers exhibit clear differences among the three Member States. In particular, German consumers have a strong preference for chocolate and other non-fruit flavoured drinks (above 50%), which is not the case for the Netherlands (5%). This observation is valid for the retail market as well as the OOH market. (665)

1025. In addition, respondents to the market investigation confirmed that the packaging format in the non-health related fresh flavoured dairy drink market in the Netherlands is unique and is hardly offered by suppliers outside the Netherlands. One competitor replied that "since the 1 l is unique for NL, it is not very easy to produce". (666) A customer noted in this respect that "branded, fresh flavoured, products sold in 1 litre gable top packaging satisfy a special demand which differs from the demand for the other products referred to in the present questionnaire." (667)

1026. Even if competitors - as indicated by the notifying parties in their reply to the Article 6(1)(c) Decision - were able to supply fresh flavoured dairy drinks in the 1 litre gable top format, they still would lack the brand recognition and brand awareness to be able to enter the Dutch market, which is dominated by brands.

1027. This is further supported by the fact that private label products - contrary to the submission of the notifying parties in their reply to the Article 6.(1)(c) Decision - have been rather constant and stable (not growing) in the Dutch retail segment. In the OOH segment, private labels are not present at all.

1028. In view of these elements, it is concluded that the relevant geographic market for the upstream market of non-health related fresh flavoured dairy drinks is national.

11.2.3. Competitive Assessment

1029. The market investigation received several strong complaints, in particular from customers who consider that the merger will ultimately lead to higher prices because the two leading suppliers of brands would merge and no alternative supplier would be available. Given the strong position of the notifying parties they would have an incentive to increase prices for branded non-health related fresh flavoured dairy drinks.

11.2.3.1. The notifying parties have high market shares

1030. According to the notifying parties the downstream market for non-health related fresh flavoured dairy drinks covering the Netherlands has a total value of EUR 51 201 000 in 2007 (EUR 44 205 000 retail segment, 6 997 000 OOH), 4.8% of it coming from private label products and the remaining 95.2% by branded fresh flavoured dairy drinks (668).

1031. According to their submission Campina's brand portfolio achieved a market share of [50- 60]*% (Yogho Yogho [20-30]*%, Mona [20-30]*%), followed by Friesland Foods' brands with [30-40]*% (Fristi [0-5]*%, Milk&Fruit [20-30]*% and GO! [0-5]*%). Danone has a market share of [5-10]*% and Dr.Oetker [0-5]*%.

1032. In the OOH markets the notifying parties had a combined market share of [90-100]*% in 2007 (Campina [20-30]*%, Friesland Foods [70-80]*%).

1033. Looking at the upstream market, where retailers/OOH wholesalers source non-health related fresh flavoured dairy drinks from dairy producers, the situation is as follows according to the notifying parties: In the branded retail market the notifying parties achieve a combined market share of [80-90]*% - increased by [0-5]* percentage points since 2005 - with Campina at [50- 60]*% and Friesland Foods at [20-30]*%.

Table 11-3: Sourcing of branded fresh flavoured drinks, indulgence, retail, The Netherlands - Source: Form CO.

1034. Since there is no private label available in the OOH segment, the market shares at the upstream level are similar to those for the downstream one.

1035. However all the figures submitted by the notifying parties correspond in fact to a different definition of the market as to that adopted in this Decision. In particular, the notifying parties are assuming that Campina's Optimal fresh flavoured dairy drink would belong to the health related product market.

1036. On the basis of the IRI data, which excludes discounters - although inferring from the notifying parties submissions discounters do not have any significant presence in this market - the market for branded non-health fresh flavoured dairy drinks has evolved as shown in the Table 11-4:

Table 11-4: Value Sales per segment, non-health fresh flavoured dairy drinks, retail, The Netherlands - Source: IRI.

1037. The market shares of the various suppliers brands are as follows:

Table 11-5: Market shares supplier brands, non-health related fresh flavoured dairy drinks, retail, The Netherlands - Source: IRI.

1038. Since the private label sales are excluded, the market shares at the upstream level are identical to those for the downstream one.

1039. The larger the market share, the more likely a firm is to possess market power. Moreover, the larger the addition of market share, the more likely it is that the merger will lead to a significant increase in market power. The proposed transaction would combine the current market leader, Campina, with the second strongest player in the market, Friesland Foods,

1040. As shown in the Table 11-5 Campina sells three brands: Mona, Optimel and Campina, generally targeting the kind, wealth management and regular, respectively. The combined market share in 2007 was [80-90]*%. Friesland Foods has lost a part of its market share over the years but is still the second largest and virtually the only competitor to Campina with a [10-20]*% market share. The merged entity would enjoy a quasi-monopoly.

1041. For the OOH market the strong market leader Friesland Foods ([70-80]*%) would join forces with the number two in the market achieving [90-100]*% - overlap [20-30]*%.

1042. The increase in market share combined with the fact that only negligible competition would exist from Danone, which has also lost part of its small market share over the last four years indicates that the merger would, as a result of non-coordinated effects, lead to a significant impediment of effective competition in the market for branded non-health fresh flavoured dairy drinks.

11.2.3.2. The notifying parties are the closest competitors

1043. Another important aspect is the degree of substitutability between the notifying parties' products. The higher the degree of substitutability, the more likely it is that the notifying parties will raise prices significantly. In particular, almost all customers consider the notifying parties as the closest competitors for fresh flavoured dairy drinks in gable top packaging, and retailers/OOH wholesalers regarded them as their first and second choice, and their rivalry as an important source of competition on the market.

1044. For example, one customer stated that "for fresh flavoured dairy products sold in 1 litre gable top packs, the merger will de facto lead to the elimination of all competition without the possibility [...] to switch to other suppliers" and added that "for all gable top packaging Friesland is closest competitor of Campina." (669) Similar CUR-FD-I-5 "the only competition that could exist, is the competition on the cartons. But, the market share of Danone in this segment [...] is very small."

1045. Campina has positioned its brands across the three segments. The Mona brand is strong in the kind segment and the Optimel brand in the weight management segment. The Campina brand is strongest in the "regular" segment. It appears from Figure 11-7 that Friesland Foods has positioned its brand both in the regular and the weight management segment whereas it is largely absent in the kind segment. This is the segment where Danone is most active (but it is worth noting that the kind segment is approximately one fourth and one fifth of the size of the regular and the weight management segments, respectively).

Figure 11-7: Allocation of brands in different segments based on volume. Source: IRI

1046. In terms of prices the notifying parties brands are also aligned in the two segments where Friesland Foods has a notable presence. In contrast, Danone's prices seem far apart as shown in Figure 11-8. [...]*

Figure 11-8: Mean average volume prices important brands, retail, The Netherlands - Source: IRI.

1047. The Commission has attached a set of econometric results on estimation of retail demand systems to its Statement of Objections. These results complemented the other, larger set of qualitative evidence and descriptive statistics. The goal of the econometric models was to assess whether supplier brands exert a greater competitive constraint than private labels on each other. In the case of fresh flavoured products, the Commission's finding from the econometric evidence was that the notifying parties' brands mutually exert a significant competitive constraint on each other and, furthermore, private labels do not significantly constrain the Friesland Foods or Campina brands. The notifying parties have raised a number of criticisms regarding the interpretation and the robustness of the econometric results in their response to the Statement of Objections. (670)

1048. Most of the criticisms put forward by the notifying parties are invalid or based on a misunderstanding of the methodology. Moreover, a number of criticisms regarding the robustness of the results can be addressed by extending the econometric model. Such an extension is also justified on economic grounds albeit for reasons other than those raised by the notifying parties. In the case of fresh flavoured products, the results of the extended model support those results put forward in the SO, that is, that the brands of the notifying parties are not significantly constrained by private label brands relative to the constraint they exert on each other. (671)

1049. On the basis of the analysis presented in section 11.2.3.2, it can be concluded that the merger would remove from the market a close competitor of Campina. This could lead to significant price increases or a reduction in the efforts to compete in the introduction of new brand features.

11.2.3.3. Customers have limited possibilities of switching suppliers

1050. In their response to the market investigation customers have indicated that besides viewing the notifying parties as close competitors, they would have difficulties in switching to other suppliers. This is valid for the retail segment as well as the OOH segment where retailers/OOH wholesalers usually source from both notifying parties at the same time.

1051. The market investigation revealed that currently only one supplier, the Dutch company Katshaar Zuivel, would be able to supply fresh flavoured dairy drinks in gable tops. However, Katshaar only supplies private label products and is capacity constrained (according to customers) and thus not able to supply sufficient volumes to offset a price increase postmerger (672).

1052. When asked to comment on past examples of suppliers switching from either Campina or Friesland Foods to other suppliers of fresh flavoured dairy drinks, the notifying parties provided no single example for branded products and referred only to a change in private label supply from the notifying parties to Katshaar (673).

1053. Notwithstanding the lack of evidence concerning other viable suppliers of branded products, in their reply to the Article 6.(1)(c) Decision the notifying parties argued (674) that foreign suppliers like Danone as well as private labels could in fact emerge as an alternative to the notifying parties. The market investigation suggests, however, that this is highly unlikely.

1054. First, Danone is hardly active in the non-health related market and is strong in the healthrelated market, in particular with its portion packs. These products belong to a separate market, consequently Danone and for the same reason Yakult cannot be considered to form a competitive constraint to the notifying parties since customers would not switch between these product types (portion packs/gable tops).

1055. Second, no foreign suppliers are perceived by customers as a credible alternative since it would be necessary for such supplier to create and develop brands. One customer replied that because "of consumer demand, we cannot change the current branded products to foreign products" (675). Another one explained it was unable to switch suppliers in the case of a 5-10% price increase as all A-brands would come from the Netherlands (676). Moreover, foreign suppliers indicated that they would either not produce fresh flavoured dairy drinks in gable tops or would have no capacity left to supply additional volumes (677).

1056. Finally, private label supply from abroad was not considered as a competitive constraint to branded suppliers as claimed by the notifying parties. One retailer (678) noted that "private label production of fresh flavoured dairy drinks does not offer an alternative to the branded products. These drinks [...] have not affected the sales of the branded drinks sold by the merging parties. These brands are very strong and have not suffered from the arrival of private label".

1057. These considerations apply mutatis mutandis to the OOH market, where customers currently rely exclusively on branded products supplied by the notifying parties, with no alternative producer available at all.

11.2.3.4. Lack of countervailing buyer power

1058. The notifying parties do not in the Form CO nor in their reply to the Article 6.(1)(c) Decision explicitly refer to the existence of countervailing buyer power from large retailers and OOH wholesalers that would restrain them from raising prices when it comes to fresh flavoured dairy drinks. However, since Campina and Friesland Foods have made this claim in general (679), this is assessed in what follows.

1059. According to the Commissions' horizontal merger guidelines, countervailing buyer power should be understood as the bargaining strength that the buyer has vis-à-vis the seller in commercial negotiations as a result of its size, commercial significance to the seller and its ability to switch to alternative suppliers.

1060. In a market structure where the notifying parties have a market share of [80-90]*% (retail) and [90-100]*% (OOH), are perceived as the closest competitors and customers would have only limited switching possibilities post-merger, the countervailing buyer power of retailers and OOH wholesalers would be very limited.

1061. Furthermore, it is not sufficient that buyer power exists prior to the merger, it must also exist and remain effective following the merger. This is because a merger of two suppliers may reduce buyer power if it thereby removes a credible alternative. In the present case, as described in section 11.2.3.2, the merger will remove one of the two main suppliers considered to be the closest competitors in the market, since alternative suppliers with a similar size could not be identified in the course of the market investigation.

1062. Therefore, it is concluded that countervailing buyer power post-merger would not be sufficient to off-set potential adverse effects of the merger.

11.2.3.5. Entry unlikely to occur

1063. Despite the high market shares, the notifying parties argue that in case of a price increase other competitors would enter the market constraining the notifying parties.

1064. The market investigation did not point to the existence of a single potential competitor with the ability of incentives to enter the Dutch market for branded non-health fresh flavoured dairy drinks. For example, one competitor clearly explained that it would supply its branded products only in its home market. (680)

1065. Retailers also argued (681) that transport costs would be higher if such a product were sourced from abroad. The longer supply distance would increase costs for the retailer (and their customer) and place foreign suppliers at a disadvantage in relation to freshness because Dutch suppliers would always be able to supply fresher products as they are located closer to the distribution centres of the retailers. Since Dutch customers consider freshness as a key variable in their purchasing decision, Dutch incumbents have a clear competitive advantage.

11.2.3.6. Conclusion on competitive assessment

1066. For the reasons set out in sections 11.2.3.1 to 11.2.3.5, it is concluded that the notified concentration is likely to significantly impede effective competition on the market for branded non-health fresh flavoured dairy drinks in the Netherlands, which is a substantial part of the common market, separated according to the distribution channel in retail and OOH.

11.3. Long-life flavoured dairy drinks

1067. Long-life flavoured dairy drinks (LLDDs) are mixtures of (fermented or non-fermented) milk and flavours, notably aromas, fruit juices and ccocoa. Only a certain amount of the aromas/fruit juices or cocoa is added so that the product does not fully lose its dairy characteristics. LLDDs are available in different formats, including 1 litre glass or plastic bottle, 1 litre tetrapak or portion-packed smaller formats (for example 0.2 litre bottle).

1068. The production process for LLDDs is similar to the production process used to produce longlife milk apart from the ingredients which are added during the process. As for long-life milk, the milk can be sterilised or subject to Ultra-High Temperature treatment. Ingredients (fruit juices, aromas, sugar) are then added before the product is poured into glass, polyethylene bottles or brick packs. LLDDs can be kept at room temperature for months.

1069. Campina and Friesland Foods sell their products under different brands and (for Campina) private labels in Belgium, Germany and the Netherlands. Campina sells fruit-flavoured dairy drinks under the Yogho Yogho brand in Belgium and the Netherlands, the Fruttis Yogho Yogho brand in Germany, the Yazoo brand in Belgium and the Netherlands, the BonOmel brand in the Netherlands and under private label in Belgium and the Netherlands. Campina sells chocolateflavoured dairy drinks under the Yazoo brand in Belgium and the Netherlands, the BonOmel brand in the Netherlands, the Stabilac brand in Belgium, the Campina Choco Choco brand in Belgium and the Netherlands, the Südmilch and Gastro brands in Germany and under private label in Belgium and the Netherlands.

1070. Friesland Foods sells fruit-flavoured dairy drinks under the Fristi brand in Belgium and the Netherlands. It sells chocolate-flavoured dairy drinks under the Chocomel brand in Germany and the Netherlands and the Cécémel brand in Belgium. In Germany, Friesland Foods sell also small volumes of long-life flavoured dairy drinks under the brand Domo.

11.3.1. Relevant Product Market

11.3.1.1. Chocolate-flavoured LLDDs and fruit-flavoured LLDDs belong to different product markets

11.3.1.1.1. Product market definition proposed by the notifying parties.

1071. As mentioned in recital 1067, dairy drinks are chocolate-flavoured or fruit-flavoured (mainly strawberry or raspberry-flavoured but also peach-flavoured, banana-flavoured or apple/pearflavoured). There are also some LLDDs which are neither chocolate nor fruit flavoured (for example vanilla-flavoured, coffee-flavoured dairy drinks). The notifying parties' activities do not overlap in these segments.

1072. The notifying parties submit that chocolate-flavoured and fruit-flavoured dairy drinks belong to separate product markets. They do not elaborate whether LLDDs which are neither chocolate nor fruit-flavoured belong to the same market as chocolate-flavoured, or to the same market as fruit-flavoured, or to a distinct market.

1073. Apart from the taste, which is obviously different, the notifying parties submit that fruitflavoured LLDDs are mainly aimed at, and consumed by, children. For example Fristi (Friesland Foods) is mostly consumed by 4-9 year old children during breakfast, lunch or in between meals as snacks. Yogho Yogho (Campina) is consumed by children that are somewhat older (as of 8-9 years). (682)

1074. Based on a marketing study conducted by Friesland Foods in 2002 (683), the notifying parties submit that consumers perceive chocolate-flavoured dairy drinks as drinks that emphasise the creamy and indulgent side of dairy. The taste and texture of chocolate-flavoured dairy drinks are perceived as very distinctive and therefore almost irreplaceable. According to the notifying parties, fruit-flavoured dairy drinks have more a typical dairy orientation (which implies that they are more health-oriented) and are more in direct competition with fruit juices or still drinks. The notifying parties submit that still drinks have shown tremendous growth over the last ten years in the Netherlands, to the detriment of fruit-flavoured long-life dairy drinks.

1075. In their response to the 6(1)(c) Decision, the notifying parties repeat their claim that chocolate-flavoured and fruit-flavoured are not substitutable in view of the fact that there is significant demand for chocolate drinks from adults whereas fruit-flavoured dairy drinks are hardly consumed by adults at all (684). This claim was reiterated in the reply to the Statement of Objections.

11.3.1.1.2. Assessment of the Commission

1076. With regard to a possible distinction between chocolate flavoured LLDDs and fruit-flavoured LLDDs, the market investigation showed mixed results. While some respondents submit that these products are not substitutable due to strong differences in taste and mouth-feel, others underline that both product categories are targeting the same consumer groups (children) and that the choice of taste is rather dynamic for the consumers: in other words, if the price of for example fruit-flavoured dairy drinks would increase by 5-10%, it could not be excluded, according to these respondents, that customers would switch to chocolate-flavoured dairy drinks (685).

1077. The market investigation showed that there are some elements which support the argument that chocolate-flavoured and fruit-flavoured LLDDs are in direct competition with each other. First, the products are, in the retail segment, presented on shelves or in supermarket space in an ambient area in direct proximity to one another and displayed as one "family" of products. Endconsumers can therefore easily choose as an impulse purchase between chocolate-flavoured LLDDS and fruit-flavoured LLDDs. They are also usually subject to only one category manager or category management team.

1078. The notifying parties also provided elements with regard to the development of drinks consumption in the Netherlands (686) which demonstrate in their view that fruit-flavoured dairy drinks are more in competition with other non-dairy drinks, in particular still drinks, than chocolate-flavoured drinks. However this data showed that if still drinks and fruit nectars have grown sharply in the few last years, the categories of "liquid cultured milk" (which would include fruit-flavoured LLDDs such as Fristi according to the notifying parties) and "flavoured milk" (which would include chocolate-flavoured LLDDs such as Chocomel according to the notifying parties) have also increased, albeit more moderately. Decreasing categories are carbonates, hot drinks, juices and alcoholic beverages, especially beer. It does not appear from this data, as submitted by the notifying parties, that still drinks have mainly gained share on fruit-flavoured LLDDS compared with chocolate LLDDS; both categories of LLDDs have followed roughly the same trend.

1079. On the other hand, other elements collected during the market investigation point towards a distinction between fruit-flavoured dairy drinks and chocolate-flavoured dairy drinks.

1080. First, the taste is clearly different. As mentioned in recital 1177, several respondents in the market investigation indicated that fruit-flavoured dairy drinks and chocolate-flavoured dairy drinks are not substitutable from the point of view of the end-consumer due to strong differences in taste and mouth-feel (687)

1081. Secondly, with regard to the consumption patterns, it is true that among children aged 1-11 years, the group of fruit-flavoured LLDDs users (23%) is larger than the group of children that consumes chocolate-flavoured LLDDs (11%). In general the main part of volume of fruitflavoured LLDDS is consumed by children. With respect to all other age groups, differences in consumption patterns between fruit-flavoured and chocolate-flavoured drinks are less pronounced. It follow that fruit-flavoured LLDDs are preferred by young children whereas older children and adults consume fruit -flavoured and chocolate-flavoured alternatively. The fact that fruit-flavoured dairy drinks target children is demonstrated in Figure 11-11. Fristi and Yogho Yogho (fruit brands) are aimed at children whereas Chocomel (chocolate brand) is not aimed at a specific target group.

1082. The notifying parties originally explained (688) that fruit-flavoured and chocolate-flavoured LLDDs follow the same consumption pattern as they are both consumed at breakfast, lunch and in-between meals as snacks. In their reply to the Statement of Objections, the notifying parties slightly revised their position and submitted a graph showing that total volumes fruit-flavoured LLDDs are mostly consumed at breakfast and lunch (689) (33% of the volume is consumed at breakfast and 30% is consumed at lunch) while chocolate-flavoured LLDDs are mainly consumed at lunch (27%) and during in-between meals as snacks in the afternoon (22%). Only 17% of the volume of chocolate-flavoured dairy drinks is consumed at breakfast. These elements show that both categories of long-life dairy drinks follow different consumption patterns.

1083. In light of the mixed evidence concerning the question of whether chocolate and fruit flavoured LLDDs belong to the same market, the elasticity matrix for a number of aggregate categories of flavoured dairy drinks was estimated in the Statement of Objections. An AIDS model of demand was specified to depict substitution patterns across different segments of flavoured dairy products. The result of this analysis with respect to chocolate-flavoured and fruitflavoured dairy drinks showed that these two long-life segments were good candidates to be aggregated together in a single product market.

1084. The notifying parties have raised a number of criticisms regarding the interpretation and the robustness of the econometric results in their response to the Statement of Objections (690). On the one hand, some of the criticisms put forward by the notifying parties are invalid or based on a misunderstanding of the methodology. On the other hand, a number of criticisms regarding the appropriateness of the model specifications related to the long life dairy drinks segment have merit. On balance, the Commission decided not to attach weight to the econometric evidence. (691)

1085. In conclusion, although the results of the market investigation are mixed and the qualitative evidence available goes in both directions, it is concluded, following the arguments of the notifying parties, that fruit-flavoured and chocolate-flavoured dairy drinks are separate product markets.

11.3.1.2. It can be left open whether private label and supplier brands of long-life flavoured dairy drinks belong to separate product markets

11.3.1.2.1. Delineation proposed by the notifying parties

1086. The notifying parties agree that a vertical segmentation of a particular product market for consumer products - in particular foodstuffs - can be made between an upstream market for the sourcing of products by retailers (and OOH wholesalers serving hotels, restaurant, catering services and filling stations) and a downstream market for the sale of foodstuffs by retailers to consumers for two reasons.

1087. First, this approach recognises both that on the retail-to-consumer level private label products compete with (producer) branded products, and that the market share of private label products should be attributed to the retailers that own the private label brands.

1088. Secondly, the approach recognises that the competitive conditions on the markets on which retailers and OOH wholesalers source their products are fundamentally different from the competitive conditions on the retailer-to-consumer markets. (692)

1089. As explained earlier the notifying parties agree that the procurement of private label products and the procurement of branded products by retailers (and OOH wholesalers) might constitute two neighbouring, but separate markets, because the competitive conditions under which retailers source private label products and branded products would be fundamentally different. This may be the case in situations where specific branded products are perceived as "must carry" products by retailers, in the sense that significant sales might be lost in the event of delisting such branded products.

1090. In the context of LLDDs, the notifying parties acknowledge that Chocomel/Cécémel and Fristi are perceived as "must-carry" brands (693) in the retail and the OOH segment. However, the notifying parties submit that this in itself does not justify identifying distinct sourcing markets. According to the notifying parties, private label products are flourishing in LLDDs and occupy such a strong position that retailers are in a position to threaten branded products manufacturers with temporary delistings. According to the notifying parties, the private label volume-based market share was 45% in the Netherlands whereas in Belgium this was 51% (figures for the retail segment only) in 2007. On the basis of IRI data the share of private label is however only between 20-25% for the Netherlands. Presumably this large discrepancy could be explained by the fact that IRI data excludes deep discounters (Aldi, Lidl).

Table 11-6 : Private label penetration in long-life dairy drinks (20004-2008, Source IRI)

1091. More generally, the notifying parties hold that a differentiation between an upstream market for branded and for PL products is artificial in markets where the retail segment is so heavily concentrated as in the Netherlands. It would be unrealistic, according to the notifying parties, to assume that suppliers of branded products have strong bargaining power against giants like Albert Heijn (approximately 30% market share) and Superunie regrouping over 35% of retail demand. According to the notifying parties, customers have alternative sources for branded or PL products depending on supply conditions.

1092. The logical consequence of having no separate markets at the upstream level as put forward in the notification would be to aggregate sales by the notifying parties at that level for private label and branded products. However, as earlier mentioned in their reply to the 6.(1)(c) Decision (694) the notifying parties criticised the Commission for aggregating the notifying parties' sales at the sourcing level for private label and branded products, because for long-life dairy drinks "the market shares are based on the sum of PL and branded sales which gives an incorrect picture of market" and "the Commission does not sufficiently distinguish branded from PL sales". Instead the notifying parties proposed to follow the analytical framework put forward in the Lactalis-Nestle/JV decision (695).

1093. In their reply to the Statement of Objections, the notifying parties reiterated their claim that branded and PL products belong to the same market at upstream level (696). In particular, they argued that the competition between branded and PL products is underscored by the resemblance of the packaging between branded and PL products, which has given rise to legal proceedings from Friesland Foods in a number of instances. They also referred to the strong promotional campaign carried out by retailers, particularly Albert Heijn, for PL products. Finally, the notifying parties stressed the importance of PL to retailers because of higher margins that can be obtained by selling PL instead of branded products.

1094. In their reply to the Statement of Objections, the notifying parties also provided a study (socalled "duplication analysis") conducted by GFK (697) and based on household panel data for both Belgium and the Netherlands. As explained by the notifying parties, this analysis essentially tracks which households bought which range of LLDD brands. According to the notifying parties, this analysis shows that PL sold by Aldi and Lidl and PL sold by other retailers have a high percentage of "exclusive buyers" (698) which prove "the strong competition between branded and PL products downstream" (emphasis added) (699).

1095. As a preliminary point, these elements put forward by the notifying parties, such as the similarity of packaging of branded and PL products and the percentage of exclusive buyers of some PL products refer to downstream competition between branded and PL products at the level of the endconsumer, and not the upstream market where retailers source their products and where the merger is likely to have an impact. As will be explained in section 11.3.1.2.2 and on the basis of the market investigation, branded and PL LLDDs compete downstream. However, this does not mean that they would automatically belong to the same market upstream at the level where retailers and OOH wholesalers source their products. Competition upstream is linked to a different set of factors, which will be described in the following paragraphs.

11.3.1.2.2. Assessment of the Commission

1096. As explained in section 7 and 11-2 in relation to fresh basic dairy products and freshflavoured dairy drinks, all LLDDs producers sell their products to retailers or OOH wholesalers which, in turn, sell these products to consumers. Therefore, there are two stages in the supply chain: the upstream level, where LLDDs are produced and supplied to retailers/OOH wholesalers and the downstream level with the supply to consumers. Campina and Friesland Foods are only active on the production and supply level.

1097. Likewise, LLDDs are available in two broad categories: brands owned by the dairy manufacturer and private label products of the retailer. The two categories are sourced separately, but are displayed next to each other on the shelves. Campina produces both branded and private label LLDDs whereas Friesland Foods produces only branded LLDDs (700).

1098. In line with the approach taken in other chapters, the issue whether private labels and supplier brands belong to the same product market upstream also depends on the following factors. It must in particular be assessed: (i) whether both types of brands, in general, compete closely with each other from the perspective of the end-customer and (ii) the extent to which upstream suppliers of private label and/or brands as well as the purchasing retailers take into account in their negotiations upstream the competitive pressure that private labels and supplier brands mutually exert on each other at consumer level

1099. With respect to the first element, the majority of customers and competitors indicated during the market investigation that private label and branded products compete at the retail level and that the quality of private label products is similar to that of branded products. (701) Almost all retailers in the Netherlands and Belgium have private labels in addition to the brands offered by Campina and Friesland Foods and position them in relation to these brands, which have a price premium over private label products, as shown in Figure 11-9.

Figure 11-9 : Prices of branded and private label products, Source IRI

1100. Looking at the upstream level, the available evidence available points towards a distinction between sourcing of private label and branded products.

1101. First, as for other dairy products (and in general consumer goods), the procurement of branded products is different than the one for private label products. Suppliers and retailers agree about the gross price, discounts, listing fees, and promotions in the case of branded products in bilateral negotiations. For private label products a tender procedure is usually undertaken, followed by a selection of preferred suppliers which agree with the retailer on a net price in final negotiations. This point has not been contested by the notifying parties.

1102. This in itself would not be decisive if the same suppliers are active in both segments. In a situation where the bulk of both the private label and the branded products would be supplied by the same firms, it could indeed be presumed that these firms would take into account in supplying customers the substitutability of private label and supplier brands among end-consumers. However, this is not the case for LLDDs. The market presence of competitors and even that of the notifying parties vary considerably whether it comes to branded or PL products. According to notifying parties' estimates, German competitors Immergut, Hochwald and Sachsenmilch only supply PL products in the Netherlands. Belgian company Inex sells 96% of its volume of longlife dairy drinks under PL in the Netherlands and only 4% under brands.

1103. In Belgium, German dairy undertakings Milch Union Hocheifel (MUH) and Hochwald, as well as the Belgian company Eurofit are exclusively active in the PL segment, according to notifying parties' estimates confirmed by the market investigation. Belgian competitor Inza achieves more than 80% of its turnover in long-life dairy drinks with PL products. Only Campina ([...]* branded products and [...]* PL) and to a lesser extent Inex (but in Belgium only) have a more balanced portfolio between private label and branded products. One important reason for this limited presence in the branded segment for these producers is probably the relatively high entry cost into the branded segment of the market, with considerable investment and time needed to "build" a brand and raise consumers' awareness thereof.

1104. The notifying parties have contested these elements in the reply to the Statement of Objections arguing that in the first phase investigation the majority of competitors (3 out of 5) replied that they supply branded and PL products to the same customers (702). Two of these three answers emanate from German companies whose sales data shows that they supply only PL products in Belgium and the Netherlands.

1105. On the other hand, as mentioned in recital 1070, Friesland Foods is basically not active in the private label segment. This is not surprising since Friesland Foods can rely on strong brands (Chocomel/Cécémel and Fristi) which are considered by the notifying parties themselves as "must-carry" brands. This is also the case for Belgian supplier Vitalac which is only active in branded products with its brand Chocovit.

1106. It follows that, as in the case of fresh flavoured dairy drinks, and in contrast with fresh basic dairy products most of the suppliers of private label products do not supply branded products, with the significant exception of Campina. In a situation where the private label offering is supplied by a different set of producers compared to the branded market, conditions of competition become "asymmetric"- the branded producers could constrain the private label producers but the opposite is not possible as a private label producer would have to invest into a brand and marketing before being able to compete against a strong brand portfolio such as Friesland Foods'. Therefore the branded producer such as Friesland Foods which carry must-have brands would hardly take the substitutability of private label and supplier brands among end customers into account in supplying retailers and OOH wholesalers, since this substitutability will affect the sales and the relative margins it can expect to achieve in its negotiations vis-à-vis its customers to only a limited extent.

1107. With regard to profit margins, it follows from the data submitted by Campina (703) that margins for products marketed under the brands Campina, Yogho Yogho and Yazoo in Belgium were consistently above [...]*% in the period between 2005 and 2005 whereas margins for PL products have remained below [...]*% over the same period. In contrast to the fresh basic dairy markets, this data does not show that the difference in margins has significantly declined over the last two years.

1108. In the Netherlands, Campina's margins are also higher for branded products than for private label products: on average for 2007, the difference is [...]*% for chocolate-flavoured dairy drinks and slightly above [...]*% for fruit-flavoured dairy drinks (overall [...]*%). Although the difference in margins has steadily declined over the last three years in chocolate-flavoured dairy drinks, it remained stable for fruit-flavoured dairy drinks. For all long-life dairy drinks, the difference in margins dropped between 2005 and 2006 but remained stable between 2006 and 2007, showing that the competitive pressure stemming from private label products producers to branded products' suppliers is still limited.

1109. In their reply to the Statement of Objections, the notifying parties did not contest the Commission's findings with respect to stability of difference in margins between branded and PL products but emphasized that it was not "made clear" (704) why this stability indicates a lack of competitive pressure from PL products.

1110. It is emphasised that this stability of margins does not necessarily prove that there is a lack of competitive pressure from PL products at the downstream level. However, at the upstream level, this stability of margins is a significant element. For a company which produces both branded and PL (Campina), if differences in margins remain the same, this producer would normally prefer to use the capacity for branded production. It will only produce PL to the extent necessary to utilize spare production capacity and achieve economies of scale. Companies which produce branded and PL products will therefore remain focused on brands where higher margins are expected. In that respect, it is worth noting that Campina's turnover in private label LLDDs has dropped by [...]*% between 2005 and 2007 whereas its branded activities have experienced a lower decrease ([...]*% between 2005 and 2007).

1111. Finally, no decisive indications were found that private label products are growing in the LLDD market. According to the figures provided by the notifying parties, private label products represented in 2007 39.2% in volume and 30.6% in value on the downstream markets in the Netherlands. In 2005, they represented 39.7% and 31.6% respectively (705). In Belgium, growth has been more significant (from 37.2% to 41.8% in volume and from 16.8% to 22.4% in value) (706) but still almost 80% of all LLDDS sold currently in Belgium are branded.

1112. With regard more specifically to chocolate-flavoured LLDDs, the notifying parties themselves concede that the penetration rate of PL products in Belgium has not significantly changed in the last ten years - it represented 45.3% (volume) and 29.9% (value) in 1997 and 47.2% (volume) and 29% (value) in 2007 (707).

1113. The notifying parties did not explicitly contest these figures, which illustrate no significant growth, but indicate that on the basis of the GFK data, PL would account for the "lion's share" (708) of the market (53% in the Netherlands and 60% in Belgium). It is understood that these figures are in volume for the retail segment. However they are very different from those which have been submitted in the notification (Netherlands 45% in volume in the retail segment, Belgium 51%). Since the figures mentioned in the Statement of Objections originate from the notifying parties in the notification and are based on IRI and GFK data for the Netherlands and IRI and Nielsen data in Belgium (709), it is unclear how the notifying parties found these different results.

1114. The elements taken together indicate that at the sourcing level, competitive conditions differ sufficiently between private label and branded products with respect to long life dairy drinks. In any event, it is not necessary to reach a definitive position on this issue given that even if the Commission defines a product market including branded and PL products at sourcing level, the transaction leads to a significant impediment of effective competition with respect to long-life dairy drinks.

11.3.1.3. Whether the markets for LLDDS have to be further separated according to distribution channel can be left open

1115. Friesland Foods and Campina both sell fresh basic dairy products to the retail and the foodservice channel (OOH). In the Friesland Coberco-Nutricia decision, a distinction was made between these two distribution channels for LLDDs. However, in the later Arla Foods-Express Dairies decision, no reference to the distinction between the various distribution channels was made.

1116. The notifying parties did not clearly specify in the notification whether in their view they believe that the retail segment and the OOH segment belong to different product markets for LLDDs. They nevertheless claimed that there is a market difference between both segments since the 0.2 litre glass bottle packaging LLDD is exclusively used in the OOH segment as an alternative for soft drinks and fruit juices. Further, some products sold under the Joy brand by Heineken's subsidiary Vrumona are only available in the OOH segment. More generally drink wholesalers would be active in the OOH segment for LLDDs and therefore competitive conditions would be different from the retail segment.

1117. As explained in Sections 7 and 11-2 on fresh basic dairy products or fresh-flavoured dairy drinks, the market investigation reveals that differences between the retail and the OOH segment exist. These elements are: (i) different forms of distribution, (ii) differences in packaging size (iii) differences in the services demanded by OOH customers, (iv) logistics requested by OOH wholesaler, and (v) smaller order volumes in OOH compared to retail. For LLDDs, the market investigation also broadly confirmed that there are differences in market players and packaging used in the retail and the OOH segment although some respondents indicated that the 0.2 litre glass bottle packaging should disappear in the near future (710).

1118. There are therefore indications that the markets for LLDDs could be separated according to OOH and retail distribution channels. In any event, given the strong presence of the notifying parties on a combined market of the supply of retail and OOH, the question whether the OOH segment should be considered as a separate market for the supply of LLDDs can be left open as even on the broader market the proposed transaction would significantly impede effective competition.

11.3.1.4. Conclusion on relevant product market

1119. On the basis of the elements discussed in Sections 11.3.1.1. to 11.3.1.3, it is considered that chocolate-flavoured dairy drinks and fruit-flavoured dairy drinks belong to distinct product markets. Further there is no need to conclude whether sourcing of branded and private label longlife dairy drinks belong to different product markets. A distinction according to the distribution channel between retail and OOH can also be left open.

11.3.2. Relevant Geographic Market

11.3.2.1. Delineation proposed by the notifying parties

1120. The notifying parties explain (711) that the relevant geographic market for long-life dairy drinks at the downstream level is not wider than national.

1121. According to the notifying parties, the market for long-life flavoured dairy drinks would be wider than the Netherlands at the upstream level and would in any event include Germany and Belgium due to several factors: (i) the "strong penetration" of the Dutch market by foreign competitors in particular from Belgium and Germany, (ii) the absence of barriers to cross-border trade (the notifying parties mention in that regard that some brands such as Fristi or Chocomel/Cécémel are used internationally) (iii) the long-life characteristics of the products which allow sourcing over long distances.

1122. In their reply to the Statement of Objections, the notifying parties noted that the factors that are put forward in the Statement of Objections rather indicate that the sourcing market is wider than national (712).

11.3.2.2. Assessment of the Commission

1123. In a previous decision (713) concerning the United Kingdom indications were found that the relevant geographic market for long-life flavoured milk might be broader than national.

1124. The market investigation in the present case yielded indications that the relevant geographic markets for branded chocolate-flavoured dairy drinks and fruit-flavoured dairy drinks could be wider than national if this market includes branded and PL products, as the notifying parties claimed in their reply to the Statement of Objections. On the other hand, if the products markets include only branded products, these markets have a national dimension.

1125. First, with regard to the Netherlands, several German producers such as Sachsenmilch, Immergut and Hochwald already supply Dutch retailers with private label long-life dairy drinks. Belgian producers Inex and Inza supply also Dutch retailers with private label products.

1126. In Belgium, the bulk of the supply of branded products is made by Dutch producers (the notifying parties) and some smaller Belgian suppliers (Inex, Inza and Vitalac). German companies such as Milch Union Hocheifel and Hochwald are present as well but only with PL products.

1127. It follows from these elements that retailers in the Netherlands and in Belgium source their PL products from their own Member States but also, to a significant extent from other Member States, especially from Germany.

1128. However, when it comes to branded products, the situation is different. The branded markets of chocolate-flavoured and fruit-flavoured dairy drinks are dominated by national brands, in particular the ones of the notifying parties in the Netherlands but also in Belgium (albeit with sometimes different names such as Chocomel/Cécémel). Friesland Foods' fruit-flavoured dairy drinks brand Fristi is considered as a must-have brand in Belgium and the Netherlands (see Section 11.3.3, competitive assessment). Chocolate-flavoured LLDD brands Chocomel (Netherlands) and Cécémel (Belgium) have a different name but are in essence the same product with similar packaging. They are also considered as must-have brands in these Member States. There are also some national brands in Belgium and the Netherlands exclusively sold in their home Member States: BonOmel in the Netherlands (Campina, used in chocolate and fruit), Stabilac in Belgium (chocolate) and Chocovit in Belgium (Vitalac, chocolate).

1129. Brands available in Germany differ to some extent from the ones sold in Belgium or in the Netherlands. For example, Fristi, which is regarded as a must-carry brand of Friesland Foods in Belgium and the Netherlands, is not available in Germany. The same is true for Yazoo, one of Campina's main brands in fruit-flavoured and chocolate-flavoured dairy drinks in Belgium (also used to a small extent in chocolate in the Netherlands). By contrast, some of the notifying parties' brands are only available in Germany and not in Belgium or the Netherlands, such as Südmilch (Campina), Gastro (Campina) and Domo (Friesland Foods).

1130. On the basis of these elements, it is concluded that on a market including PL and branded products, the geographic scope is wider than national and includes Belgium, the Netherlands and Germany. If the product market at upstream level is limited to branded products, given that brands differ to a large extent between Member States, these markets are national in scope.

11.3.3. Competitive assessment

1131. The concerns expressed by market players and the arguments put forward by the notifying parties have been carefully assessed and the conclusion is that the proposed transaction would lead to non-coordinated effects on the upstream markets for branded long-life dairy drinks in the Netherlands (section 11.2.3.1) and in Belgium (section 11.2.3.2). The conclusion would not be different if the upstream market were to include branded and PL products (section 11.2.3.3) with a wider geographic scope (Belgium, Germany and the Netherlands).

1132. As Friesland Foods is not active in PL, the upstream markets for sourcing of PL dairy drinks (fruit or chocolate-flavoured) only is not addressed in the decision. Furthermore, since the geographic scope of the possible sourcing market comprising branded and PL products includes Belgium, the Netherlands and Germany, national markets for the sourcing of branded and PL chocolate-flavoured dairy drinks and for the sourcing of branded and PL fruit-flavoured dairy drinks are not relevant and are therefore not assessed in the decision.

1133. In Germany, Friesland Foods is not active in the OOH segment. In the retail segment, the only affected market at upstream level would be the market for chocolate-flavoured dairy drinks where the notifying parties hold a combined market share of [20-30]*% in value, albeit with a small overlap ([0-5]*%). The notifying parties would furthermore face competition from German companies such as Müller, Zott and Nordmilch. Therefore Germany is not assessed as a national market in this decision and the only national geographic markets that are assessed at upstream level are Belgium and the Netherlands.

11.3.3.1. Branded LLDDs - The Netherlands

11.3.3.1.1. Merging firms have large market shares

1134. According to the notifying parties the upstream market for branded chocolate long-life dairy drinks (retail and OOH) covering the Netherlands had a total value of EUR 70 782 000 in 2007. Market shares of the notifying parties and their competitors in value are shown in Table 11-7.

Table 11-7: Market Shares upstream for sourcing of branded chocolate-flavoured long-life dairy drinks The Netherlands - Source: Form CO, Section 7.F, table 79.

1135. On this market, the new entity would hold a market share of [90-100]*% (Friesland Foods [90-100]*%, Campina [0-5]*%) which represents an increase of more than [0-5]* percentage points ([80-90]*%) since 2005. Main competitors of the merged entity are Belgian companies Inza ([0-5]*%) and Inex ([0-5]*%). Market shares would be of the same magnitude if the retail and the OOH segment were to be separated (714).

1136. It therefore appears that Friesland Foods with its brand Chocomel dominates the Dutch market for chocolate flavoured dairy drinks since Chocomel accounts for more than [90-100]*% of the sourcing of chocolate flavoured dairy drinks by retailers and OOH wholesalers. By contrast, Campina's presence appears limited ([0-5]*%) but it remains the strongest branded alternative to Chocomel in this market especially in light of the fact that Inex and Inza are not considered as credible brands in the Netherlands. Under these market conditions, even if it is true that Campina is a weak actor on this market, the removal of the main alternative to a dominant brand is detrimental to competition.

1137. With respect to fruit-flavoured dairy drinks, the upstream market for branded fruit-flavoured dairy drinks (retail and OOH) covering the Netherlands had a total value of EUR 36 986 000 in 2007 according to the notifying parties. Market shares of the notifying parties and their competitors in value are shown in Table 11-8.

Table 11-8: Market Shares upstream for sourcing of branded fruit-flavoured long-life dairy drinks The Netherlands - Source: Form CO, Section 7.F, table 73.

1138. On this market, the new entity would hold a market share of [90-100]*% (Friesland Foods [50-60]*%, Campina [30-40]*%) which represents an increase of [0-5]* percentage points ([90- 100]*%) since 2005. Main competitors of the merged entity are Belgian companies Inza ([0- 5]*%) and Inex ([0-5]*%). Market shares would be of the same magnitude if the retail and the OOH segment were to be separated (715).

1139. The merger would therefore create a quasi-monopoly in the market for branded fruitflavoured dairy drinks in the Netherlands by eliminating the main alternative to the Friesland Foods brand Fristi. For the same reason than in chocolate-flavoured dairy drinks, Inex and Inza are not considered as possible alternatives.

11.3.3.1.2. Merging Firms are the closest competitors

1140. Another important aspect is the degree of substitutability between the merging firms' products. The higher the degree of substitutability between the products of the notifying parties, the more likely it is that the merging firms will significantly raise prices after the merger.

1141. A large majority of Dutch competitors and customers (retailers and OOH wholesalers) explained during the investigation that they view Campina and Friesland Foods as the closest competitors on the market, especially for fruit-flavoured dairy drinks (716). Customers viewed them as their first and second choice, and their rivalry as an important source of competition on the market. For example one customer indicated, with reference to the main brands of Friesland Foods and Campina in fruit-flavoured dairy drinks, that "Fristi (Friesland Foods) and Yogho Yogho (Campina) have been competing from the get go. Both of them tried to be in the shelves exclusively" (717). This was confirmed by a second competitor which stated that "Yogho Yogho is indeed the main competitor of Fristi" (718).

1142. A third customer explained that "In the branded segment, Friesland and Campina are each others' closest competitors" (719). This customer explained more in detail with respect to the notifying parties' brands in fruit-flavoured dairy drinks that "Yogho Yogho (Campina) and Fristi (Friesland Foods) are both yoghurt drink brands which are aimed at the same consumer group and have the same placing on the shelves".

1143. With respect to chocolate flavoured dairy drinks, respondents to the market investigation consider that no brands can bear the comparison with Chocomel since "Chocomel has turned into a generic. For instance a customer in a restaurant will typically order a Chocomel instead of asking for a chocolate milk drink" (720). However, as mentioned in recital 1136, the merger would eliminate the main branded alternative to Chocomel.

1144. With respect to this issue of closeness of competition, the Commission has attached a set of econometric results on estimation of retail demand systems to its Statement of Objections to assess in particular whether supplier brands exert on each other a greater competitive constraint than private labels. The notifying parties have raised a number of criticisms regarding the interpretation and the robustness of the econometric results in their response to the Statement of Objections. (721)

1145. On the one hand, some of the criticisms put forward by the notifying parties are invalid or based on a misunderstanding of the methodology. On the other hand, a number of criticisms regarding the appropriateness of the model specifications related to the long life dairy drinks segment has merit. On balance, the Commission decided not to attach weight to the econometric evidence. (722) It is important to emphasise that this conclusion does not in any way contradict or affect the merits of the other, qualitative and quantitative evidence put forward in this section with respect to the issue of closeness between suppliers brands in the long-life dairy drinks segment.

1146. In their reply to the Statement of Objections, the notifying parties argued that this conclusion is rebutted by the purchasing patterns based on GFK data mentioned in recital 1094. This data shows that end-consumers that usually buy Chocomel or Fristi do not see a Campina brand as their second best choice. These customers prefer PL as their second choice (723).

1147. First, with respect to the results of this study, a preliminary remark is that it refers to purchasing behaviour at downstream level, when end-consumers choose between different products available, and not to the sourcing patterns of retailers and OOH wholesalers, where the merger is likely to have an impact.

1148. More importantly, this study shows that for example, an average buyer of a Friesland Foods fruit-flavoured LLDD in the Netherlands sources in volume [20-30]*% of its other fruitflavoured LLDDs purchase from Campina ([10-20]*% from Yogho Yogho and [10-20]*% from Bonomel fruit), [20-30]*% from Aldi/Lidl chocolate-flavoured private label, [20-30]*% from other fruit-flavoured PL and [10-20]*% from other fruit-flavoured brands.

1149. These results show that Campina fruit-flavoured brands are purchased more by Dutch households at downstream level than other brands. With respect to private labels, it remains unclear how these elements are indicative of the fact that private label products are closer competitors from Friesland Foods than Campina products are, and to which extent this split of volumes from non-exclusive buyers between branded and PL products measures the competitive pressure that Campina brands or PL products currently exercise on Friesland Foods brands. Therefore these results, even with respect to purchasing patterns of Dutch consumers downstream, are not conclusive on the issue of closeness of competition.

11.3.3.1.3. Merging firms' brands are strong brands.

1150. In their response to the 6(1)(c) Decision (724), the notifying parties strongly contested that Campina has strong brands. While they acknowledge that Friesland Foods has strong, "mustcarry" brands, they claimed that Campina has only weak brands. For fruit-flavoured dairy drinks, they submit that Yogho Yogho, one of Campina's brands, is sold at a lower price than private labels, while Chocomel and Fristi (Friesland Foods) are more expensive than PL, which in itself would show that the Campina products are in a completely different league to Friesland Foods'. For chocolate-flavoured dairy drinks, the notifying parties argued that Campina's share is very small, even negligible.

1151. In the reply to the Statement of Objections, the notifying parties reiterated that Campina brands in the Netherlands are small low-priced products that have no impact in the marketplace whatsoever (725).

1152. The Commission's market investigation did not support the notifying parties' view as regards the alleged limited strength of Campina's brands.

1153. With regard to fruit-flavoured LLDDs in the Netherlands, retailers unanimously confirmed during the market investigation that they view Yogho Yogho as an A-brand (726) which competes with Fristi, also an A-brand. This is also supported by the fact that historically, as can be observed in Figure 11-10, Campina Yogho Yogho was priced above fruit-flavoured private label, except for the beginning of 2008, reflecting the generalised price hike experienced by private labels since April 2007 in particular.

[...]*

Figure 11-10: Evolution of Average weighted prices by brand

1154. The fact that Yogho Yogho was priced for the first months of 2008 below private labels' fruitflavoured products cannot be seen as evidence for a long-term trend since it has consistently been priced above private labels in the last four years and the difference has been reduced sharply only as of September 2007.

1155. Retailers confirmed that Yogho Yogho is an important brand that they need to have in their assortment much like Fristi and Chocomel (727). Furthermore, Yogho Yogho and Fristi are aiming at children as shown in Figure 11-11.

Figure 11-11: Segment Share by Brand

1156. In the Netherlands, the price of Friesland Foods' brands - Fristi and Chocomel - fluctuate heavily, reflecting steep temporary discounts. Such discounts appear particularly effective in expanding sales as can be observed by the strong fluctuations in market shares over time, as shown in Figure 11-12. It is worth mentioning that in the Netherlands, Campina Yogho Yogho and Friesland Foods Fristi have comparable and stable market shares.

Figure 11-12: Evolution of Market Shares

1157. Likewise, with regard to Chocolate-flavoured dairy drinks in the Netherlands, customers confirmed that Campina Choco Choco and Yazoo should be viewed as A-brands (728) whereas Campina's brand BonOmel is viewed as a C-brand (brand priced below private label). Inex and Inza brands, on the other hand, are viewed as C-brands in the Netherlands.

1158. The notifying parties claimed in the reply to the Statement of Objections that these brands have a low percentage of exclusive buyers ([5-10]*% for Yogho Yogho, [5-10]*% for Campina Choco Choco) - especially compared to Chocomel or Fristi and that therefore these brands are not strong (729).

1159. However, the notifying parties themselves admit that brand loyalty is not very high in the LLDDs markets. For example the percentage of exclusive buyers of Chocomel is [...]*% and it is [...]*% for Fristi. In light of these figures, percentages of [...]*% for Yogho Yogho and [...]*% for Campina Choco Choco are not really indicative that these brands are much weaker than Chocomel or Fristi. Moreover, market shares have also to be taken into account when analysing these purchasing patterns since it is more likely in the case of a brand with a high market share that a household which has initially purchased this brand, purchases the same brand in all its subsequent purchases. In sum, the relatively low percentage of exclusive buyers for Yogho Yogho or Campina Choco Choco mainly shows that their market share is relatively low compared to Fristi and Chocomel. This is not surprising given the strength of Fristi and Chocomel brands but does not mean that Campina brands do not exert competitive pressure on Friesland Foods brands.

1160. In conclusion, the quantitative and qualitative evidence described in Section 11.3.3.1.3. show that Campina brands are in the Netherlands considered as A-brands in direct competition with Friesland Foods brands. This assessment is valid for Yogho Yogho and Campina Choco Choco.

11.3.3.1.4. Customers are unlikely to switch supplier

1161. Customers have not only informed the Commission that they would view the notifying parties as closest substitutes to each other, but also that they would have difficulties in switching to other suppliers. This is valid for the retail segment as well as the OOH segment.

1162. In the Netherlands, since there are only two alternative suppliers currently active on the Dutch market (Inex and Inza), which are mainly active as suppliers of private label and hold each a market share below [0-5]*%, retailers and OOH wholesalers would have no alternative in the branded segment to the notifying parties. This is true for the chocolate-flavoured dairy drinks segment and the fruit-flavoured dairy drinks segment.

1163. While it is correct that private label products compete on the shelves against the brands of the notifying parties, switching from branded products to private label products in case of a price increase for branded LLDDs would not be an acceptable solution because all retailers still rely to a significant part on branded products (60% of total supply in 2007 in the Netherlands, in value). For OOH wholesalers, the issue is more acute since there are no private label products in the OOH segment. As recognized by the notifying parties themselves, Chocomel and Fristi are mustcarry brands in the Netherlands and the market still has a very strong brand-orientation.

1164. Retailers in the Netherlands indicated that they might switch to private label products in order to replace C-brands by PL (this is particularly valid for BonOmel in fruit-flavoured and chocolate-flavoured LLDDs) as an alternative but that it would be impossible to substitute Abranded products (Yogho Yogho or Campina Choco Choco) by private labels (730). It appears that these brands of the notifying parties are too strong to either delist or otherwise reduce their position.

11.3.3.1.5. Lack of countervailing buyer power

1165. The notifying parties also argue that countervailing buyer power from the largest retail chains (Albert Heijn, SuperUnie, Schuitema, or Super de Boer in the Netherlands) would be a disciplining factor of great importance.

1166. According to the Commission's horizontal merger guidelines, countervailing buyer power should be understood as the bargaining strength that the buyer has vis-à-vis the seller in commercial negotiations due to its size, its commercial significance to the seller and its ability to switch to alternative suppliers.

1167. In a market structure where the notifying parties would have a quasi-monopolistic position in the Netherlands and where, as demonstrated in Section 11.3.3.1.4, switching possibilities would be very scarce and even, for some categories of products, non-existent, the countervailing buyer power of retailers and OOH wholesalers is very limited.

1168. Furthermore, it is not sufficient that buyer power exists prior to the merger, it must also exist and remain effective following the merger. This is because a merger of two suppliers may reduce buyer power if it thereby removes a credible alternative. In the present case, as described in Section 11.3.3.1.2, the merger will remove one of the two main suppliers considered to be the closest competitors in the market, since alternative suppliers could not be identified in the course of the market investigation.

1169. The notifying parties argued that because of the strong presence of private label products, retailers have a strong range of instruments to influence negotiations with branded suppliers, even if there would only remain a sole branded supplier on the market. The notifying parties list strategies that can be used by retailers, such as charging additional payments for stocking suppliers' products or creating production and financial uncertainties that complicate their planning decisions.

1170. These instruments described by the notifying parties are rather general and the market investigation did not show that these strategies were employed for LLDDs in the Netherlands. The notifying parties claim that retailers have the possibility to delist products and indeed consider such a measure as delisting. However, unlike in other market where examples of delistings have been presented, the notifying parties fail to provide any elements which would show that this measure has been used against Campina or Friesland Foods LLDDs. Given the alleged limited strength of Yogho Yogho or Campina Choco Choco, it is unclear why retailers did not delist them. One possible explanation is that they consider necessary to have another brand in their shelves to balance with Friesland Foods' must-have brands Chocomel or Fristi. This possibility would disappear after the merger.

1171. Therefore, it can be concluded that countervailing buyer power post-merger would not be sufficient to off-set potential adverse effects of the merger.

11.3.3.1.6. Entry unlikely to occur

1172. The market investigation confirmed that there have been no recent entries in the Dutch market for LLDDs in recent years (except for the Heineken's products Joy Fruity and Joy Choco, only available since beginning of 2008 in the Netherlands in the OOH segment).

1173. It also highlighted that market players do not expect any new entry in the Dutch branded market for LLDDs, especially in the branded segment. Given the strong brand orientation of the market, brand/company image among consumers plays a significant role in these markets. Therefore, entering these branded markets require considerable investment and time needed to "build" a brand and raise consumers' awareness thereof. This is all the more difficult as Friesland Foods strongly promotes its Fristi and Chocomel brand (Friesland Foods spends consistently [...]*% of its annual turnover in long-life dairy drinks on advertisement and promotion (731)).

1174. During the market investigation, it was investigated whether German competitors would be in a position to enter the branded segment in the Netherlands by supplying retailers with branded products in the case of a price increase. However, the market investigation (732) showed that in such a market structure (must-carry brands for Friesland Foods and strong brands for Campina +, private labels), there is little room for a newcomer to break through with a branded product, unless it positions its brand as a "C-brand" which would not compete directly against the notifying parties' brands. Entering the branded segment with an A brand would involve promotion and listing expenses and therefore margins would be too small to sustain successful entry.

1175. Finally, the notifying parties indicated that the markets for LLDDs are mature and commoditized markets in the Netherlands, making an entry less attractive.

11.3.3.1.7. Conclusion on the market for branded LLDDS in the Netherlands

1176. For these reasons, it is concluded that the transaction is likely to lead to a significant impediment of effective competition in the markets for branded long-life dairy drinks in the Netherlands.

11.3.3.2. Branded LLDDs - Belgium

11.3.3.2.1. Merging firms have large market shares

1177. According to the notifying parties the upstream market for branded chocolate long-life dairy drinks (retail and OOH) covering Belgium had a total value of EUR 36 116 000 in 2007. Market shares of the notifying parties and their competitors in value are shown in Table 11-9.

Table 11-9: Market Shares upstream for branded chocolate-flavoured long-life dairy drinks Belgium - Source: Form CO, Section 7.N, table 97.

1178. On this market, the new entity would hold a market share of [70-80]*% (Friesland Foods [60- 70]*%, Campina [5-10]*%) which has slightly increased since 2005 ([70-80]*%). Main competitors are Belgian companies Vitalac ([5-10]*%), Inex ([5-10]*%) and Inza ([0-5]*%).

1179. If the retail and the OOH segment are separated, the combined market share would be in value [80-90]*% in the retail segment (with an overlap of [10-20]*%) and [60-70]*% in the OOH segment (with an overlap of [0-5]*%).

1180. With respect to fruit-flavoured dairy drinks, the upstream market for branded fruit-flavoured dairy drinks (Retail and OOH) covering Belgium had a total value of EUR 8 593 000 in 2007 according to the notifying parties. Market shares of the notifying parties and their competitors in value are shown in Table 11-10.

Table 11-10: Market Shares upstream for branded chocolate-flavoured long-life dairy drinks Belgium - Source: Form CO, Section 7.N, table 91.

1181. On this market, the new entity would hold a market share of [70-80]*% (Friesland Foods [60- 70]*%, Campina [5-10]*%) a slight increase from 2005 ([70-80]*%). Main competitors are Inex ([10-20]*%) and Inza ([5-10]*%).

1182. If the retail and the OOH segment are separated, the combined market share would be in value [70-80]*% in the retail segment (with an overlap of [10-20]*%) and [80-90]*% in the OOH segment (with an overlap of [5-10]*%).

11.3.3.2.2. Merging Firms are close competitors

1183. A large majority of Belgian competitors and customers (retailers and OOH wholesalers) explained during the investigation that they view Campina and Friesland Foods as the closest competitors on the market, both for chocolate and fruit-flavoured dairy drinks (733). Like in the Netherlands, customers viewed them as their first and second choice and their rivalry as an important source of competition on the market (734).

1184. With respect to the study submitted by the notifying parties (which allegedly indicates that PL products are closer competitors to Friesland Foods brands than Campina brands), the same comments as for the Netherlands can be made. First, it refers to purchasing behaviour at downstream level, when end-consumers choose between different products available, and not to the sourcing patterns of retailers and OOH wholesalers, where the merger is likely to have an impact. More importantly, it remains unclear how these elements are indicative of the fact that private label products are closer competitors to Friesland Foods than Campina products are and to which extent this split of volumes from non-exclusive buyers between branded and PL products measure the competitive pressure that Campina brands or PL products currently exercises on Friesland Foods brands. Therefore these results, even with respect to purchasing patterns of Belgian consumers downstream, are not conclusive on the issue of closeness of competition

11.3.3.2.3. Merging firms' brands are strong brands.

1185. In Belgium, Campina is active with several brands, namely Yazoo (chocolate and fruit), Yogho Yogho (fruit), Campina Choco Choco (chocolate) and Stabilac (Chocolate). During the market investigation, customers (735) indicated that these brands, with the exception of Stabilac, are positioned as A-brands and compete directly with Fristi and Cécémel. With regard to fruitflavoured dairy drinks, Yazoo is considered as the closest competitor of Fristi and one retailer underlined that its basic assortment for fruit-flavoured dairy drinks comprises only Fristi and Yazoo.

1186. Other A-brands for Belgian retailers include Inex (fruit and chocolate), Inza (fruit and chocolate) and Chocovit (chocolate only). However Inex and Inza are hardly available in the French-speaking regions of Belgium and are therefore not considered as close competitors to Fristi and Cécémel (736). Chocovit is only available in the chocolate-flavoured segment and therefore it does not cover the whole range of LLDDs.

1187. The notifying parties claimed in the reply to the Statement of Objections that these brands have a low percentage of exclusive buyers (Campina Choco Choco [5-10]*%, Yazoo [10-20]*% and Yogho Yogho [10-20]*%) - especially compared to Cécémel and that therefore these brands are not strong (737).

1188. However, like in the Netherlands, Cécémel and Fristi do not exhibit a strong degree of brand loyalty (respectively [30-40]*% and [20-30]*%) and therefore no definitive conclusions can be drawn from the percentage of exclusive buyers of Campina brands. Moreover, as mentioned in Section 11.3.3.1 regarding the Netherlands, market shares must also be taken into account when analysing these purchasing patterns since it is more likely in the case of a brand with a high market share, that a household which has initially purchased this brand purchases the same brand in all its subsequent purchases. In sum, a relatively low percentage of exclusive buyers for Yogho Yogho, Yazoo or Campina Choco Choco mainly shows that their market share is relatively low compared to Fristi and Cécémel. This is not surprising given the strength of Fristi and Cécémel brands but does not mean that Campina brands do not exert competitive pressure on Friesland Foods brands.

1189. In conclusion, the quantitative and qualitative evidence described in this section show that Campina brands are in Belgium considered as A-brands and are in direct competition with Friesland Foods brands. This assessment is valid for Yogho Yogho, Yazoo and Campina Choco Choco.

11.3.3.2.4. Customers are unlikely to switch supplier

1190. Customers have not only informed the Commission that they would view the notifying parties as closest substitutes, but also that they would have difficulties in switching to other suppliers. This is valid for the retail as well as the OOH segment.

1191. In Belgium, two alternative suppliers are active in the fruit-flavoured dairy drinks segment (Inex and Inza) and three other companies supply retailers and OOH wholesalers in the chocolate-flavoured dairy drinks segment (Inex, Inza, Vitalac). However, as mentioned in recital 1186, Inex and Inza have a limited geographic coverage and are mainly active in the private label segment. Moreover, their offerings do not comprise portion packs of 0.2 litre plastic bottle which are available in Campina and Friesland Foods' range.

1192. Vitalac supplies only branded products. However, Vitalac has a limited assortment since it does not supply portion packs of 0.2 litre brick or plastic bottle, which are available in Campina and Friesland Foods' range.

1193. While it is correct that private label products compete on the shelves against the brands of the notifying parties, switching from branded products to private label products in the case of a price increase for branded LLDDs would not be an acceptable solution because all retailers still rely to a significant part on branded products (68% of total supply in 2007 in Belgium in value). For OOH wholesalers, the issue is more acute since there are no private label products in the OOH segment. As recognized by the notifying parties themselves, Cécémel and Fristi are must-carry brands in Belgium and the market still has a very strong brand-orientation.

1194. In Belgium, the answers were more varied than in the Netherlands with regard to the possibilities of substitution. For chocolate LLDDs, retailers submit that switching to private labels could be a possibility but argued that for fruit-flavoured LLDDs, alternatives to the branded products of the notifying parties are very limited and that switching to PL is excluded since end-consumers are willing to buy these specific brands (738).

11.3.3.2.5. Lack of countervailing buyer power

1195. The notifying parties also argue that countervailing buyer power from the largest retail chains (Delhaize, Colruyt or Carrefour in Belgium) would be a disciplining factor of great importance.

1196. According to the Commission's horizontal merger guidelines, countervailing buyer power should be understood as the bargaining strength that the buyer has vis-à-vis the seller in commercial negotiations due to its size, its commercial significance to the seller and its ability to switch to alternative suppliers.

1197. However, in a market structure where the notifying parties would have very high market shares in Belgium and where, as demonstrated in Section 11.3.3.2.4, switching possibilities would be very scarce, the countervailing buyer power of retailers and OOH wholesalers appear to be very limited.

1198. Furthermore, it is not sufficient that buyer power exists prior to the merger, it must also exist and remain effective following the merger. This is because a merger of two suppliers may reduce buyer power if it thereby removes a credible alternative. In the present case, as described in Section 11.3.3.2.2, the merger will remove one of the two main suppliers considered to be the closest competitors in the market, since alternative suppliers could not be identified in the course of the market investigation.

1199. Like in the Netherlands, the notifying parties argued that because of the strong presence of private label products, retailers have a strong range of instruments to influence negotiations with branded suppliers, even if there would only remain a sole branded supplier on the market. These instruments described by the notifying parties are rather general and the market investigation did not highlight that these strategies were employed for LLDDs in Belgium. The notifying parties claim that retailers have the possibility to delist products and indeed consider such a measure as delisting. However, unlike in other market where examples of delistings have been presented, the notifying parties fail to provide any elements which would show that this measure has been used against Campina of Friesland Foods LLDDs. Given the alleged limited strength of Yogho Yogho, Yazoo or Campina Choco Choco, it is unclear why retailers did not delist them.

1200. Therefore, it can be concluded that countervailing buyer power post-merger would not be sufficient to off-set potential adverse effects of the merger.

11.3.3.2.6. Entry unlikely to occur

1201. The market investigation confirmed that there had been no recent entries in the Belgian market for branded LLDDs in recent years.

1202. It also highlighted that market players do not expect any new entry in the Belgian market for LLDDs, especially in the branded segment. Given the strong brand orientation of the market, brand/company image among consumers plays a significant role in these markets. Therefore, like in the Netherlands entering these branded markets require considerable investment and time needed to "build" a brand and raise consumers' awareness thereof. This is all the more difficult as Friesland Foods strongly promotes its Fristi and Cécémel brand (Friesland Foods spends consistently [...]*% of its annual turnover in long-life dairy drinks on advertisement and promotion (739)).

1203. Like in the Netherlands, German competitors are not in a position to enter the branded segment in the case of a price increase. The Belgian market structure exhibits the same features as the Dutch one (must-carry brands for Friesland Foods and strong brands for Campina) which gives little room for a newcomer to enter the market with an A-brand which could successfully compete with the merged entity. Entering the branded segment with an A brand would involve promotion and listing expenses and therefore margins would be too small to sustain successful entry.

1204. Like in the Netherlands markets for LLDDs are mature and commoditized markets in Belgium, making an entry less attractive.

11.3.3.2.7. Conclusion on the branded market for LLDDS in Belgium

1205. For these reasons, it is concluded that the transaction is likely to lead to a significant impediment of effective competition in the branded market for long-life dairy drinks in Belgium.

11.3.3.3. Branded + PL LLDDs - Belgium, the Netherlands and Germany

1206. In a market including both brands and PL with a geographic scope covering Belgium, Germany and the Netherlands, the notifying parties would hold a combined market share in value for chocolate-flavoured dairy drinks of [60-70]*% (Friesland Foods [50-60]*%; Campina [5- 10]*%) which would represent an increase of [0-5]*% since 2005. Sourcing of private label products would represent 23.0% of the market and sourcing of branded products 77%. The combined market share of the notifying parties would be [60-70]*% in the retail segment and [60-70]*% in the OOH segment (740).

1207. Main competitors of the notifying parties would be much smaller in size than the new entity. The second player would be German company Immergut with a market share of [5-10]*% thanks to a very significant volume of sales of PL products (Immergut achieved a turnover above EUR [...]* in the private label segment in 2007) and the third one would be Müller which would hold a market share of [5-10]*% (branded + PL products).

1208. With respect to fruit-flavoured dairy drinks, the notifying parties would hold a combined market share of [50-60]*% (Friesland Foods [20-30]*%; Campina [20-30]*%) which has decreased by [5-10]*% since 2005. Sourcing of private label products would represent 32.9% of the market and sourcing of branded products 67.1%. The combined market share of the notifying parties would be [40-50]*% in the retail segment and [60-70]*% in the OOH segment (741).

1209. Main competitors of the notifying parties would be much smaller in size than the new entity. The second player in the fruit-flavoured dairy drinks market would be Müller with a market share of [5-10]*% (brands + PL) and the third one would be Inex with [0-5]*%.

1210. It follows that even on a market including both branded and PL covering Belgium, the Netherlands and Germany, market shares would still be very high since the merged entity would account for [60-70]*% of the chocolate-flavoured dairy drink market and [50-60]*% of the fruitflavoured dairy drinks market (slightly less in the retail segment and more in the OOH segment).

1211. Hence it appears that market data does not confirm the statements of the notifying parties that "Market shares of the parties are (very) low if the geographic market is the combination of the Netherlands, Belgium and Germany" (742) or that "The high market shares are a direct result of the Commission's narrow market definition" (743).

1212. With respect to the closeness of competition in this wide product and geographic market, the market investigation in first phase indicated that for some customers, Inex (mainly in fruitflavoured), Inza (mainly in chocolate-flavoured) and Immergut (mainly in chocolate-flavoured) could be considered as equally close competitors to Campina as Friesland Foods. This is particularly due to the fact that these companies supply large volumes of PL products to the market. However, with respect to the branded segment, it is beyond doubt that Campina and Friesland Foods are the closest competitors. Since the branded segment still represents 77% of the market in value for chocolate-flavoured dairy drinks and 67.1% for fruit-flavoured dairy drinks in 2007, Campina and Friesland Foods are the closest competitors for the main part of these markets.

1213. The conclusion with regard to the strength of brands equally applies in this wide market.

1214. With respect to switching possibilities, it appears at first sight that retailers and OOH wholesalers would have the possibility to switch to alternative suppliers of PL should the notifying parties increase prices. However, it is also clear that the notifying parties are the closest competitors in this wide market and therefore switching to alternative suppliers would not be an easy task. Furthermore, customers would have difficulties to switch to PL to replace the notifying parties' A-brands or reduce their shelf space since end-consumers expect these brands to appear on the shelves. Even if PL and branded products are in the same market, the fact that the notifying parties are the closest competitors in the branded segment does not enable customers to change their sourcing patterns easily.

1215. As mentioned in the section regarding product market definition, the notifying parties argued that Friesland Food brands are primarily constrained by private labels and not by Campina brands. The notifying parties put forward that competition between brands and PL is underscored by the resemblance of packaging between branded and PL products - the packaging being sometimes almost identical. The notifying parties point out also that Dutch retailer Albert Heijn has launched recently a "blind-test" campaign where consumers can test branded and PL products in an Albert Heijn shop, in order to convince consumers that these products have the same quality, PL products being cheaper. This campaign concerns a wide range of products and hence not only LLDDs.

1216. It should first be mentioned that this promotional campaign has triggered strong reactions by producers of branded goods in the Netherlands, including Friesland Foods, aimed at persuading consumers that branded products have specificities which distinguish them from private label products. It is worth noting that Friesland Foods chose specifically Chocomel for that particular campaign (744).

1217. Moreover, it appears that this resemblance of packaging or the promotional campaigns carried out by retailers mainly shows that competition exists between branded and PL products at the downstream level, where end-consumers choose between different products available. These elements are not indicative that retailers consider that the main alternative to a price increase of must-carry brands of Friesland Foods lies in the sourcing of more private label products and the reduction of shelf space for these must-carry brands. On the contrary, as mentioned in Sections 11.3.3.1 and 11.3.3.2, retailers have consistently indicated in the market investigation that Campina and Friesland Foods are the closest competitors with respect to brands and that switching to PL is not a potential alternative given that end-consumers expect to find these branded products on the shelves.

1218. The existence of countervailing buyer power is mainly linked to the availability of alternatives in the branded segment and although the set of potential suppliers would enlarge if Germany were included, no German supplier has a comparable strength to Friesland Foods or even Campina in the market including branded and PL products. Moreover German suppliers do not envisage widening their presence (745) and the entry of a "foreign" supplier (neither Dutch, nor Belgian or German) is unlikely.

1219. For these reasons, it is concluded that the transaction is likely to lead to a significant impediment of effective competition in the market for branded and PL long-life dairy drinks in Belgium, the Netherlands and Germany.

11.3.3.4. Conclusion on competitive assessment

1220. For the reasons set out in Section 11.3.3, the notified concentration is likely to significantly impede effective competition on the market for branded long-life chocolate-flavoured dairy drinks in the Netherlands, the market for branded long-life fruit-flavoured dairy drinks in the Netherlands, the market for branded long-life chocolate-flavoured dairy drinks in Belgium, the market for branded long-life fruit-flavoured dairy drinks in Belgium, the market for branded and private label long-life chocolate-flavoured dairy drinks in the Netherlands, Belgium and Germany and the market for branded and private label long-life fruit-flavoured dairy drinks in the Netherlands, Belgium and Germany, these three Member States being a substantial part of the common market, regardless of whether these markets need to be further segmented according to the distribution channel.

12. FRESH DAIRY DESSERTS

1221. Campina and Friesland Foods are both active in the production and supply of fresh dairy desserts such as fresh custard, porridge and portion packed desserts like mousse, puddings or tiramisu.

1222. The product portfolio of Campina consists mainly of different types of custard ("vla"), a variety of (chocolate) mousse and puddings and some porridges ("pap"). Campina sells fresh dairy desserts in the Netherlands under - among others - the Campina, Mona, Vifit and Optimel brands as well as private label. In Germany and Austria Campina sells fresh dairy desserts mainly under the Landliebe, the Campina and a variety of smaller brands and also private labels. In general, Campina sells fresh dairy desserts under various brands and private labels in many Member States.

1223. The product portfolio of Friesland Foods consists of a variety of custard and porridges. Friesland Foods markets its fresh dairy desserts only in the Netherlands, under the Friesche Vlag brand and private labels.

12.1. Relevant Product Markets

1224. The notifying parties submit that the market for fresh dairy desserts includes a variety of ready-made sweet dairy desserts including custard, portion packs such as mousse, (rice) puddings, profiteroles, tiramisu and porridges.

1225. Fresh dairy desserts are generally produced from dairy products such as milk or cream, in which - among others - eggs, thickening agents (such as starch), sugar, flavours and/or fruit are added.

1226. Custard is a sweet dairy product produced from raw milk, which is first partially skimmed and pasteurised. Subsequently aromas, sugar and thickening agents are added which provides taste and increases viscosity. It is a spoonable product with a shelf life of 16-25 days if stored and transported refrigerated. The notifying parties distinguish between basic custard and value added custard. Basic custard is all custard containing only one aroma such as vanilla or chocolate and all custard containing two aromas such as Dubbelvla. Value added custard is custard which compared to the basic custard needs special processing and/or additional ingredients, and is also sold in portion packs. 1227. Fresh dairy desserts which are sold in portion packs includes added value custard in portion packs, mousse, pudding and "luxury" desserts such as tiramisu, crème brûlée etc. The average shelf life of portion packed desserts is 25-35 days if stored and transported refrigerated.

1228. Porridges are produced from milk or buttermilk, whey and a type of grain which determines the taste of the product. Grain types which are often used include oatmeal, barley and wheat. The average shelf life of porridges is 12-19 days. Total sales of porridges in the Netherlands do not exceed EUR 9.8 million per year. The notifying parties are not aware of substantial volumes of ready made porridge being sold in other Member States.

12.1.1. Custard in gable top, porridge and fresh dairy desserts in portion packs each form a single product market

12.1.1.1. Product market definition proposed by the notifying parties

1229. Friesland Foods and Campina in their notification (746) submit that the market for dairy desserts is heterogeneous, but should be considered as one product market. The range of products constituting the relevant market are so differentiated in part because that would be part of the fun of consuming desserts.

1230. While the notifying parties put forward that the production process are to a large degree the same, they also indicate that the filling and packaging process for custard, portion pack desserts and porridge would differ due to different recipes, packaging types and require additional equipment (additional boiling and packaging equipment for porridge, special designed filling machines for custard). (747)

1231. Despite the heterogeneity of the market, the notifying parties submit that from a demand side perspective fresh dairy desserts are substitutable as they fulfil the same needs of consumers (completing a meal), although the consumption pattern as well as consumer group would be different for porridge compared to custard and portion pack desserts. Porridge would also be consumed at breakfast and has an important nutritional value for (mainly) elderly people. Custard in gable top is a typical Dutch product sold in the Netherlands only and mainly consumed by families.

1232. The described heterogeneity would not give rise to separate markets according to the notifying parties, but may result in segmentation. They submit (748), that where a further segmentation is made on the market for fresh dairy desserts, the market would have to be subdivided in three segments (i) custard in gable top packing, (ii) portion packed desserts and (iii) porridge.

1233. The notifying parties also submit that fresh dairy desserts and value added yoghurts belong to separate relevant product markets, although yoghurts may also be consumed after a meal as a dessert (749). They put forward that the main consumption of value added yoghurt takes place during the day for breakfast, lunch and as an in-between snack. Moreover, the taste of desserts is normally sweeter and the texture differs. Finally, yoghurt is a fermented product while desserts are non-fermented and have more artificial ingredients, and are perceived as less healthy than value added yoghurt.

1234. During the market investigation, a majority of customers confirmed that the two products should belong to separate markets because of different consumption patterns, taste differences and motives for consumption (health vs indulgence). Therefore, value added yoghurt is considered to be a separate market from dairy desserts.

12.1.1.2. Assessment of the Commission

1235. The market investigation has not confirmed the broad market definition proposed by the notifying parties.

1236. Respondents to the market investigation (750) argued that custard in 1 litre gable tops are usually consumed by families and considered to be a basic product non-substitutable by portion pack desserts. One customer (751) for example stated that "vla does not compete with value added more luxurious desserts. They have different consumer group and different consumption moments." Similar a second retailer (752) "We see Vla more as a basic product which does not directly compete with the luxury dessert segment" (753), while porridge is mainly consumed by elderly people and also at breakfast because of its nutritional value. Portion pack desserts in contrast represent more of a premium segment with products consumed at weekends or for special occasion and by single or two person households.

1237. Prices differ significantly per unit of volume between custard and porridge and portion packs: custard/porridge has a retail price around EUR 1.5 per litre, while portion packs are sold on average at EUR 3 per litre. (754)

1238. Asked whether a 5-10% price increase for custard would result in a switch to portion packs, the majority of respondents - customers as well as competitors - expressed the opinion that a switch would be unlikely as the price difference between the two types of desserts would still be high and custard would be consumed to cover basic needs. (755)

1239. In relation to porridge, the replies received during the market investigation were mixed. (756) Roughly half of the respondents were of the opinion that porridge does not belong to fresh dairy desserts as it has a different nutritional value and consumption pattern, while others concluded that despite of these different pattern porridge still falls into the desserts category. (757)

1240. Based on the results of the market investigation, this Decision considers fresh custard in gable top, porridge and portion pack dairy desserts as separate relevant product markets. Because Friesland Foods is not active in portion pack desserts and the proposed merger would not lead to a significant impediment of effective competition in the market for portion pack desserts, the focus is only on custard in gable top and porridge.

12.1.2. Private label and supplier brands belong to the same relevant upstream market

12.1.2.1. Position of the notifying parties

1241. As explained in section 7.1.1.2 the notifying parties agree in general to a vertical segmentation of a particular product market for consumer products between an upstream market for the sourcing of products by retailers and OOH wholesalers serving hotels, restaurant, catering services and filling stations) and a downstream market for the sale by retailers to consumers for two reasons.

1242. The approach recognises that on the retail-to-consumer level private label products compete with (producer) branded products, and that the market share of private label products should be attributed to the retailers that own the private label brands.

1243. In addition, the approach recognises that the competitive conditions on the markets on which retailers and OOH wholesalers source their products are fundamentally different from the competitive conditions on the retailer-to-consumer markets. (758)

1244. The notifying parties explain (759) that the procurement of private label products and the procurement of branded products by retailers (and OOH wholesalers) might constitute two neighbouring, but separate markets, because the competitive conditions under which retailers source private label products and branded products are fundamentally different. This may be the case in situations where specific branded products are perceived as "must carry" products by retailers. However, although the notifying parties have well known brands in the fresh dairy desserts markets this is not sufficient for a separation at the upstream level according to the notifying parties as both Campina and Friesland Foods would be under constant pressure from private label products. This is exemplified by the consistent decline of branded products over the years and the possibility of retailers to carry out delistings of branded products or retaliation in the area of private label. (760)

1245. Thus, private label and supplier brands would belong to the same relevant product market at the upstream level according to the notifying parties.

12.1.2.2. Assessment of the Commission

1246. All fresh dairy desserts producers sell their products to retailers or OOH wholesalers which, in turn, sell these products to consumers. Therefore, there are two stages in the supply chain: the upstream level, where fresh dairy desserts products are produced and supplied to retailers/OOH wholesalers and the downstream level with the supply to consumers. Campina and Friesland Foods are only active on the production and supply level.

1247. In a recent consumer goods case (761), the upstream level, where retailers source their products, has been distinguished from a downstream level, where the products are sold on to the final customer. In the present case, the same approach is followed for custard. For porridge this assessment is not necessary since private labels are absent from the market.

1248. Custard is available in two broad categories: brands owned by the dairy manufacturer and private label products of the retailer. (762) The two categories are sourced separately, but are displayed next to each other on the shelves. Campina and Friesland Foods both produce branded and private label custard.

1249. As explained in section 7.2.1.2 (concerning fresh basic dairy products) whether private labels and supplier brands belong to the same product market upstream depends, in particular on: (a) whether both types of brands, on the aggregate, compete closely with each other from the perspective of the end-customer; and (b) the extent to which upstream suppliers of private label and/or brands and the purchasing retailers, take into account the competitive pressure that private labels and supplier brands mutually exert on each other, in their upstream negotiations.

1250. With respect to the first criterion, there is significant evidence suggesting that private label and supplier brands compete in the downstream market and exert a competitive constraint on each other.

1251. In their responses to the market investigation the majority of customers and competitors indicated that private label and branded products compete at the retail level and that the quality of private label products is to a large extend similar to branded products. (763)

1252. In addition, on the downstream markets for fresh custard the share of private label products was 35.5% in value terms in 2007 and has been increasing over the past few years. (764) This can also be observed in the IRI market-level data, provided by the notifying parties. Focusing on IRI data, which excludes discounters, supplier brands have lost some market share in the recent past to private labels.

Figure 12-1: Evolution of Value Sales by Brand Category - Custard. Source: IRI DATA

1253. Other features of the products also suggest that private labels and supplier brands compete within the same market. First, no differentiation can be observed with respect to packaging format and size in fresh custard. Branded products do not distinguish themselves by more expensive packaging or deviating sizes. Campina and Friesland Foods brands are sold primarily in gable top format in containers of mainly 1 litres and some 0.5 litres, with 1 litre accounting for around [...]*% in both cases (765). A similar pattern can be observed for private label brands in the aggregate as shown in Figure 12-2.

Figure 12-2: Package Size by Brand Category - Custard Source: IRI DATA

1254. This broad analysis of IRI market-level data suggests there is a sufficient degree of competitive interaction between supplier brands and private labels for fresh custard, for both to be considered as belonging to the same relevant product market at retail level.

1255. However, additional evidence indicates that upstream suppliers of private label and the purchasing retailers take the competitive pressure that private labels and supplier brands mutually exert on each other into account, in their upstream negotiations. One factor to be considered is that private labels account for a relevant portion - at least 30% (766) - of the overall market, which is growing over time. Given the relative importance of private label sales, suppliers cannot ignore the competitive pressure that private labels exert on their brands. This is specially the case when the suppliers of branded products are also the supplier of private labels and both compete in the downstream market. In this case, the bulk of both the private label and the branded products are supplied by the same firms, namely the notifying parties. It can thus be presumed that the notifying parties would take the substitutability of private label and supplier brands among end customers into account when supplying retailers. (767)

1256. The same is true with respect to supermarket chains that offer private label. Indeed, the major retailers in the Netherlands have two private labels in addition to the premium brands offered by Campina and Friesland Foods. (768) The existence of both high and low-level private labels indicates that retailers consider and condition their private labels in relation to supplier brands. (769) Albert Heijn's prices are similar to the ones of Campina and Friesland Foods for example, while other retailers position their private label further away from supplier brands.

Table 12-1: Average Weighted Prices by Brand - Custard. Source: IRI DATA

1257. In fact, supermarket chains appear to pursue different strategies with respect to the positioning of their private labels vis-à-vis branded suppliers. For example, Albert Heijn sells the majority of custard under its private label ([60-70]*%), in addition to significant Campina products ([30-40]*%) and only a small number of Friesland Foods' branded custard. In contrast, Laurus carries primarily Campina ([50-60]*%) and Friesland Foods ([20-30]*%), while its own private labels (with a share of [10-20]*%) lag behind. Schuitema sells Campina, Friesland Foods and its own brand equally. Finally, purchasing group Superunie relies on private labels ([20- 30]*%), Campina ([40-50]*%), and Friesland Foods ([30-40]*%).

Table 12-2: Value Market Shares by Supermarket Chain - Custard Source: IRI DATA

1258. As confirmed by customers and competitors (770), the procurement of branded products follows a different procedure than the one for private label products. Suppliers and retailers agree on gross price, discounts, listing fees, and promotions in the case of branded products in bilateral negotiation. For private labels, a tender procedure is usually undertaken, followed by a selection of preferred suppliers which agree with the retailer on a net price in final negotiations. This in itself is not decisive if the same suppliers and retailers participate in both segments.

1259. For custard in the Netherlands, Campina and Friesland Foods both offer well-known brands and private label products representing more than [60-70]*% of the respective market (771), and therefore would take the competitive pressure that private labels and supplier brands mutually exert on each other into account in their negotiations.

1260. Taken together, these elements - the competitive interaction between branded and private label products downstream, the differentiated strategy of Dutch retailers with respect to private label and branded products and the activities of the notifying parties in private and branded products - support the view that despite the lower penetration by private label compared to fresh milk, it is necessary to consider private label and branded products as belonging to the same product market upstream.

12.1.3. Whether the markets for fresh custard and porridge have to be further separated according to distribution channel can be left open

12.1.3.1. The position of the notifying parties

1261. Friesland Foods and Campina both sell fresh dairy desserts to the retail and the foodservice channel. The OOH segment serves restaurants, cafes, hotels, catering services, hospitals but also small businesses like bakers or other food processing entities. The notifying parties explain that the OOH market is typically served by foodservice wholesalers who deliver at the doorstep of the customer and by cash-and-carry business. In the Netherlands, direct sales to OOH users constitute only a very small proportion of the entire market segment. (772)

1262. In previous decisions (773) a distinction was made between these sales channels because of differences in services, sales force, price structure, packaging sizes and health and safety regimes. The notifying parties argue that such a distinction is no longer necessary, for several reasons: most of the products sold through the OOH segment are more or less identical to those in the retail segment; both channels partially supply the same customers who cross over depending on the circumstances; in part, players in the segments overlap; prices in both segments show similar developments and the absence of private label in OOH would not be determinative; the logistics of supply and additional services are largely the same.

12.1.3.2. The assessment of the Commission

1263. Several respondents during the market investigation indicated that some differences between the retail and the OOH segment still exist (774):

1264. It has been confirmed that contrary to the notification of the notifying parties, the distribution in the OOH segment is different from the retail segment in the sense that dairy producers try to target end customers directly and use OOH wholesalers only as logistic providers. Friesland Foods and Campina confirmed (775) that they indeed target end-customers directly and these direct contractual relationships represent [60-70]*% of Friesland Foods annual turnover and about [70- 80]*% of Campina's annual turnover. These contracts cover bonus fees, promotional budgets, but also price, category management. OOH wholesalers would be in charge of the delivery. Thus, contrary to the retail segment where the end customer is approached through the retailer, in the OOH segment a significant sub-set of final consumers is target directly by the dairy manufacturers.

1265. In addition, the market investigation confirmed that packaging sizes and services differ compared to the retail segment. Clients in OOH demand daily deliveries including weekends and special packages, which are not available in the retail segment like small portions consumable immediately after purchasing or larger sizes to be used in the catering or hotel business. At least for these products, which represent up-to [20-30]*% of their turnover according to OOH wholesalers, alternatives and therefore a cross-over to the retail segment does not seem possible.

1266. The service requested from an OOH customer has direct implications for the wholesaler. Since OOH customers expect timely and flexible delivery from their wholesaler, they therefore look for a supplier able to mirror the customers' demand - which implies that unless products can be stored over longer periods (like long-life dairy or butter) such a need favours local suppliers.

1267. Moreover, order volumes in the OOH segment are significantly lower than in the retail segment - notifying parties estimate that 4% of the overall market is covered by the OOH segment in fresh dairy desserts. Orders by wholesalers are usually small and shipments over longer distances do rarely occur as transport costs per unit are too high. This fact limits sourcing possibilities from abroad.

1268. Contrary to the argument of the notifying parties, prices in the retail segment do not constrain the prices in the OOH segment since OOH clients are not willing to cross-over and buy at a retailer because the additional services will not be offered by the retailer. Demand seems therefore to be inelastic as customers have no alternatives.

1269. Several competitors as well as customers have argued that the notifying parties have the possibility to offer "package deals", thus entering the market with only one dairy product seems to be difficult as OOH wholesalers like to get as much as possible from one supplier for logistical reasons - achieving economies of scale in deliveries.

1270. Competitors already present in the Netherlands in the retail segment indicated that additional investments into a distribution network and logistics would be necessary to supply the OOH segment.

1271. On the basis of the replies, there are indications that the markets for fresh custard and porridge could be separated according to distribution channels in OOH and retail. In any event, given the strong presence of the notifying parties on a combined market of the supply of retail and OOH and the alleged small size of the OOH segment of roughly 4% in value, the question whether the OOH segment should be considered as a separate market for the supply of fresh custard and porridge can be left open as even on the broader market, the view is taken that the proposed transaction would significantly impede effective competition.

12.1.4. Conclusion on relevant product market

1272. On the basis of these elements, separate relevant product markets for fresh custard, porridge and portion pack desserts are defined. For custard, a further separation into private label/branded products is not necessary. Whether the market has to be further separated according to distribution channels can be left open as it would not affect the competitive assessment.

12.2. Relevant Geographic Market

12.2.1. The relevant geographic market for custard and porridge is national in scope

12.2.1.1. Relevant geographic market proposed by the notifying parties

1273. The notifying parties submit (776) that the relevant geographic market for fresh custard and porridge at the downstream level is national.

1274. At the upstream level the market for custard is wider than the Netherlands and includes Germany and Belgium. This is due to several factors including (i) advances in logistics due to centralised distribution by retailers and extended shelf life allowing for transportation over longer distances, (ii) procurement done on an international scale and (iii) the possibility of foreign suppliers to switch their fresh milk, fresh buttermilk or plain yoghurt production to the production of custard in gable tops for the Dutch market. All these elements prevent suppliers in the Netherlands from increasing prices profitably by 5-10% as they would lose significant volumes to foreign competitors.

1275. For porridge the notifying parties consider that the market may be confined to the Netherlands. (777)

12.2.1.2. The assessment of the Commission

1276. The market investigation in the present case has yielded strong indications that the relevant geographic market for fresh custard is narrower than submitted by the notifying parties.

1277. First, demand characteristics across Belgium, Germany and the Netherlands are strikingly different. No custard is consumed outside the Netherlands. The notifying parties themselves submitted that "vla in gable top is a typical Dutch product and is sold in the Netherlands only". (778) According to the replies in the market investigation there are hardly any dairy companies currently producing custard outside and supplying the Netherlands. (779)

1278. In addition, competitors were asked in the second phase market investigation if they could produce custard in 1 litre gable top packaging. The replies were predominantly negative: Most foreign competitors informed the Commission that they would not only be unable to produce custard, but also not interested in this product: One competitor said "we are not able to produce this pack and we will not invest in such categories - it isn't strategic." (780) Another one replied "[...] do not have filling equipment suitable for Vla. Vla does not fit into our concept." (781)

1279. With respect to the procurement, competitors and customers have not confirmed the view of the notifying parties that tenders for custard are usually applied covering an area that is wider than national. (782) Therefore, the role of imports seems to be limited and is focused on the harddiscounters only. This has widely been confirmed by Dutch retailers and is mirrored in the supply data the Commission has received during the market investigation. The only non-Dutch supply in 2007 went to the discounters Aldi and Lidl, representing roughly 1% of total supplies. (783)

1280. Concerning porridge, the market investigation confirmed that the market should be viewed as national. Several retailers indicated that they exclusively rely on Friesland Foods and/or Campina when sourcing porridge. (784) None of the competitors which replied to the market investigation is currently producing any porridge. This has also been indirectly admitted by the notifying parties when they stated that "they are not aware whether any other supplier of porridge is active in the Netherlands". (785)

1281. In view of these elements, it is concluded that the relevant geographic market is national for the markets of custard and porridge.

12.3. Competitive Assessment

12.3.1. Market Structure and Market Shares

1282. According to the notifying parties, the downstream market for fresh dairy desserts covering the Netherlands had a total value of EUR 196 086 000 in 2007 (EUR 188 060 000 retail segment, EUR 8 025 000 in OOH), 26.5% of it coming from private label products and the remaining 73.5% by branded fresh dairy dessert. (786) Both private label products and branded products have lost in value over time.

1283. Campina's brand portfolio achieved a market share of [40-50]*% ([20-30]*% Campina brand, Mona [10-20]*%, Optimel [0-5]*%), followed by Friesland Foods' Friesche Vlag [10-20]*%. The leading retailer in the Netherlands Albert Heijn was able to achieve a share of [5-10]*% with its private label brand AH.

1284. Looking at the markets which are considered to be affected by the proposed transaction, in fresh custard the Campina brand portfolio had a share of [40-50]*% in 2007, followed by Friesland Foods with [20-30]*%, private label products achieved a share of [30-40]*% - for details see table below. (787)

1285. In the OOH segment for custard, the notifying parties had a combined market share in 2007 of [90-100]*% (Campina [80-90]*%, Friesland Foods [10-20]*%). (788)

Table 12-3: Downstream market shares custard in gable top, retail, The Netherlands - Source: Form CO.

1286. The Commission has relied on scanner data provided by the notifying parties to cross-check the figures submitted in the Form CO. As shown in the table 12-4 (based on the IRI data, excluding discounters), Campina brands obtained a market share of [40-50]*% and Friesland Foods' brand [20-30]*%.

Table 12-4: Market shares based on value, custard, retail, The Netherlands - Source:IRI.

1287. The discrepancies between the market shares in value reported by the notifying parties and those computed on the basis of IRI, are not of an order of magnitude that would significantly alter the competitive assessment. The difference is probably due to the fact that the scanner data do not include the sales of Aldi, Lidl and Koopconsult, sellers of private label products, and are based on the prices retailers charge vis-à-vis their customers, while the notifying parties have based their estimates on the prices they charge from retailers. Nonetheless, for the descriptive analysis of the downstream market, in particular prices, as well as the estimation of demand elasticities, it is necessary to rely on the IRI data

1288. For porridge - independent of the distribution channel - the notifying parties together account for 100% of the market at the downstream level, with Campina ([50-60]*%) accounting for slightly more than Friesland Foods ([40-50]*%).

1289. Looking at the upstream market, where retailers/OOH wholesalers source custard and porridge from dairy producers, the situation looks as follows according to the notifying parties. In the retail segment for fresh custard in gable top packaging, the notifying parties had a combined market share of [80-90]*% - representing an increase of [0-5]* percentage points since 2005 - because of their strong position in the branded segment (Campina [60-70]*%, Friesland Foods [30-40]*%) as well as in the private label segment. Here, Campina had a [40-50]*% market share in 2007, Friesland Foods [20-30]*%, with Inex [5-10]*%, and the German suppliers Humana accounting for [5-10]*% and Molkerei Ammerland [5-10]*%. The notifying parties have significantly increased their market share since 2005 (by [10-20]*% percentage-points) and the replies from the market investigation (789) seem to indicate that the share of the German suppliers in particular has been overestimated.

Table 12-5: Market shares upstream, custard in gable top, retail, The Netherlands - Source: Form CO.

1290. Since there is no private label available in the OOH segment, the market shares at the upstream level are identical to those for the downstream one. In addition, since the OOH market does only represent roughly 4% of the overall retail plus OOH market, the market shares on a combined OOH/retail sourcing market would not be different from the one for the retail segment.

1291. In porridge, due to the absence of private labels, the upstream market share figures resemble those of the downstream market with the notifying parties combined share accounting for [90- 100]*% of the market [Campina ([50-60]*%) slightly more than Friesland Foods ([40-50]*%)]*.

12.3.2. Non-coordinated effects in the markets for fresh custard and porridge

1292. The concerns expressed by market players and the arguments put forward by the notifying parties have been assessed and the conclusion is that the proposed transaction would significantly impede effective competition on the Dutch market for the sale of fresh custard in gable top packaging to retailers and OOH wholesalers as well as on the Dutch market for the sale of porridge to retailers and OOH wholesalers - that is, in important Community markets - for the reasons set forth in this section.

12.3.2.1. Fresh Custard in gable top packaging

1293. During the market investigations several complaints were received, particularly from customers who consider that the merger will ultimately lead to higher prices because the two leading suppliers of brands and private label custard would merge and no alternative supplier would be available. Given the strong position of the notifying parties in both segments (private label and branded) they would have an incentive to increase prices for brands and private label products at the same time.

12.3.2.1.1. Notifying Parties have large market shares

1294. The larger the market share, the more likely a firm is to possess market power. Moreover, the larger the addition of market share, the more likely it is that the merger will lead to a significant increase in market power. The proposed transaction would combine the current market leader, Campina, with the second strongest player in the market, Friesland Foods, with a combined market share of [90-100]*% with an addition of [30-40]*% in the branded retail segment, [60- 70]*% with an addition of [20-30]*% in the private label retail segment, [80-90]*% with an addition of [30-40]*% in the overall retail segment and [90-100]*% with an addition of [10- 20]*% in the OOH segment.

1295. The combined market share in an overall market with no separation into branded/private label and distribution channel would be similar to the one in the overall retail segment.

1296. Clearly, this significant increase in the sales base on which higher margins can be enjoyed after a price increase makes it more likely that the notifying parties will find such a price increase profitable despite the accompanying reduction of output.

12.3.2.1.2. Notifying parties are close competitors

1297. In addition to pointing towards high market shares, most complainants (790) argued that the notifying parties are the closest competitors. In particular, with respect to the upstream market complainants indicated that the notifying parties are the only ones that currently are able to offer the full range of fresh basic dairy products in sufficient volumes and quality. (791) Moreover, they are also the only firms able to supply both in the form of supplier brands and private label. Customers often source from both notifying parties and considered the rivalry between Friesland Foods and Campina as an essential source of competition. Thus, the merger would eliminate the primary source of competitive pressure that prevails in the custard market.

1298. Currently the notifying parties compete directly both in the supply of branded products and also in the supply of private labels to supermarkets. Furthermore private labels and supplier brands compete in the downstream markets as discussed in the market definition section. As a result Campina and Friesland Foods mutually exert competitive pressure on each other through three means: (a) First, supermarkets willing to carry a private label currently are able to source a large part of their requirements from either of the notifying parties, thereby enjoying a certain degree of countervailing buyer power. (b) Second, the supplier brands of the notifying parties are the closest competitors in the downstream market thereby allowing supermarket chains that wish to carry at least one brand of custard to play one against the other in listing and pricing negotiations. (c) Third, supermarket chains can also threaten to source private label from Campina in negotiating prices for Friesland Foods brands, and vice-versa. The significance of this competitive constraint depends on the degree of substitutability of private label and supplier brands among end-customers.

1299. By bringing together the main competitors in the supply of both brands and private labels the merger would undermine all three mechanisms of competitive interaction in the custard market and impede the exercise of any significant degree of countervailing buyer power. 1300. Table 12-3 shows that the notifying parties' two brands achieved a combined share of [90- 100]*% in 2007 in the downstream market (looking at the branded segment only). Furthermore, the notifying parties' brands are both positioned as premium brands with prices historically 30% higher than low-end private labels. However, this gap has narrowed in the past year as a result of the significant increase in raw milk prices which have affected private labels relatively more than supplier brands.

1301. The graph in Figure 12-3 shows the price evolution of the notifying parties' brands and that of the major private labels. Together the brands represented account for more than [90-100]*% of the market.

Figure 12-3: Evolution of Average Weighted Prices - Custard Source:IRI.

1302. Since custard is a rather homogeneous product, the observed price differences can be regarded as a proxy for the degree of product differentiation. Comparing price levels for the different private label brands and the supplier brands shows that Campina and Friesland Foods are the only two premium supplier brands currently available in the market with a significant volume.

1303. Major supermarket chains usually offer two private label products in addition to the premium brands of the notifying parties (with the exception of Albert Heijn). The graph further confirms that while some brands are positioned at the lower-end of the market others seem to be closer to the premium brands of Campina and Friesland Foods.

1304. The Commission attached a set of econometric results on estimation of retail demand systems to its Statement of Objections. These results complemented the other, larger set of qualitative evidence and descriptive statistics. The goal of the econometric model was to assess whether supplier brands exert a greater competitive constraint on each other than private labels. The notifying parties raised a number of criticisms regarding the interpretation and the robustness of the econometric results in their response to the Statement of Objections.(792)

1305. Most of the criticisms put forward by the notifying parties are invalid or based on a misunderstanding of the methodology. Moreover, a number of criticisms regarding the robustness of the results can be addressed by extending the econometric model. Such extension is also justified on economic grounds albeit for reasons other than those raised by the notifying parties. However, the results of the extended regressions do not support the initial finding that supplier brands constrain each other more than private labels. (793)

1306. It is important to emphasise that this conclusion does not in any way contradict or affect the merits of the other, qualitative and quantitative evidence put forward in each of the affected markets.

1307. In conclusion, the brands of the notifying parties mutually constrain each other. The competitive constrain exerted by private labels on supplier brands is comparable. This provides a strong indication that the merged entity would be in a position to increase the price of fresh custard.

1308. Indeed, it is self evident that since the merged entity will supply both branded products and private labels to supermarket chains, when considering an increase in the wholesale price for the Campina and Friesland Foods brands, the merged entity will undoubtedly take into consideration potential switching of customers to private labels, allowing the notifying parties to increase the price of private label supplies to supermarket chains. Of course the converse is also true. 12.3.2.1.3. Customers have limited possibilities of switching suppliers

1309. Customers have not only informed the Commission that they would view the notifying parties as closest substitutes, but also that they would have difficulties in switching to other suppliers. This is valid for the retail as well as the OOH segments where retailers/OOH wholesalers usually source from both notifying parties at the same time. Since there are only a few alternative suppliers (such as Farm Dairy, Den Elder, Zuivelhoeve and Molkerei Ammerland (Germany)) currently active on the Dutch market, most of them as suppliers of private label, the retailers and OOH wholesalers would have no alternative in the branded segment other than the notifying parties. This is of particular importance for the OOH segment, which currently exclusively relies on branded products supplied by the notifying parties only, with no alternative producer available at all.

1310. Asked for past examples of suppliers switching from either Campina or Friesland Foods to other suppliers of fresh dairy desserts, the notifying parties replied (794) that "Campina has no examples of retailers switching from Campina to different suppliers" and "there have been hardly any switches away from Friesland Foods' products in the last year".

1311. While it is correct that private label products play an important role in the downstream market and compete against the brands of the notifying parties, switching from branded products to private label products in case of a price increase for branded custard would be limited because all retailers still rely to a significant part on branded products ([60-70]*% of total supply in 2007), but in addition, Campina and Friesland Foods are also the two largest producers of private label custard in the Netherlands with a combined market share of [60-70]*%, which represents and increase of [10-20]*% percentage points since 2005. Thus, the private label segment cannot be considered to be a competitive constraint on the branded segment since the notifying parties would be the dominant player in both segments post merger.

1312. In addition, the mentioned suppliers lack the size of the notifying parties. Retailers have therefore argued that "due to the relative volumes required by [...], it will be impossible to switch to smaller suppliers." (795) and that "finding a producer of gable top vla is very difficult" (796) and a third one stated that"Campina en Friesland Foods beleveren [...] met Vla de verpakking is uitsluitend gable top. Er is ons geen leverancier buiten Nederland bekend welke de benodigde kwaliteit in gable top kan leveren." (797)

1313. Even if enough smaller suppliers were available in the market, one contract pre-merger handled by either Friesland Foods or Campina would have to be split up in parts after the merger, lowering the benefits for the retailer coming from economies of scale in production and logistics, thereby increasing the costs and finally the price to customers. (798)

1314. Switching to foreign suppliers has not been perceived by customers as a credible alternative because there is only one current supplier of custard in Germany and Belgium, other competitors denied having the ability to produce fresh custard in gable top packaging.

1315. In conclusion, switching possibilities on all potential markets for fresh custard in gable top would be limited for customers after the merger.

12.3.2.1.4. Lack of countervailing buyer power

1316. The notifying parties also argue that countervailing buyer power from the largest retail chains (for example Albert Heijn, SuperUnie, Schuitema, or Super de Boer) would restrain them from raising prices, because of the retailers' strong demand for private label products relative to brands. "In this respect", the notifying parties note (799), "any post-merger price increase sought by the Parties would be retaliated by retailers who will be able to rapidly expand the share of PL products or delist brands." 1317. As the general structure of the market and the competitive interaction for custard is similar to that of fresh milk, the arguments put forward in section 7.1.3.1.7.1 concerning buyer power also apply in relation to custard.

1318. Since the notifying parties are perceived as the closest competitors and customers would have only limited switching possibilities post-merger, an immediate switch to alternative sources of supply can be excluded. In addition, the notifying parties are the dominant producers of the main retailers' private label products with a combined market share in the Netherlands of [60-70]*%, thus threatening to switch to private label products would be a limited option for retailers as supply alternatives in that segment are limited as well.

1319. Furthermore, it is not sufficient that buyer power exists prior to the merger - it must also exist and remain effective following the merger. This is because a merger of two suppliers may reduce buyer power if it thereby removes a credible alternative. In the present case, the merger will remove one of the two main suppliers considered to be the closest competitors in the market, since alternative suppliers with a similar size could not be identified in the course of the market investigation.

1320. Therefore, it is concluded that countervailing buyer power post-merger would not be sufficient to off-set potential adverse effects of the merger.

12.3.2.1.5. Entry unlikely to occur

1321. Despite the high market shares, the notifying parties argue that in case of a price increase other competitors would enter the market constraining the notifying parties.

1322. The market investigation has not confirmed that competitors currently not active in the production of fresh custard in gable top packaging have any intention to enter the Dutch market. Almost all competitors that have replied to the market investigation stated clearly that they would have no intention to enter the Dutch market with custard in gable top packaging. One supplier for example explained (800) that "they have never considered to enter the Dutch market due to the fact that [...] brand is unknown and it will be too expensive to build a franchise on this market". Two German producers have already been quoted in the section on the relevant geographic market (see recital 1278). A third one replied (801) that "[he] as a matter of principle is not active in this product segment ".

1323. In addition, retailers argued that transport costs would be higher if such a product were sourced from abroad. The longer supply distance would increase costs for the retailer (and their customer) and put the foreign suppliers at a disadvantage concerning freshness because Dutch suppliers would always be able to supply fresher products as they are located closer to the distribution centres of the retailers. Since Dutch customers consider freshness a key variable in their purchasing decision, Dutch incumbents have a clear competitive advantage.

1324. Finally, the market for fresh custard in gable top packing seems to be a declining market - it has declined in value from EUR 118 million in 2005 to EUR 108 million in 2007 - making entry less attractive. Moreover, the notifying parties have spare capacity of around [30-40]*% (802) which would allow them to deter entry by threatening to use that capacity to depress prices.

12.3.2.1.6. Conclusion on the competitive assessment

1325. For the reasons set out above, it is concluded that the notified concentration is likely to significantly impede effective competition on the market for fresh custard in the Netherlands, regardless of whether these markets need to be further segmented according to distribution channels.

12.3.2.2. Porridge

1326. On a porridge market covering the Netherlands the merger would lead to a monopoly with Campina holding a [50-60]*% market share and Friesland Foods a [40-50]*% share - in the retail and OOH market.

1327. The market investigation revealed that the notifying parties would be the only suppliers of porridge in the Netherlands and no alternative would be available. Thus customers, in particular retailers, expect that due to the merger prices would increase.

1328. The notifying parties - although not aware whether any other supplier of porridge is active in the Netherlands - have argued (803) that "there are several small producers who are already able to manufacture small quantities of porridge (mainly small organic farms) and/or who would be in a position to expand their operations".

1329. This claim - not supported with specific company names - has not been supported during the market investigation. (804) For example one customer informed (805) the Commission that "there are no alternatives for porridge as [...] is fully dependent for all porridge on Campina and Friesland Foods."

12.3.2.2.1. Conclusion on the competitive assessment

1330. Given the lack of alternative suppliers to the notifying parties it is concluded that the notified concentration is likely to significantly impede effective competition on the market for porridge in the Netherlands, regardless of whether these markets need to be further segmented according to distribution channels.

12.3.2.3. Overall conclusion on the competitive assessment

1331. For these reasons, it is concluded that the notified concentration is likely to significantly impede effective competition on (i) the market for fresh custard in the Netherlands and (ii) the market for porridge in the Netherlands, which is a substantial part of the common market, regardless of whether these markets need to be further segmented according to distribution channel.

13. CREAM

13.1. Introduction

1332. Cream is a mix (emulsion) of fat and water, with varying fat percentages. The fat may be dairy or non-dairy or a blend thereof. Dairy cream is a by-product of dairy production. Non-dairy cream is made of vegetable fat (such as coconut or sunflower fat). Dairy cream is in essence a way to add value to excess milk fat.

1333. According to the notifying parties within cream a distinction should be made between spray cream and liquid cream, which should be treated as separate relevant product markets.

1334. The notifying parties submit that spray cream (which is cream combined with pressurised gas, packed in a can) and liquid cream have to be treated separately for the reasons set forth below.

1335. After the production of cream as an ingredient (dairy- and non-dairy based), the filling step is different for liquid cream and spray cream. For liquid cream, filling can be performed in packages, being bottles, bag-in-boxes or tetra bricks and in sizes varying from 125 ml up to 10 litres. In the industry segment, returnable packaging is used in the form of containers, tanks and tank wagons and as such delivered in bulk to the customer. For spray cream, the resulting cream is often combined with vanilla flavour, sugar or others sweeteners, poured into a sterilised can, which is then closed with a sterilised valve and gassed. It is packaged in aluminium and tin cans.

1336. Spray cream and liquid cream also differ in shelf life. Shelf life of spray cream ranges generally from 6 to 9 months, although small quantities of spray cream with a shorter shelf life are also sold. Spray cream can generally be stored in an ambient environment.

1337. Liquid cream is produced in pasteurised and UHT-treated form. Packaged pasteurised liquid cream has a shelf life of generally four weeks (low pasteurised up to 16 days, high pasteurised four to six weeks). UHT-treated liquid cream has a shelf life of several months. Pasteurised cream has to be refrigerated (and is therefore "fresh"). UHT-treated cream can be stored at an ambient temperature (and is therefore "long-life"). The notifying parties argue that, in practice, most UHT-treated cream is sold refrigerated and therefore perceived as fresh.

1338. Based on the remarks made by respondents to the market investigation, (806) the basic distinction between liquid cream and spray cream in the relevant product market is accepted.

13.2. Liquid Cream

1339. Both notifying parties are active in the production and sale of liquid cream to retailers, OOH wholesalers and the industry.

1340. In the retail segment Campina sells most of its liquid cream under the Campina brand and under private labels. Friesland Foods sells liquid cream under the Friesche Vlag brand and under private labels. In this sales channel, liquid cream is mostly sold in units of 125 ml to 250 ml.

1341. In the OOH segment, Campina sells liquid cream under the Polderland and Campina brands. Campina sells liquid cream to the OOH customers in bottles and cans of 800 ml to 5kg and bagin- boxes (BIB) of 5 to 10 Kg with a shelf life of approximately four weeks, with and without added sugar. Friesland Foods sells liquid cream under the Debic and Hollandia brands. It sells its products in bottles of 1 and 2 Kg, bags of 1.75 Kg, bag-in-box of 5 or 10 Kg and bottles of 5 Kg and cartons of 5 Kg.

1342. In the industrial segment, supplies of liquid cream are delivered directly to customers and therefore not under a particular brand. Campina sells liquid cream to the industrial segment in tanks and containers from 275 litres to 1000 litres. Friesland Foods sells liquid cream in shipments of 200 Kg, 500 litres and 650 Kg, in containers and in tanks ranging from 500 to 1000 Kg and more. Shelf life is generally 7 days.

13.2.1. Relevant Product Markets

13.2.1.1. Dairy liquid cream and non-dairy liquid cream each form a single product Markets

13.2.1.1.1. Product market definition proposed by the notifying parties

1343. The notifying parties argue that, although taste and colour of cream may differ slightly, depending on whether it is dairy based or non-dairy based, for most purposes, dairy and nondairy based cream are substitutable.

1344. With respect to supply-side substitutability, the notifying parties submit that the first steps of the production process of non-dairy liquid cream are slightly different from those for dairy liquid cream and that, therefore, the investment required to transform a dairy cream manufacturing plant to a manufacturing plant suitable for the manufacturing of non-dairy cream is negligible because the main components of the line are already present (costs are approximately EUR 0.5 million). In particular, with respect to supply-side substitutability, plants that are suitable for producing dairy cream can also, but not at the same time, be used for producing non-dairy cream. To be able to produce non-dairy cream, warm fat needs to be inserted in the plant. This requires the use of a heater on-site or the use of heated transporting (trucks). Currently, Friesland Foods uses such heated trucks to have non-dairy fats delivered for the production of cream. Installing a heater takes about two months. Ordering non-dairy fat in heated trucks takes a few days. If a manufacturer of dairy cream wants to switch production, the plant must be rinsed before starting a non-dairy production run. This rinsing can be done automatically and takes between one and three hours. Costs for rinsing are negligible.

1345. The notifying parties regularly switch from producing dairy to non-dairy cream and vice versa, when this is commercially interesting. Friesland Foods uses both its Lummen (Belgium) and Nuenen (Netherlands) plants as dual sourcing plants (the same production lines for dairy and non-dairy), Campina uses its plant is Oud-Gastel as a dual sourcing plant (the same production lines for dairy and non-dairy). Competitors, according to the notifying parties, are also able to switch from producing dairy to non-dairy cream.

1346. However, the notifying parties did not provide details as to the time and costs necessary to switch from the production of non-dairy liquid cream to that of dairy liquid cream.

1347. In conclusion, the notifying parties submit that dairy and non-dairy liquid cream belong to the same product market.

13.2.1.1.2. Assessment of the Commission

Demand-side substitution

1348. It has been verified, during the market investigation, if customers are able and willing to substitute non-dairy liquid cream to dairy liquid cream. The results of the investigation is that customers consider non-dairy and dairy liquid cream as different products and are not able/willing to switch from one to the other in case of price increase. In particular, customers have noted a quality gap between the two products and a significant difference in taste (taste of non-dairy products is considered poor). (807) Retail customers have noted that non-dairy cream may be a substitute for dairy cream for some uses (for example, cooking), but not for others (whipping) and have generally submitted that their customers would not switch between dairy and non-dairy products. (808)

Supply-side substitution

1349. With respect to supply-side substitutability, the notifying parties' competitors have consistently denied such a possibility. Switching between the production of dairy and non-dairy liquid cream requires the development of a new recipe. In addition, the technology involved is significantly different (vegetable fats require specific heating equipment) and a long and investment-intensive effort would be necessary. (809) In particular, the market investigation has not confirmed the notifying parties' submission that the switch from dairy to non-dairy cream may occur in a limited timeframe and with a small financial investment.

1350. In addition, although there are isolated examples of liquid cream producers manufacturing both dairy and non-dairy cream or which are able to switch between the to productions, competitors in general submit that they are not willing or able to switch from dairy cream to nondairy cream, even if the latter's price increased by 5-10%. Some of them consider such a switch a complete "u turn" of their market strategy, something against their business philosophy, which would, in any event, require between 1 and 2 years (810) and another respondent (811) indicated as "important" the necessary investment.

1351. For the reasons set out in recitals 1348, 1349 and 1350, namely the lack of both demand-side and supply-side substitutability, it is concluded that non-dairy and dairy liquid cream belong to separate markets.

1352. Since there is no overlap in non-dairy liquid cream, the assessment focuses only on the market for dairy liquid cream. 13.2.1.2. Low-fat and high-fat cream belong to the same relevant product market

13.2.1.2.1. Product market definition proposed by the notifying parties

1353. The notifying parties submit that low-fat liquid cream for use in cooking is substitutable by high-fat liquid cream, as the main difference is the fat percentage. However, low-fat liquid cream is not a valid substitute for high-fat liquid cream to be used for whipping, since the lower fat content does not provide for a sufficiently stable whipped cream.

1354. From a supply-side perspective, the notifying parties explained that liquid cream with various fat percentages can be produced on the same production line (being sufficient to merely dilute the cream with milk, in order to adjust the fat level), so there is perfect supply-side substitutability in this respect.

1355. The notifying parties therefore submit that low-fat and high-fat liquid cream should be considered as being part of the same product market. 13.2.1.2.2. Assessment of the Commission

1356. The market investigation has confirmed that a number of customers consider low-fat and high-fat liquid cream as substitutes, especially for cooking purposes, and are able and willing to switch between the two. (812)

1357. Similarly, on the supply-side, the market investigation has indicated that there is complete supply-side substitutability, as most liquid cream producers already manufacture both low-fat and high-fat cream and those which don't are able to swiftly switch production, in the case of a price increase. (813)

1358. Therefore, it is concluded that low-fat and high-fat liquid cream are part of the same product market.

13.2.1.3. It can be left open whether the market for dairy liquid cream have to be further separated in fresh and long-life dairy liquid cream

13.2.1.3.1. Product market definition proposed by the notifying parties

1359. With respect to the distinction between fresh (or pasteurized) and long-life (or UHT treated) liquid cream the notifying parties argued that, with existing technologies, the differences in taste between the two products are minimal. Considering also that UHT cream is often sold refrigerated, the notifying parties argue that customer do not perceive a difference between fresh and long-life and that thus this distinction is not relevant.

1360. During the procedure, the notifying parties have qualified their statement on consumers' attitude towards fresh and long-life cream. In particular, they have explained that, due to historical reasons, Dutch customers have a strong preference for fresh liquid cream, while Belgian and German customers remain in principle indifferent to the distinction between the two varieties of liquid cream. (814)

1361. The notifying parties however maintain that there is supply side substitutability between fresh and long-life liquid cream. Indeed, UHT equipment essentially consists of a heating device which is "attached" to the line for producing fresh pasteurized cream. Therefore, a line equipped to produce UHT cream would also be capable to produce fresh cream, once the UHT device is switched off. The notifying parties note that most recent cream production plants have both technologies available. The notifying parties then argue that fresh and long-life liquid cream belong to the same product market.

13.2.1.3.2. Assessment of the Commission

Demand-side substitution

1362. The replies of customers with respect to substitutability between fresh and long-life cream are mixed. Industrial customers do not consider them as substitutes, since they are used to buy mainly fresh cream. (815) Short shelf life is not a constraint for them, since they buy fresh liquid cream with delivery "just in time". Retailers have also provided mixed replies. It seems that fresh liquid cream and long-life liquid cream are not fully substitutable to each other, although some customers would be willing to switch (816) because of the similar characteristics and limited substitutability of the two products for cooking purposes. (817) In the OOH segment, the replies are mixed. Two OOH wholesalers deny that fresh and long-life liquid cream would be interchangeable. On the other hand, one respondent explains that only the professional, highly sophisticated purchasers like professional cooks and bakers would not switch, while other OOH final customers would do it. (818)

Supply-side substitution

1363. The replies of respondents to the questions on supply-side substitutability indicated that the switching between the production of fresh cream and long-life cream is asymmetric. Producers of long-life cream can switch to producing fresh liquid cream, (819) but the reverse is not possible, as it would require setting up an UHT equipment. Some respondents to the investigation noted that such upgrade would be expensive (EUR 500 000) and time-consuming (2 years) and would only be profitable in case the price for long-life cream increased by more than 10%. (820)

1364. In light of the inconclusive replies, the question whether fresh cream and long life cream belong to the same relevant product market is finally left open as such a distinction would not have an impact on the competitive assessment.

13.2.1.4. It can be left open whether private label and supplier brands belong to the same relevant upstream market

13.2.1.4.1. Product market definition proposed by the notifying parties

1365. With respect to the distinction between branded products and private labels, the notifying parties acknowledge that, in some cases, it was argued that the procurement of private label products and the procurement of branded products by retailers (and OOH wholesalers) would determine two neighbouring, but separate markets, because the competitive conditions under which retailers source private label products and branded products would be fundamentally different.

1366. In that respect, the notifying parties do not deny that purchasing conditions may be different where specific branded products are perceived as "must carry" products by retailers. This, however, is not the case in the retail markets for liquid cream, (i) which in essence is a commodity product used as an ingredient in households, (ii) where there is no intrinsic difference between branded and private label products, and (iii) where the main differentiating factor is price, not other characteristics. In the OOH segment, the notifying parties expect private label to become more important in the future as well.

1367. In conclusion, the notifying parties argue that branded and private label liquid cream should be part of the same market, as there would be no room for suppliers of liquid cream to carry through a sustainable price increase of 5 to 10% for their branded products.

13.2.1.4.2. Assessment of the Commission

1368. In the market for fresh liquid cream in Belgium, Germany and the Netherlands, private labels represent more than 50% of the total market (EUR 205 million on a total market of EUR 375 million). In the OOH market, private labels are absent, as the whole market consists of branded products. (821) In the retail segment, the same suppliers produce branded and private label cream for the most part, while this is of course not the case in the OOH segment, where private labels are not present.

1369. In long-life liquid cream in Belgium, Germany and the Netherlands, private labels represent more than 50% in the retail market (EUR 72 million on a total market of EUR 130 million). In the OOH market, private labels account for approximately 4.5% of the total market (EUR 9 million on a total market of EUR 204 million). In both the retail and OOH segments, the same players produce branded and private label cream for the most part.

1370. The market investigation revealed that the procurement pattern between private label and branded products is different. Some respondents indicated that branded products are sourced through bilateral negotiations, which aim at setting not only the basic price for the product, but also the marketing and promotion mark-ups, the bonuses and the volume rebates. Private labels, instead, are often tendered and sold at a net/net price. (822)

1371. However, whether private labels and supplier brands belong to the same product market upstream also depends on other factors. In particular: (a) Whether both types of brands, in general, compete closely with each other from the perspective of the end-customer; and (b) The extent to which upstream suppliers of private label and/or brands as well as the purchasing retailers, take into account in their negotiations upstream the competitive pressure that private labels and supplier brands mutually exert on each other at consumer level.

1372. With respect to downstream competition between branded and private labels liquid cream, customers and competitors consistently indicated that the two categories compete on retail downstream markets, while they are virtually absent in the OOH market. In this respect, the main ground for competition is price, with a large price difference between branded and private label products of up to 30-50%. (823)

1373. As to the second issue - whether upstream negotiations for liquid cream take into account the competitive interaction between the two categories downstream - respondents to the market investigation indicated that the price of private label is determined on the basis of other retailers' behaviour, that the price of branded products is not constrained by the price of private labels and that the shelf space allocation between the two brands has remained stable. (824)

1374. Although there are indications that at the sourcing level, where dairy manufacturers sell cream to retailers, a distinction between private label and branded products could be made, the question is finally be left open, as it would not have any impact on the competitive assessment.

13.2.1.5. The retail and OOH sales channel belong to separate product markets

13.2.1.5.1. Product market definition proposed by the notifying parties

1375. With respect to the liquid cream market, the notifying parties submit that three types of sales channels can be distinguished: (i) the retail sales channel, (ii) the OOH sales channel and (iii) the industry sales channel. Differences between these sales channels mainly concern the brands on the product sold, the packaging and the sales practices. In particular, suppliers use specialized sales force in the OOH channel, considering that purchasers (cooks and bakers) are sophisticate buyers. The notifying parties, therefore, submit that the retail and the OOH sales channel identify separate product markets, although, for reasons further explained below, the notifying parties believe that the retail channel puts competitive pressure on the OOH channel.

1376. The industry segment, according to the notifying parties, would form a separate segment within the liquid cream market due to the absence of cross sales between the channels. The notifying parties explained that industry is supplied in bulk (tanks of 200 litres and more), which are not substitutable with packages of up to 10 litres.

13.2.1.5.2. Assessment of the Commission

1377. The results of the market investigation have confirmed several elements put forward by the notifying parties. In the first place, packaging is different between retail and OOH, as retailers mainly source 250 ml packages, while OOH rely mostly on 10 litre BIB (Bag-in-Boxes). In addition, prices (more volatile in OOH) and brands are different between channels. Furthermore, in the OOH channel there are specific procurement processes (based on individual negotiations in the retail channel and on best net price selection in OOH and industry channels). (825)

1378. Therefore, it is concluded that, with regard to liquid cream, a distinction should be made according to distribution channels in retail, OOH and industry.

13.2.1.6. Conclusion on relevant product market

1379. In conclusion the product markets for liquid cream are defined as follows: (i) Dairy liquid cream sold to retail customers; (ii) Dairy liquid cream sold to OOH customers; (iii) Dairy liquid cream sold to industrial customers; (iv) Non-dairy liquid cream sold to retail customers; (v) Non-dairy liquid cream sold to OOH customers; (vi) Non-dairy liquid cream sold to industrial customers;

1380. It is also concluded that the market for dairy liquid cream includes both low fat and high fat liquid cream, as there is a high degree of demand-side substitutability and supply-side substitutability.

1381. Within the market for dairy liquid cream, the distinction between fresh cream and long-life cream can be left open, as it will not have an impact on the competitive assessment.

1382. Similarly, the question as to whether it is necessary to differentiate between branded and private label liquid cream is left open as it does not have an impact on the final conclusion on the effects of the transaction.

1383. As indicated in recital 1352, the notifying parties' activities do not overlap in the market for non-dairy liquid cream. The assessment, therefore, will focus only on dairy liquid cream.

13.2.2. Relevant Geographic Market

13.2.2.1. Relevant geographic market proposed by the notifying parties

1384. On the upstream market for the supply of liquid cream - independent of the distribution channel - the notifying parties submit that the relevant geographic market is wider than national and includes at least Belgium, Germany and the Netherlands. In the notifying parties' opinion, it may also include Denmark, France and Italy. In order to support their conclusion, the notifying parties refer to Eurostat figures on the import and export of large packages of liquid cream (two litres and more), showing that significant cross-border flows of cream take place in the Community. Cream is imported into the Netherlands from Belgium, Denmark, France and Germany, and cream from the Netherlands is exported into Belgium and Germany.

1385. In addition, the notifying parties make reference to the fact that some of their customers have announced their intention to source liquid cream from other Member States. The notifying parties also argue that they themselves and their competitors are targeting other national markets and face competition from foreign producers.

1386. The notifying parties finally note that domestic producers would not be in the position to profitably increase prices without losing sales to foreign competitors. This would, inter alia, result from the fact that: (i) liquid cream is a commodity product and there are no national preferences; (ii) shelf life of liquid cream (minimum 4 weeks) is not an obstacle to shipping over long distances; (iii) retailers and large OOH customers avail themselves of distribution centers, which optimizes transport costs; (iv) there are no regulatory barriers to cross-border sales of liquid cream.

13.2.2.2. Assessment of the Commission

1387. In assessing the geographic scope of the relevant geographic market for liquid cream, multiple factors have been taken into consideration. In particular: (i) the sourcing patterns currently followed by market players, (ii) the possibility that customers source from neighbouring Member States, in case of a 5-10% price increase, (iii) the competitive landscape in the various Member States, in terms of presence of liquid cream suppliers, and (iv) hurdles to cross-border sales of liquid cream (if any) depending on transport costs and other logistic considerations.

1388. With respect to the current sourcing patterns, the market investigation revealed that retailers source their supply of liquid cream from both their own country and a broader area including the Netherlands, Belgium and Germany. (826) Dutch OOH customers mainly source cream from the Netherlands. Industrial customers, with some exception, mainly source liquid cream from their own Member State. (827) Although such findings apply to fresh liquid cream in particular, with respect to long-life cream and from a supply-side angle, producers of long-life dairy liquid cream generally indicated long shipping distances for their products and supply customers in distant Member States (Italy, Portugal and Spain). (828) Retail respondents have not indicated sourcing differences between branded and private label cream (private labels are not present in the OOH segment).

1389. As to the possibility of switching to suppliers located in neighbouring Member States in the case of a 5-10% price increase, retail customers have indicated that they would look for alternative sources of supply in neighbouring markets. (829) OOH customers have provided mixed answers. In this respect, the main obstacle to switching seems to be the lack of sugar in cream produced outside the Netherlands, a requirement suppliers may easily comply with. (830) Industrial customers which are not already doing it, declared themselves willing and able to source from across their national borders should the need arise. (831) Retail respondents have not indicated sourcing differences between branded and private label cream (private labels are not present in the OOH segment).

1390. As to the competitive landscape in the markets for dairy liquid cream in the various Member States, the situation is as follows:

(a) Both notifying parties are active in the Netherlands and Campina is also active in Germany.

(b) In the Dutch branded retail market, German producers are active both directly (Nordmilch in fresh and long-life cream, Milchunion Hocheifel in long-life cream) and through importers (Koningszuivel).

(c) In the Dutch private label retail market for fresh cream, three German producers - Nordmilch, Starmilch and Neuberger - have significant market shares, which have increased over the last three years (from [5-10]*% to [10-20]*% individually) and which place them second only to Campina.

(d) In the Dutch retail market for branded long-life cream, Milchunion Hocheifel is the market leader.

(e) In the Dutch retail market for private label long-life cream, two German producers and a Belgian supplier split the market equally between themselves. It must be observed, in relation to the foregoing, that the presence of German producers on the Dutch markets for liquid cream strengthens the conclusion that the market includes three Member States, regardless a distinction between fresh and long-life liquid cream, as such producers are equally able to supply fresh and long-life cream.

1391. As to transport costs and logistics the market investigation indicated that retail customers optimize their distribution system using distribution centres, which allow them to aggregate shipments and centralize supply, (832) with substantial savings on transport costs and an increase of the sourcing distance. In addition, retailers indicated that, in the case of a price increase, they are ready to adjust their sourcing solutions and source from neighbouring Member States. (833) OOH customers have not provided information as to their transport costs and logistic arrangements. Industrial customers have indicated that they can source over a distance ranging between 50 and 150 Km. (834)

1392. In light of these considerations, it is concluded that the OOH market, the retail and the industry markets for dairy liquid cream exceed the national boundaries and include at least Belgium, Germany and the Netherlands.

13.2.2.3. Conclusion on the relevant geographic market

1393. With respect to the definition of the relevant geographic market in liquid cream, therefore, it is concluded that the markets for dairy cream sold to OOH, to retail and to industrial customers exceed the national boundaries and include at least the Netherlands, Belgium and Germany.

13.2.3. Competitive assessment

1394. During the market investigation, a number of market players have voiced concerns in relation to the proposed merger, concerning possible unilateral effects stemming out of the enhancement of the notifying parties' market position after the transaction. Such concerns and the arguments put forward by the notifying parties have been assessed and the conclusion is that the proposed transaction would not significantly impede effective competition on the market for dairy liquid cream sold to OOH, retail and industrial customers in the geographic market including Belgium, Germany and the Netherlands for the reasons set forth below.

13.2.3.1. Market position of the notifying parties in dairy liquid cream

1395. In the retail market for dairy liquid cream (including fresh and long-life together), in the same geographic area, the notifying parties' estimated market share would be as follows:

Table 13-1: Market shares liquid cream, retail, Belgium, Germany and the Netherlands - Source: Notifying parties.

1396. Where fresh and long-life dairy liquid cream are taken separately, in the fresh retail market the notifying parties would have a combined market share of [10-20]*% in fresh cream (brands and PL together, slightly different if they are separated) and [0-5]*% (brands and PL together, slightly different if they are separated) in long-life cream.

1397. In the OOH market for dairy liquid cream (including fresh and long-life together), which extends to Belgium, Germany and the Netherlands, the notifying parties' estimated market share are as follows:

Table 13-2: Market shares liquid cream, OOH, Belgium, Germany and the Netherlands - Source: Notifying parties.

1398. If fresh and long-life dairy cream are taken separately, in the OOH segment the notifying parties would have a combined market share as follows:

Table 13-4: Market shares long-life liquid cream, OOH, Belgium, Germany and the Netherlands - Source: Notifying parties.

1399. In the industrial market for dairy liquid cream (where no distinction between branded and private label products exists), in Belgium, Germany and the Netherlands, the notifying parties' estimated market share would be as follows:

Table 13-5: Market shares liquid cream, industry, Belgium, Germany and the Netherlands - Source: Notifying parties.

1400. In the industry segment, the distinction between fresh and long-life cream does not materially affect the competitive positioning of the notifying parties on the market, as 97% of the purchases are purchases of fresh cream with "just in time" delivery (840).

1401. The market shares in Tables 13-1 to 13-5 indicate that the merger will prevent the notifying parties from exercising market power. Especially on the retail and OOH markets, there are a number of strong competitors with close or even higher market shares (mainly Nordmilch, but also Frischli in the OOH market for long-life liquid cream). Also, in the industry segment there are several other players present on the market.

1402. In addition, the analysis of market factors other than the market shares reveals that the transaction is not likely to result in a significant restriction of effective competition on such markets.

13.2.3.2. Ability of customers to switch to alternative suppliers

1403. In the industrial channel, several customers indicated the ability to switch between different suppliers. (841) In particular, these customers procure liquid cream through multiple sourcing. Although so far they have not switched supplier very often, they can, when necessary, procure liquid cream from alternative suppliers and from traders also in neighbouring geographic areas. (842)

Switching is easy also due to the fact that supply contracts are usually entered into (or renegotiated) for a short duration (a few months up to one year). (843) For example, one of the industrial customers indicated that he switched from Friesland Foods to Vreugdenhil, in order to obtain a better price for its supply of liquid cream. (844) Another industrial customer switched from Friesland Foods to Polderland, in order to obtain a better services, better price and marketing assistance. (845)

1404. Overall, it is considered that, should the notifying parties attempt to increase the price of liquid cream to industrial customers after the transaction, customers would have sufficient alternatives in order to procure the cream they need and make the price increase worthless.

13.2.3.3. Spare capacity available

1405. The market investigation has also revealed that, in general, there is a significant degree of spare capacity in the market for liquid cream. (846)

1406. With particular respect to the industrial segment, a number of suppliers which normally supply industrial customers have reported that they have spare capacity in the production of liquid cream. Although the data are fragmented, by aggregating the replies available, it is estimated that such spare capacity is in the range of 28-55% of these respondents' overall capacity. (847)

1407. On the basis of these considerations, it is concluded that the notifying parties would not find it profitable to reduce output and increase prices after the merger, as competitors would have the ability and the incentive to react by expanding production and restrain any attempted price increases by the notifying parties.

13.2.3.4. Conclusion

1408. For the reasons developed in section 13.2.3, it is concluded that the concentration is not likely to lead to a significant impediment of effective competition in the field of dairy liquid cream to retail customers, OOH customers and industrial customers in the market including the Netherlands, Belgium and Germany.

13.3. SPRAY CREAM

1409. Both notifying parties are active in the production of spray cream. In the retail segment, Campina sells its spray cream under the brands Campina, Cessibon, Banket, Menkomel, Tastou, Dorlay, Landgold and under private labels. Friesland Foods sells its spray cream under the Friesche Vlag brand and under private labels. Spray cream is mostly sold in 250 ml cans.

1410. In the OOH segment, Campina sells spray cream under the Polderland and Campina brands via the Polderland division of Campina. Polderland is the premium brand and Campina is the mainstream brand. Friesland Foods sells spray cream under its Debic and Hollandia brands. The Hollandia brand is primarily intended for the Dutch market, and the Debic brand is primarily intended for the Belgian market and the rest of the world (including Germany). Spray cream is sold by both notifying parties generally in 500 ml and 700 ml cans.

13.3.1. Relevant Product Market

13.3.1.1. Dairy and non-dairy spray cream each form a single product market

13.3.1.1.1. Product market proposed by the notifying

Parties

1411. In the Form CO, the notifying parties explained that spray cream, like liquid cream, is sold in dairy- and non-dairy types (dairy and non-dairy cream is often also sold as a blend). For most purposes, dairy and non-dairy based creams are substitutable according to the notifying parties. Non-dairy creams are commercialised by dairy producers because customers demand it as part of the same package. The notifying parties submit that the two belong to the same market.

13.3.1.1.2. Assessment of the Commission

1412. In their replies to the market investigation, retail customers have consistently indicated that their customer would not switch from dairy to non-dairy spray cream and, therefore, they do not consider the two products as substitutes. The reason for the lack of switching is the significant difference in taste and structure (or texture) between the two spray cream varieties. (848) The same replies have been given by OOH customers, which have noted differences in taste and quality. (849)

1413. With respect to supply-side substitutability, cream producers have been asked whether they are able to switch from the production of dairy and non-dairy cream and what the investment and the timeframe necessary were. They have consistently highlighted their inability to switch swiftly from the production of one product to the production of the other, although indicating different estimates of the necessary investments (EUR 500 000 - 2 million) and timeframe (6 months - 2 years). (850)

1414. In addition, one of the producers explained that non-dairy cream was initially conceived as a product to be sold on those markets (including the Community market) where import duties were imposed on food products of animal origin. As the Community enlargement was taking place and such duties were progressively abandoned, the market for non-dairy cream reduced mainly to non-European markets, where such duties are still in force. (851)

1415. In the light of the investigation, it is concluded that dairy and non-dairy spray cream belong to separate product markets. The notifying parties' respective activities do not overlap in non-dairy spray cream (only Campina is active in the retail market for non-dairy spray cream). Therefore, the non-dairy market is not further discussed below.

13.3.1.2. Private label products and branded products are in the same product market.

13.3.1.2.1. Product market proposed by the notifying

Parties

1416. The notifying parties acknowledge that in some cases it was argued that the procurement of PL products and the procurement of branded products by retailers (and OOH wholesalers) would constitute two neighbouring, but separate markets, because the competitive conditions under which retailers source PL products and branded products would be fundamentally different. The notifying parties do not deny that purchasing conditions may be different where specific branded products are perceived as "must carry" products by retailers. This, however, according to the notifying parties, is not the case in the retail market for spray cream (i) which in essence is a commodity product used as an ingredient in the kitchen, (ii) where there are no intrinsic differences between branded and PL products, and (iii) where the main differentiating factor is price, not other characteristics.

1417. In addition, the notifying parties consider that a differentiation between a sourcing market for branded and for PL products is artificial in markets where the retail segment is so heavily concentrated. The notifying parties noted that the Netherlands, suppliers of branded products have little bargaining power against giants like Albert Heijn (over 30% market share) and Superunie (with new member Dirk van den Broek over 35% of retail demand). The same dominance of the retailers is present in Belgium (Carrefour, Delhaize and Colruyt), in Germany (Rewe, Edeka, Metro, Aldi and Lidl) and in France (Carrefour, Leclerc, Auchan, Lidl and Aldi).

1418. With respect to the OOH segment, where private label products have a lower penetration rate, the notifying parties expect private labels to become important in the future as well. The notifying parties, therefore, submit that branded and private labels are, at least in the retail segment, in the same product market.

13.3.1.2.2. Assessment of the Commission

1419. In the market for spray cream in Belgium, Germany and the Netherlands, private label spray cream represents more than 50% of the total retail market (EUR 53 million on a total market of EUR 90 million) and approximately 20% of the total OOH market (EUR 9 million on a total market of EUR 42 million). In the retail market for spray cream, the same producers manufacture branded and private label cream, while in the OOH segment only the notifying parties, Codap and Incopack produce both branded and private label spray cream.

1420. Whether private labels and supplier brands belong to the same product market upstream depends on multiple factors. In particular:

(a) Whether both types of brands, in general, compete closely with each other from the perspective of the end-customer; and

(b) The extent to which upstream suppliers of private label and/or brands as well as the purchasing retailers, take the competitive pressure that private labels and supplier brands mutually exert on each other at consumer level into account in their negotiations upstream.

1421. The market investigation revealed that, in the retail segment, brands and private labels compete downstream. (852) Notwithstanding differences in procurement patterns and in price structure, private labels put a constraint on branded products. (853) In addition, some retail customers have launched their private label in what is called "a price war" in the Netherlands, which was started by retailers in 2003 and aimed at obtaining lower supply prices from producers of branded cream. (854)

1422. One of the notifying parties' competitors explained that branded products are declining and that the spray cream market is dominated by private label spray cream. Such development depends on the circumstance that, while food producers decided not to concentrate their marketing efforts in the cream market, retailers have made large investments in the creation and promotion of private labels. (855) In addition, looking at the upstream market, the same producers are active in the manufacturing of branded and private label spray cream - that is, the notifying parties, Codap, Incopack and Hochwald.

1423. Therefore, the conclusion is that, in the retail segment, branded and private label products are part of the same product market. In the OOH market, the question can be left open, as private label products represent a minimal portion of the total market (approximately 2% in volume).

13.3.1.3. The retail and OOH sales channel each form a separate product market

13.3.1.3.1. Product market proposed by the notifying

Parties

1424. In the Form CO, the notifying parties explained that spray cream is sold to retail customers and to OOH customers, while there are no industrial customers for this product. The notifying parties also submitted that the retail and OOH segment are separate markets. In particular, the notifying parties note that the brands, the packaging (250 ml cans for retail customers, 500-700 ml cans for OOH customers) and the sales arrangement between the two channels are different (specialized sales force is used to place products with OOH customers, as these are more sophisticate buyers).

13.3.1.3.2. Assessment of the Commission

1425. The results of the market investigation confirmed differences between the retail sales channel and the OOH sales channel. In particular, the market investigation indicated that there are differences concerning the brands, the packaging (formats sold to OOH customers are larger than formats sold to retail customers and also the packaging materials differ), the services provided to customers and also concerning wholesale prices (prices negotiated with retailers include bonuses and mark-ups, while prices negotiates with OOH customers are often net/net prices). (856) It is, therefore, concluded that such channels each belong to separate product markets.

13.3.1.4. Conclusion on the relevant product market

1426. In the light of these considerations, it is concluded that two relevant product markets exists: dairy spray cream sold to retail and dairy spray cream sold to OOH customers. The retail spray market includes both branded and private label products, while the questions as to the distinction between branded and private labels can be left open for the OOH market, as it will not have an impact on the competitive assessment.

13.3.2. Relevant Geographic market

13.3.2.1. Geographic market proposed by the notifying parties.

1427. The notifying parties submit that on the upstream market for the supply of spray cream, the geographic market is EEA-wide, with five strong competitors active in the entire EEA region.

1428. The notifying parties explained that in 2007, Campina sold spray cream in France, in Hungary, in Spain and in the United Kingdom. On the other hand, in 2007, Friesland Foods sold spray cream in France, in Germany, in Greece, in Hungary and in Spain. The notifying parties further explained that, since transportation costs of spray cream are low and shelf life is not a constraint, spray cream can be easily transported over the entire EEA. Furthermore, spray cream is, according to the notifying parties, bought through international (web-based) tenders by large retail and wholesale chains, such as Metro, Markant, Lidl, Aldi, Rewe and Edeka (Agenor).

13.3.2.2. Assessment of the Commission

1429. The market investigation showed that both the retail and the OOH market for spray cream include Belgium, Germany and the Netherlands. First, there are a number of similarities between these Member States, in terms of consumption habits, available product range, packaging and price. (857) Such similarities allow customers to source from any of these Member States (858) and suppliers to import spray cream across national borders. (859) Suppliers of spray cream have noted that product specifications across Member States mainly relate to the addition of sugar to spray cream and to the level of fat, two parameters suppliers can easily adjust to (time and costs are respectively estimated in 6-12 weeks and EUR 20-30 000, depending on the number of tests necessary). (860)

1430. These indications of a homogeneous market are also confirmed by sourcing patterns. Indeed, in retail segment, several retailers source spray cream not only from their own Member State (one between Belgium, Germany and the Netherlands), but also from all three of these Member States. (861) Moreover, most customers are able and willing to switch to neighbouring Member States, in the case of a price increase. In addition to the notifying parties, customers named as alternatives of supply Hochwald (Germany), Incopack (Belgium) and Codap (Italy). In this respect, all the sourcing alternatives indicated by the respondents, with the sole exception of Codap, are located in the area including Belgium, Germany and the Netherlands. For spray cream, Meggle (Germany) and Cooperlat (Italy) are also active, but the former mainly sells in Germany and Eastern Europe, while the latter mainly limits its sales to Italy and neither one has been indicated as a viable sourcing alternative by customers. (862)

1431. Replies from OOH customers indicated that they are currently sourcing the large majority of spray cream from the Netherlands (mainly from Friesland Foods and Campina), but are able and willing to switch to neighbouring Member States in case of price increase. (863)

1432. On the other hand, the suppliers of spray cream addressed during the market investigation have indicated that both their retailer and OOH customers are located throughout the EEA. (864) A closer look to their replies, however, shows that two major German producers focus their sales in Germany (865) and only one producer of spray cream sells its production on a truly EEA-wide basis. (866)

1433. Based on these considerations, it is concluded that both the retail and the OOH market for spray cream include at least Belgium, Germany and the Netherlands.

13.3.3. Competitive assessment

13.3.3.1. The marketshares of the notifying parties

1434. The notifying parties' combined market shares in the retail and in the OOH market for spray cream are reported in Tables 13-6 and 13-7. The table on the OOH market also reports figures concerning branded and private label products separately, as the distinction has been left open in this market.

Table 13-6: Market shares spray cream, retail, Belgium, Germany and the Netherlands - Source: Notifying parties.

1435. With respect to the market shares reported in Table 13-6, the notifying parties seem to have underestimated the market shares of Codap, Hochwald and Incopack. Indeed, by looking at the electronic data submitted by the notifying parties, the spray cream sales of the three main competitors in the branded market and in the branded plus private label market are exactly the same.

1436. In the retail market, there is a substantial overlap between the notifying parties' activities resulting in a post-merger market share of approximately [40-50]*%. On this market, however, there are a number of competitors with significant market shares, which are already investing on production expansion, foreseeing that some customers might switch away from the notifying parties.

Table 13-7: Market shares spray cream, OOH, Belgium, Germany and the Netherlands - Source: Notifying parties.

1437. The figures in Table 13-7 show that the merger does not substantially alter the competitive landscape in the OOH market for branded or private label spray cream, considering the very small overlap between the notifying parties ([0-5]* % of the OOH market equals to approximately [...]* tons of spray cream).

13.3.3.2. Ability of customers to switch to alternative suppliers

1438. The spray cream market in the relevant geographic area includes a number of active competitors to the notifying parties, namely, the German suppliers Hochwald and Meggle, the Belgian Incopack and the Italian producers Codap and Cooperlat.

1439. The market investigation sought to ascertain whether these competitors provide a credible sourcing alternative to the notifying parties, so that customers, especially in the retail market, have the possibility to switch to spray cream suppliers, in the case of a price increase.

1440. Indeed, the large majority of retailers consider that the notifying parties' competitors provide viable sourcing alternatives for both branded and private label spray cream. (867)

1441. Some retail customers are already relying on multiple sourcing of spray cream and, even in cases where single-sourcing applies, customers have indicated that they have switched to alternative suppliers in the past. (868)

1442. Switching is also possible in light of the circumstance that supply contracts are negotiated for an average duration of 6 months to 1 year.

1443. Replies from OOH customers indicated that they are currently sourcing a large majority of spray cream from the Netherlands (mainly from Friesland Foods and Campina), but are able and willing to switch to suppliers in neighbouring Member States in the case of price increase. (869) The fact that they are not currently switching does not seem to depend on unfavourable market conditions or other structural restraints, but rather on the fact that OOH customers are rather loyal to the notifying parties as their current suppliers. Indeed, some OOH customers multi-source their spray cream supplies and consider the notifying parties' competitors as viable supply alternatives, especially for private labels products. (870)

1444. It is therefore concluded that in the retail and in the OOH segment customers have the ability to switch and would not encounter significant obstacles to do so, in the case of a price increase by the merged entity.

13.3.3.3. Spare capacity available

1445. The market investigation also looked at the ability of competitors to expand their production of spray cream, following a price increase by the merged entity. In particular, it was sought to verify the amount of spare capacity available to such competitors.

1446. In this respect, the Commission directly contacted the spray cream producers, in order to verify the effective amount of capacity currently available, any plan to improve such capacity and promptly expand production and, more in general, their expectations as to future developments of the spray cream market.

1447. The result is that competitors indeed have the possibility to expand production. Some can expand production to cover an additional 20% of the market (sufficient to supply two or three large customers) without incurring additional investments. Other can expand production by additional 1.500 tons (approximately 5% of the market) on short notice. Respondents have not clarified whether their additional production would be directed toward the retail market or the OOH market. The first option, however, seems the most likely, as these respondents are currently focusing on supplying retail customers.

1448. It was also investigated whether any of the spray cream producers are planning to invest in additional production capacity. In this respect, the market investigation revealed that at least one supplier is planning large investments in capacity enhancement, foreseeing that some customers might desert the notifying parties after the merger. Such new capacity is expected be on-stream by the year 2009.

1449. It is therefore considered that, in the event of a post-merger price increase, the notifying parties' competitors would have the ability to provide a viable sourcing alternative for the notifying parties' customers willing to switch.

13.3.3.4. Conclusion

1450. For these reasons, the concentration is not likely to lead to a significant impediment of effective competition in the field of dairy spray cream sold to OOH and retail customers in the geographic market including Belgium, Germany and the Netherlands.

14. LIQUID COFFEE WHITENERS

14.1. Introduction

1451. Campina and Friesland Foods are both active in the production and sale of liquid coffee whiteners ("LCW") comprising coffee milk and coffee cream. Coffee whiteners are dairy and non-dairy products which are used as a taste enhancer in coffee.

1452. According to the notifying parties, plain (fresh and long-life) milk is sometimes used as a taste enhancer in coffee. However, the notifying parties consider that plain milk should not be regarded as part of a broad market for coffee whiteners, due to a higher price for LCW and the fact that plain milk is mainly used for the preparation of special coffee types (such as cappuccinos or café latte). Overall, the investigation has confirmed the submission of the notifying parties that plain milk should be distinguished from liquid coffee whiteners (871). The markets for plain milk (fresh or long-life) are discussed in the relevant sections of the decision.

1453. Friesland Foods (but not Campina) also produces powdered coffee whiteners which can be dairy (with skimmed milk powder) or non-dairy (with only sodium caseinate). The notifying parties submit that powdered coffee whiteners form a distinct product market from liquid coffee whiteners since they can be kept at an ambient temperature (whereas LCW once opened have to be stored refrigerated), they are produced through a completely different process and the price is almost twice the price of liquid coffee whiteners. Respondents to the market investigation

confirmed almost unanimously that powdered coffee whiteners have to be separated from liquid coffee whiteners (872). As mentioned in this recital, there is no overlap in powdered coffee whiteners and thus these markets will not be discussed in what follows. 1454. Therefore, the assessment will focus on the market for liquid coffee whiteners, where the activities of the notifying parties overlap.

14.2. Relevant product markets

14.2.1. Coffee milk and coffee cream belong to different product markets

1455. According to the notifying parties, liquid coffee whiteners include coffee milk and coffee cream.

1456. These products can be described as follows. Coffee milk is produced from cow's milk. In general, coffee milk is produced by evaporating water out of plain milk and has a shelf life of maximum 180 days. There are three types of coffee milk depending on the fat content and the dry matter percentage: Whole coffee milk is produced from whole milk and has a fat content of 7.5 to 15% and a dry matter of at least 20%; semi-skimmed coffee milk has a fat content of 1 to 7.5% and a dry matter of at least 20%: skimmed coffee milk has a fat content of maximum 1% and a dry matter of at least 20% (873).

1457. Non-dairy coffee milk is a particular kind of skimmed coffee milk in which vegetable fats, and not milk fat, are added. This production process (pre-treatment, evaporation, homogeneisation, standardisation, sterilisation or UHT-treatment) is the same for all types of coffee milk, be it dairy or non-dairy. Coffee milk is offered in different types of packaging: glass or polyethylene bottles, brick packs and cups (874).

1458. Coffee cream is derived from dairy based cream, which is itself produced from raw milk. The raw milk is first separated into skimmed milk and cream with a 40% fat content. Then the cream is collected in a tank and UHT-treated (6 seconds at 140°C) and after that it is homogenised, resulting in a shelf life of 136 days. Alternatively, the cream can be homogenised first and subsequently sterilised (10 minutes at 118°C), resulting in a shelf life of more than 136 days. Coffee cream is produced by the decrease of the fat content to a level of at least 10%, in order to fulfil the applicable legal requirements with regard to coffee cream in Belgium, the Netherlands and Germany (875). Coffee milk comes in the same packaging as coffee milk, namely glass or polyethylene bottles, brick packs and cups.

14.2.1.1. Product market definition proposed by the notifying parties

1459. The notifying parties argue that the main differences between coffee milk and coffee cream are the higher fat content of coffee cream and the differences in the production process. Nevertheless, they argue that from a demand-side perspective, coffee milk and coffee cream are close substitutes and that the majority of customers in the Netherlands do not perceive any differences between the two types (876).

1460. This statement is also supported by the fact that none of the notifying parties approach coffee milk and coffee cream separately in their strategic planning but rather together -as liquid coffee whiteners- even if they have a stronger position in one or the other product (877). Besides, according to the notifying parties, coffee cream is a small part of the global liquid coffee whiteners market (4% in the Netherlands).

14.2.1.2. Assessment of the Commission

1461. In a previous case (878), the Commission assessed the market for liquid coffee whiteners and concluded that it is not necessary to precisely delineate the relevant product market as it would not have affected the competitive assessment. Based on the preliminary market investigation, in the 6(1)(c) Decision, it was indicated that the majority of respondents (but not all of them) confirmed that coffee cream and coffee milk belong to the same product markets.

1462. In the present case, the Phase II market investigation produced mixed results with regard to this issue of a possible segmentation between coffee milk and coffee cream. Roughly half of the respondents endorsed the notifying parties' view that although both products exhibit differences in terms of fat content, these variations are hardly noticeable at the level of the end-consumer in terms of taste and functionality (whitening effect) of the coffee whitener. Therefore, according to these respondents, a slight but significant increase of prices of one of these coffee whiteners would lead customers to switch to the other.

1463. However, a significant number of respondents did not share the notifying parties' views and underlined that end-consumers are very loyal to the specific liquid coffee whitener they use, even in the case of a price increase of 5-10%. One competitor indicated that "Coffee milk and coffee cream are two totally different products. The impact of each in view of coffee taste is definitely different. So end-consumer which currently using usually coffee milk would not be satisfied with taste by using coffee cream and the same in the other direction" (879). This is also the opinion of a second competitor which stated that "The category of coffee whitener is regarded as price inelastic with very loyal consumers. A price change of 5-10% would not trigger a shift in consumption and change in consumer behaviour" (880).

1464. Similarly a Dutch retailer explained that "Coffee whiteners users are fairly loyal to their product, even if the prices of milk will get to the point where they are not that different from creamers anymore" (881). This opinion is shared by most retailers in the Netherlands.

1465. With regard to supply-side substitutability, the clear differences between both production processes do not lead to the conclusion that suppliers of one product are able to switch production to the other in the short term without incurring significant additional costs or risks.The notifying parties explained that coffee milk can be produced by producers of milk powder and producers of long-life milk with an additional equipment required ("evaporation unit"). Part of the equipment used for the production of coffee milk is uniquely suited to that purpose and therefore cannot be used for the production of other dairy products (for example coffee cream). This has been confirmed by one competitor who underlined that switching from coffee cream to coffee milk would require strong investment in an evaporation unit (882).

1466. Lastly, while it is true that coffee cream represents a limited part of the overall market for liquid coffee whiteners in the Netherlands (roughly 4%), this is not the case in Belgium, the other Member State where the activities of the notifying parties overlap with regard to liquid coffee whiteners. In Belgium, coffee cream represents roughly 25% of the overall market for liquid coffee whiteners. The notifying parties themselves acknowledge that "in Belgium and in Germany there is more awareness on the difference between coffee milk and coffee cream" (883).

1467. Within coffee milk, it is not considered appropriate to distinguish between so-called "dairy" and "non-dairy" coffee milk or according to the fat content. The reason for this is that the production process is exactly the same, with a small percentage (4%) of vegetable fats incorporated in the non-dairy coffee milk. With regard to fat content, all producers already cater for different tastes by producing a whole range of coffee milk with different fat contents. There is thus a high degree of supply-side substitutability between coffee milks with various fat contents (884).

1468. The limited demand-side substitutability between both products put forward by several market participants and the absence of supply-side substitutability point towards a separation between two product markets with regard to LCWs, namely coffee milk (regardless whether it is dairy or non-dairy or of the fat content) on the one hand and coffee cream on the other hand. Therefore, for the purpose of the present case, it is concluded that coffee milk and coffee cream constitute distinct product markets.

14.2.2. Private label and supplier brands belong to the same relevant upstream market.

14.2.2.1. Product market definition proposed by the notifying parties

1469. The notifying parties agree that a vertical segmentation of a particular product market for consumer products - in particular foodstuffs - can be made between an upstream market for the sourcing of products by retailers (and OOH wholesalers serving hotels, restaurant, catering services and filling stations) and a downstream market for the sale of foodstuffs by retailers to consumers for two reasons.

1470. First, this approach - taken by the Commission in previous decisions (885) - would recognise that on the retail-to-consumer level private label products compete with (producer) branded products, and that the market share of private label products should be attributed to the retailers that own the private label brands.

1471. Secondly, it recognises that the competitive conditions on the markets on which retailers and OOH wholesalers source their products are fundamentally different from the competitive conditions on the retailer-to-consumer markets. (886)

1472. The notifying parties agree (887) that the procurement of private label products and the procurement of branded products by retailers (and OOH wholesalers) might constitute two neighbouring, but separate markets, in the event that the competitive conditions under which retailers source private label products and branded products would be fundamentally different. This may be the case in situations where specific branded products are perceived as "must carry" products by retailers, in the sense that significant profits might be lost in the event of delisting such branded products.

1473. The notifying parties claim that in the context of the market for liquid coffee whiteners, both Campina and Friesland Foods are under constant pressure from private label products. Although in particular, Friesche Vlag in the Netherlands and Nutroma in Belgium are relatively strong brands, private label products would occupy such a strong position in both Member States that retailers could threaten suppliers of branded products with temporary delistings. According to the notifying parties, the combination of a highly concentrated retail segment and PL dominance ensures that the competitive conditions under which retailers source PL products and brands are similar. (888)

1474. Thus, private label and branded products would according to the notifying parties belong to the same product market at the upstream level.

14.2.2.2. Assessment of the Commission

1475. Liquid coffee whiteners are available in two broad categories: brands owned by the dairy manufacturer and private label products which are marketed by the retailer and the OOH wholesaler. Both Campina and Friesland Foods produce both branded and private label liquid coffee whiteners.

1476. In the 6(1)(c) Decision, it was indicated that a separation at the upstream level into the sourcing of private label and branded LCWs could not be excluded and that, during the second phase investigation, a potential distinction into private label and branded products would be scrutinized more closely. The elements gathered during this second phase investigation point to the conclusion that branded and PL products belong to the same sourcing market.

1477. All liquid coffee whiteners (889) producers sell their products to retailers and/or OOH wholesalers which, in turn, sell these products to consumers. Therefore, there are two stages in the supply chain: the upstream level, where LCWs are produced and supplied to retailers/OOH wholesalers and the downstream level of supply to consumers. Campina and Friesland Foods are only active on the production and supply (to retailers and/or OOH wholesalers) level.

1478. In a recent consumer goods case (890) the upstream level, where retailers source their products, from was distinguished from a downstream level, where the products are sold on to the final customer. In the present case, the same approach has been followed in the market(s) for liquid coffee whiteners.

1479. As already explained in Sections 7 and 11, the issue whether private labels and supplier brands belong to the same product market upstream also depends on other factors.

1480. It must in particular be assessed: (a) whether both types of brands, in general, compete closely with each other from the perspective of the end-customer and (b) the extent to which upstream suppliers of private label and/or brands as well as the purchasing retailers take the competitive pressure that private labels and supplier brands mutually exert on each other at consumer level into account in their negotiations upstream.

1481. For liquid coffee whiteners there exists with respect to the first element mentioned in recital 1480 significant evidence suggesting that private label and supplier brands compete in the downstream market and exert a competitive constraint sufficiently relevant for the purposes of market delineation on each other.

1482. In their responses to the market investigation the majority of customers and competitors confirmed that private label and branded products compete at the retail level and that the quality of private label products is to a large extent similar to branded products (891). No specific distinction can be made in terms of intrinsic product quality between branded and PL liquid coffee whiteners. Almost all retailers have private labels in addition to the brands offered by Campina and Friesland Foods and position them in relation to these brands, which have a price premium over private label products.

1483. In addition, on the downstream markets for liquid coffee whiteners the share of private label in volume terms was 47.5% in the Netherlands in 2007 (+0.4% since 2005), 43.9% in Belgium (+2%) and 62.7% in Germany (+1.1%) (892). PL products thus occupy a strong position at downstream level in these three Member States.

1484. In general, the LCW market can be characterised as a decreasing market since it has contracted in volume by 6.7% in the Netherlands, 3.8% in Belgium and 7% in Germany between 2005 and 2007 (893). In this context, PL products experienced during the same period a smaller decline than the overall market in the Netherlands (-6.1%) and in Germany (-5.3%) and even slightly increased in Belgium (+0.8%). These results show that PL products are gaining shares at downstream level on branded products.

1485. These elements suggest that there is a sufficient degree of competitive interaction between supplier brands and private labels for liquid coffee whiteners for both to be considered as belonging to the same relevant product market at retail level. This does not, in itself, necessarily imply that supplier brands and private labels should be considered as part of the same relevant product market at the upstream level.

1486. At the upstream level the available evidence is more mixed.

1487. First, customers and competitors (894) confirmed that the procurement of branded products follows a different procedure than the one for private label products. In the case of branded products suppliers and retailers agree on the gross price, discounts, listing fees, and promotions in bilateral negotiations. For private labels a tender procedure is usually used, followed by a selection of preferred suppliers which agree with the retailer on a net price in final negotiations.

1488. This, in itself, would not be decisive if the same suppliers were active in both segments. In that respect, it appears that the market presence of competitors and the notifying parties varies to some extent when it comes to branded or private label LCWs. According to the notifying parties' estimates, in the Netherlands, German competitors like Nordmilch, Hochwald and Milchunion Hocheifel mainly supply private label and brands only to a limited degree. On the other hand, Unilever is only active with its branded product Becel whereas Friesland Foods and Campina sell both private labels and branded products.

1489. Likewise, in Belgium, Nordmilch and Hochwald are only present with PL products and Unilever and Nestlé with branded products. On the other hand, German competitor Zott is active in both segments and Campina has a very limited presence in branded products in Belgium. 1490. Despite this "asymmetry" in the conditions of competition - the branded producers could constrain the private label producers, the opposite does not seem possible as a private label producer would have to invest into a brand and marketing before being able to compete - it should be remembered that private labels account for a significant proportion of the overall market - 47.5% in the Netherlands, 43.9% in Belgium and 62.7% in Germany- which has been growing over time. It should also be remembered that PL are gaining shares on branded products. Given the relative importance of private label sales and their growth to the detriment of sales of branded products, suppliers cannot be expected to ignore the competitive pressure that private labels exert on their brands.

1491. This growing importance of PL products has been also highlighted by respondents to the market investigation. One retailer for example explained that "The role of private label products will increase because of a) Customers have a great trust in the (name of the retailer) brand and b) difficulties for A-brand suppliers to create relevant arguments for customers to buy Abrands" (895).

1492. Taken together, all these elements, in particular the growing importance of private label liquid coffee whiteners relative to branded products and the competitive interaction between the two brands that affects the negotiations upstream, allow for the conclusion that in the context of the present case private label and branded products belong to the same product market upstream for liquid coffee whiteners.

14.2.3. Retail and OOH segments are distinct product markets

14.2.3.1. Product market definition proposed by the notifying parties

1493. Friesland Foods and Campina both sell LCWs to the retail and the foodservice channel (OOH). The OOH segment serves restaurants, cafes, hotels, catering services, hospitals but also small businesses like bakers or other food processing entities. The notifying parties put forward that a distinction between the retail channel and the OOH channel is justified with regard to coffee milk and coffee cream for four main reasons (896).

1494. First (897), according to the notifying parties the brands used in the OOH and the retail segment would be different.

1495. Second, suppliers often provide additional services to their OOH customers such as cobranding on cups on LCW. In addition, the notifying parties focus promotional activities specifically on the OOH channel (898). These promotional activities, targeted to the OOH channel, are aimed at differentiating the brands of the notifying parties from those of other suppliers and locking in OOH customers as much as possible.

1496. Third, the packaging of LCWs in the retail and the OOH segment is also different. In the OOH segment, cups are the best selling types of packaging and the least sold in the retail segment. Restaurants and cafes require cups while retail customers can easily do without and indeed mainly procure glass or polyethylene bottles.

1497. Finally, prices are significantly higher in the OOH segment, due to the additional services mentioned in recital 1495 and the more widespread use of small portion-packed LCW such as cups in the OOH segment. For instance in 2007, the average downstream price of LCWs in the retail channel was approximately 1.5 EUR/litre and 2.8 EUR/ litre in the OOH channel in the same year.

14.2.3.2. Assessment of the Commission

1498. In Friesland Coberco/Nutricia, the market investigation confirmed the distinction between the two distribution channels (retail channel and so-called food service channel) and therefore the product markets were delineated accordingly. A similar distinction was provisionally put forward in the 6(1)(c) Decision in this case.

1499. In the case at hand, respondents to the market investigation confirmed overall the existence of strong differences between both channels with regard to packaging, downstream prices and additional services provided to customers with respect to liquid coffee whiteners (coffee milk and coffee cream). Respondents underlined, however, that they use generally the same brand in the OOH and the retail channels and it appears that it is also the case for Friesland Foods (Friesche Vlag and Nutroma are brands available in the retail and the OOH segment).

1500. In conclusion, the elements mentioned in Section 14.2.3.1 and past Commission's practice allow for the conclusion that the OOH and the retail channel form two separate product markets.

14.2.4. Conclusion on relevant product market

1501. On the basis of the elements discussed in Sections 14.2.1 to 14.2.3, it is concluded that there are separate relevant product market for coffee milk and coffee cream. No separation into private label/branded products is necessary for these products. A distinction with regard to the distribution channel into retail/OOH should be made.

14.3. Relevant geographic markets

14.3.1. Relevant geographic market proposed by the notifying parties

1502. The notifying parties submit (899) that the relevant geographic market for coffee milk and coffee cream at the downstream level is not wider than national while at the upstream level it is wider than the Netherlands and includes in any event Belgium and Germany. This is due to several factors including among others (i) significant penetration of the Dutch market by German suppliers, in particular with PL products which dominate the market (ii) homogeneity of the products across Member States (iii) increasing cross-border flows from Germany to Belgium and the Netherlands due to a lower price level in Germany compared to Belgium and the Netherlands, and (iv) long shelf life of the products which allows transportation over large distances. All these elements would not allow suppliers in the Netherlands or in Belgium to increase prices profitably by 5-10% as they would lose significant volumes to foreign competitors.

14.3.2. Assessment of the Commission

1503. In Friesland Coberco/Nutricia (900), it was suggested that the relevant geographic market for long-life flavoured milk might be national because of the following reasons: there is a degree of divergence in the shares of relevant suppliers in the different Member states, consumer tastes differ between Member States, Euro brands are the exception brands, price differ between Member States and customers operate mainly on national level.

1504. In the 6(1)(c) Decision, it was indicated that differences in brands and the fat content among Belgium, Germany and the Netherlands indicate that at least for a potential branded market, the relevant geographic market might be narrower than proposed by the notifying parties. Therefore the relevant geographic market should be further examined.

1505. In their reply to the 6(1)(c) Decision, the notifying parties emphasized that differences in fat content across Member States could not serve as a basis to delineate national markets since these differences are not in themselves decisive to delineate product markets according to the fat content. With respect to brands, the notifying parties submitted that apart from a national branded market leader, PL is dominant in the various markets for LCW (901).

1506. In the present case, the second phase market investigation rather showed that the relevant geographic market would be wider than national.

1507. Several elements are relevant in that respect. First, the market investigation confirmed that several German producers are already supplying Dutch and Belgian retailers and OOH wholesalers with coffee milk and coffee cream. In the Netherlands, with regard to coffee milk, German producers Nordmilch and Hochwald deliver under PL significant volumes, in the magnitude of the volume sold by the notifying parties, into the Netherlands. The same is valid, albeit to a lesser extent, for German producers Milch Union Hocheifel and Frischli. For coffee cream, Hochwald, Nordmilch and Zott have also a significant presence in the Netherlands (902).

1508. In Belgium, Hochwald has a strong presence with PL products in coffee milk and this is also the case for Nordmilch and to a lesser extent for Zott. In coffee cream, Zott and Nordmilch are the main suppliers (903). Hence it appears that the divergences highlighted in the Friesland/Coberco Nutricia Decision are starting to erode due to the significant presence of these German suppliers, especially where PL products are included in the product market (904). Moreover, the fact that customers operate mainly at national level does not prevent them from sourcing private label at a wider level, especially in the framework of tender procedures where foreign suppliers from surrounding Member States are invited to participate.

1509. A majority of customers and competitors also informed the Commission that coffee milk and coffee cream could be considered as homogeneous products with no particular difference in taste, consumption habits or packaging according to Member States, with the limited exception of the "kännchen", a small plastic can available only in Germany (905). The market investigation did not yield strong differences in terms of consumption habits between Belgium, the Netherlands and Germany. Unlike in other areas such as fresh milk where Dutch raw milk has been considered as a key input, the wide majority of Dutch retailers who responded to the market investigation explained that coffee milk and coffee cream packagings do not refer to a potential Dutch origin of the product (906).

1510. Moreover, the fact that coffee milk and coffee cream are not perishable products and can be shipped in an ambient environment would facilitate trade between Member States. Indeed, almost all Dutch and Belgian customers source from German companies. Even Campina and Friesland Foods supply their Dutch customers from their production facilities in Belgium.

14.3.3. Conclusion on the relevant geographic markets

1511. Based on these elements, it is considered that the relevant geographic markets for coffee milk and coffee cream are wider than national and include Belgium, the Netherlands and Germany.

14.4. Competitive assessment

1512. Campina sells coffee milk primarily in Belgium and the Netherlands under the Campina and Landgold brands as well as PL. Campina also sells coffee cream in Germany under the Campina and Südmilch brands as well as PL, and a modest volume of coffee cream in the Netherlands.

1513. In Belgium and the Netherlands, Friesland Foods sells coffee milk under the Friesche Vlag and Nutroma brands as well as under PL Friesland Foods sells coffee cream under PL in the Netherlands and under the Moka cream brand in Belgium. Friesland Foods is not active in coffee milk and coffee cream in Germany. Therefore there is no overlap between the notifying parties' activities at downstream level in Germany and the only geographic markets that are affected by the merger at downstream level are Belgium and the Netherlands.

14.4.1. Coffee milk, retail market

1514. The notifying parties estimate (907) that the size of the retail market for coffee milk on the upstream level in Belgium, Germany, and the Netherlands was EUR 439 542 000 with a volume of 35 938 000 litres in 2007. In Germany the turnover in coffee milk was EUR 306 990 000, followed by the Netherlands (EUR 118 781 000) and Belgium (EUR 13 772 000).

14.4.1.1. Proposed transaction would not result in non-coordinated effects in the retail market for coffee milk.

1515. In the 6(1)(c) Decision, it was indicated that the transaction raises serious doubts as regard its compatibility with the common market because it was likely to create a dominant position in the markets for liquid coffee whiteners (908). The following elements were relevant in that respect : large market share on possible Dutch and Belgian national markets, especially for branded products in the retail and the OOH markets, the alleged closeness of the notifying parties in terms of competitive constraint and the limited switching possibilities in the branded segment.

1516. During the market investigation in the second phase, the Commission received only a few complaints from customers indicating that the merger might lead to higher prices, while most customers as well as competitors were not expecting a significant impact of the proposed transaction on the market.

1517. According to the notifying parties, the downstream retail market for coffee milk covering the Netherlands has a total value of EUR 118 991 000 in 2007, 43.8% of which was covered by private labels and the remaining 56.2% by supplier brands. The branded segment experienced a decrease in value since 2005 whereas the PL segment has slightly increased resulting in an increase of the private label share by nearly 2%. Market shares of the notifying parties and their competitors at downstream level are shown in Table 14-1.

Table 14-1: Market Shares Downstream Market for Coffee milk, Retail, Netherlands - Source: Form CO.

1518. According to the notifying parties' submission the Friesche Vlag brand was the strongest supplier brand in 2007 with a [30-40]*% market share, followed by Unilever-Becel with [10- 20]*%, and Campina brands with [0-5]*%. The two leading retailers in the Netherlands (Superunie and Albert Heijn) were able to achieve a share of [10-20]* % (Superunie) and [5-10]* % with their private label brands.

1519. The downstream retail market for coffee milk covering Belgium had a total value of EUR 13 787 000 in 2007, 23.8% of which was covered by private labels and the remaining 76.2% by supplier brands. The branded segment experienced a slight increase in value whereas the PL segment has slightly decreased resulting in a decrease of the private label share by nearly 1%. Market shares of the notifying parties and their competitors at downstream level are shown in Table 14-2.

Table 14-2: Market Shares Downstream Market for Coffee milk, Retail, Netherlands - Source: Form CO.

1520. According to the notifying parties' submission the Nutroma (Friesland Foods) brand was the strongest supplier brand in 2007 with a [40-50]* % market share, followed by Nestlé with [20- 30]* %, and Unliever-Becel with [5-10]*%. Campina has a negligible market share ([0-5]* %).

1521. Looking at the upstream market, where retailers source coffee milk from dairy producers, the picture is obviously very different because of the wider geographic scope of the market which includes Belgium, Germany and the Netherlands. According to the notifying parties' submissions, the sourcing retail market for coffee milk covering Belgium, Germany and the Netherlands has a total value of EUR 439 542 000 in 2007. Market shares of the notifying parties and their competitors at upstream level are shown in Table 14-3.

Table 14-3: Market Shares Upstream Market for coffee milk, retail Belgium, Germany and the Netherlands - Source: Form CO

1522. The notifying parties had a combined market share of [10-20]*% - basically unchanged since 2005 - because of the position of Friesland Foods in the branded segment ([10-20]*%), closely followed by Hochwald ([10-20]* %, declining) and Unilever ([0-5]*%, declining). The bulk of the market, however, is covered by private label products - 63.5% - where the notifying parties achieve [0-5]*% (Campina) and [0-5]*% (Friesland Foods). Here the German suppliers Nordmilch, Frischli, Hochwald and Turm supply roughly [90-100]*% of the entire segment.

1523. Thus based on the notifying parties own submissions, their combined market shares in the upstream market would hardly lead to an affected market in the retail market for coffee milk as they are slightly above [10-20]*%.

1524. Even if the market were narrower in geographic scope, the impact of the merger on competition on the retail market for coffee milk would be limited. Friesche Vlag (Friesland Foods) is the strongest brand on the market but Campina only has weak brands whose market share is small at downstream level in the Netherlands ([0-5]*%) and negligible in Belgium ([0- 5]*%). In terms of market presence, the closest competitor to Friesche Vlag appears to be Nestlé in Belgium and Unilever-Becel in the Netherlands.

1525. Some respondents have questioned the competitive constraint exercised by Unilever-Becel on Friesche Vlag arguing that the Becel brand is aimed at the health orientated segment and therefore does not have the same positioning as Friesland Foods and Campina. However, other products from the Friesche Vlag range have health propositions so Becel's positioning is not particularly specific (909).

1526. In addition, it appears that retailers have the possibility of switching to other suppliers. In principle, given the level of penetration of private label coffee milk in the Netherlands, and to a lesser extent in Belgium, the large supermarket chains could consider switching further volumes to private label where German companies are active. Indeed the market investigation has revealed that retailers already often source their private labels from several foreign suppliers. One Dutch retailer explained that "In the case one of the brands would disappear, private label would become an alternative to the (former) Campina or Friesland Foods brands" (910). This is also documented by the fact that this retailer delisted in Spring 2008 four Campina coffee milk products (911).

1527. Finally, the market investigation confirmed the submission of the notifying parties (912) that several main suppliers of coffee milk have additional capacity available to be able to supply the Dutch or the Belgian markets. The notifying parties explained that part of the equipment used for the production of coffee milk is uniquely suited to that purpose and therefore cannot be used for the production of other dairy products. Suppliers of coffee milk which invested in this type of equipment are faced with a declining demand of their products (913) and as a result, increasing spare capacity. This has been confirmed by one German competitor who explained that "Capacity is available due to stagnating or declining consumer demand with regard to new coffee consumption habits" (914).

1528. The notifying parties estimated the average available capacity of their competitors to be in the same range as their own available capacity, namely [...]*%. This has been broadly confirmed by the market investigation. In this respect, the consumption of coffee milk in Belgium and the Netherlands represents only 28% of the total consumption in the wider geographic market comprising Germany as well (25%, for the Netherlands and 3% for Belgium). Several German competitors indicated during the market investigation that they would have overcapacity which would cover entirely or at least a significant part of the Dutch demand. This is of course also true for Belgium, where demand is much smaller. (915)

14.4.1.2. Conclusion on competitive assessment on the retail market for coffee milk.

1529. For the reasons set out in Section 14.4.1.1, it can be concluded that the proposed transaction is not likely to lead to a significant impediment of effective competition on the retail market for coffee milk covering Belgium, Germany and the Netherlands.

14.4.2. Coffee milk, OOH market

1530. The notifying parties estimate (916) that the size of the OOH market for coffee milk on the upstream level in Belgium, Germany, and the Netherlands was EUR 132 888 000 in 2007 with a volume of 68 133 000 kg. The bulk of the turnover in coffee milk in the OOH market was achieved in Germany with EUR 99 116 000 followed by the Netherlands (EUR 22 195 000) and Belgium (EUR 11 576 000).

14.4.2.1. Proposed transaction would not result in non-coordinated effects in the OOH market for coffee milk

1531. During the market investigation in second phase the Commission received only a few complaints from customers indicating that the merger might lead to higher prices, while most customers and competitors were not expecting a significant impact as a result of the proposed transaction on the market.

1532. According to the notifying parties the downstream OOH market for coffee milk covering the Netherlands had a total value of EUR 22 195 000 in 2007, 26.5% of which was covered by private label and the remaining 74.1% by supplier brands. The total market has remained stable since 2005 whereas the PL segment has expanded a little (+0.9%) Market shares of the notifying parties and their competitors at downstream level are shown in Table 14-4.

Table 14-4: Market Shares Downstream Market for Coffee milk, OOH, Netherlands - Source: Form CO.

1533. According to the notifying parties' submission Friesland Foods with its brands Friesche Vlag and Nutroma would account for [60-70]*% of the market whereas Campina would hold [5- 10]*%. The combined market share would therefore be [70-80]*%, with competition stemming from private label products (26.5%).

1534. The downstream OOH market for coffee milk covering Belgium had a total value of EUR 11 576 000 in 2007, 27.4% of which was covered by private label and the remaining 72.6% by supplier brands. The branded segment experienced a decrease in value whereas the PL segment has grown resulting in an increase of the private labels share by slightly more than 6%. Market shares of the notifying parties and their competitors at downstream level are shown in Table 14-5.

Table 14-5: Market Shares Downstream Market for Coffee milk, OOH, Belgium - Source: Form CO.

1535. According to the notifying parties' submission, Friesland Foods with its brands Friesche Vlag and Nutroma accounts for [50-60]*% of the market whereas Campina with its brands Campina and Landgold holds [5-10]*%, resulting in a combined market share of [60-70]*%. Main competitors would be Unilever ([0-5]*%) and Nordmilch ([0-5]*%) whilst PL products represent [20-30]* % of the market.

1536. Looking at the upstream market, where OOH wholesalers source coffee milk from dairy producers, the picture is obviously very different because of the wider geographic scope of the market which includes Belgium, Germany and the Netherlands. According to the notifying parties' submission, the sourcing retail market for coffee milk covering Belgium, Germany and the Netherlands has a total value of EUR 132 888 000 in 2007. Market shares of the notifying parties and their competitors at the upstream level are shown in Table 14-6.

Table 14-6: Market Shares sourcing Market for Coffee milk, OOH, Netherlands + Belgium+Germany - Source: Form CO.

1537. The notifying parties had a combined market share of [20-30]*% - slightly in decline since 2005 - mainly because of the position of Friesland Foods in the branded segment ([10-20]*%). Market leaders are Hochwald ([20-30]*%, declining), Frischli ([20-30]*%, increasing), and Nordmilch ([20-30]*%, slightly increasing). Private labels represent 6.8% of the market (917), out of which Campina and Friesland Foods achieve less than the half and German competitors (Nordmilch, Frieschli and Turm) the other half.

1538. Thus based on the notifying parties' own submissions, their combined market shares in the upstream market would be moderate and several German companies would hold a stronger position than the new entity.

1539. Even if the market would be narrower in geographic scope, the impact of the merger on competition on the OOH market for coffee milk would be limited, for several reasons.

1540. First, with regard to brands, OOH customers did not indicate during the second phase investigation that Campina brands would be the closest competitors to Friesche Vlag and Nutroma. Although Unilever-Becel is not present in the OOH market in the Netherlands and has a low market share in Belgium, respondents underlined that it already exercises competitive pressure on Friesland Foods. For example, one OOH wholesaler active in the Netherlands and Belgium explained that the main brands in coffee milk in the OOH segment would be "Friesland Foods with Nutroma, Unilever with Becel" and that Unilever approached this wholesaler in 2007 to be listed as a supplier of branded products (918). This has been confirmed by another OOH wholesaler (919).

1541. The presence of Campina in the OOH market for coffee milk is further toned down by a respondent to the market investigation active in Belgium and the Netherlands who stated that "Campina is not big in coffee milk in OOH" (920).

1542. In addition, it appears that wholesalers have the possibility of switching to other suppliers. While it is true that the level of penetration of private label coffee milk is lower in the OOH segment than in the retail segment, customers in the OOH segment put forward that they have the possibility of switching to other branded products as well as PL products in the case of a price increase by the merged entity. These private label products are produced not only by the notifying parties but also by German companies who currently ship large volumes of coffee milk to the OOH segment in the Netherlands and in Belgium.

1543. For example, one OOH wholesaler indicated that "All brands of Friesland Foods can be replaced with Becel" and that a price increase of 10% of branded products would trigger "Opportunities for PL". With regard to the potential alternative suppliers for PL coffee milk, this wholesaler explained that "We are in contact with German suppliers such MUH. (Potential suppliers would be) Hochwald, MUH and Nordmilch" (921).

1544. Likewise, a customer in the OOH segment emphasized that "A private label product with a good price/quality rate probably has a fair chance to success on the on-premise market (LCWproducts are low-interest products)" and that "Due to the fact that these products are low interest products to end consumers, there certainly is a possibility for (...) to source elsewhere. Switching to private label products is also possible" (922). Therefore, despite an inferior market presence of PL in the OOH segment, customers would be able to switch to private label products manufactured by third parties should the merged entity increase prices.

1545. Finally, as in the retail segment, it appears that several main suppliers of coffee milk have additional capacity available to be able to supply the Dutch or the Belgian OOH markets. 14.4.2.2. Conclusion on competitive assessment on the OOH market for coffee milk.

1546. For the reasons set out in Section 14.4.2.1, it can be concluded that the proposed transaction is not likely to lead to a significant impediment of effective competition on the OOH market for coffee milk covering Belgium, the Netherlands and Germany.

14.4.3. Coffee cream, retail and OOH market

1547. The transaction does not lead to a significant impediment of effective competition in the coffee cream retail and OOH markets.

1548. With regard to the retail market, only Campina is active in Germany (with a market share of [5-10]*% in value in 2007) and in the Netherlands (with a market share of [10-20]*%) in the downstream national markets. In Belgium, Campina has no market presence and Friesland Foods has only a very small market share ([0-5]*%) (923).

1549. Looking at the upstream market, where retailers source coffee cream on a wider scale including Belgium, the Netherlands and Germany, the notifying parties achieve a combined market share of [5-10]*% with a minuscule overlap (Campina [5-10]*%; Friesland Foods [0-5]* %). Main competitors would be Hochwald ([5-10]*%), German company Gabler Saliter ([0- 5]*%) and Turm ([0-5]*%). The bulk of the market consists of private label products (75.3%) in which Hochwald, Nordmilch, Frischli and Zott achieve together more than [90-100]*% of the sales.

1550. On this basis, it is concluded that the merger is not likely to lead to a significant impediment of effective competition in the retail market for coffee cream.

1551. A similar picture arises in the OOH market for coffee cream. Looking at national downstream markets, Friesland Foods is not present in the Netherlands where Campina would hold a market share of [0-5]*%, or in Germany (Campina [0-5]*%). In Belgium, only Friesland Foods is active with a share of [10-20]*%.

1552. Looking at the upstream market, where OOH wholesalers source coffee cream on a wider scale including Belgium, the Netherlands and Germany, the notifying parties achieve a combined market share of [0-5]*% with a minuscule overlap (Campina [0-5]*%; Friesland Foods [0-5]*%). Main market players would be Nordmilch ([20-30]*%), Frischli ([20-30]*%), Hochwald ([10- 20]*%), Zott ([10-20]*%) and German company Meggle ([10-20]*%). Private labels account for 2.9% of the sales.

1553. On this basis, it is concluded that the merger is not likely to lead to a significant impediment of effective competition in the OOH market for coffee cream.

14.4.4. Conclusion on competitive assessment

1554. In the light of these elements, it is concluded that the transaction is not likely to lead to a significant impediment of effective competition in Belgium, the Netherlands and Germany in the retail market for coffee milk, the OOH market for coffee milk, the retail market for coffee cream and the OOH market for coffee cream.

15. SPRAY DRIED EMULSIONS

15.1. Introduction

1555. Both Campina and Friesland Foods are active in the production and sales of spray-dried emulsions ("SDEs") for use in the food sector. Six segments can be identified in the sector of SDE with regards to the level of know-how involved, the applications and the production processes: creamers, foamers, toppings, batter stabilisers, fat concentrates and encapsulated nutritional oils (924). Campina only sells creamers, foamers, toppings and batter stabilisers through its subsidiaries DMV International and Satro (925). Friesland Foods supplies SDEs through its subsidiary Kievit and the product portfolio of Friesland Foods consists of the six segments of SDEs mentioned in this recital.

1556. In general, an emulsion is a finely dispersed mix of one liquid with another liquid which, in principle, do not mix (oil and water). Spray-dried emulsions (SDE) are emulsions in which the water is evaporated through spray-drying. For food applications, SDEs consist of a combination of oil or fat, carbohydrates (as a carrier) and an emulsifier (usually proteins) and possible other ingredients such as sugar and additives.

1557. SDEs are used for different applications such as beverages, soups, sauces, bakery products and infant nutrition. All SDEs can be represented on a sliding scale of added value. Some SDE such as creamers can be considered as basic while other SDEs such as encapsulated nutritional oils are technology advanced and know-how driven. However, even some allegedly basic SDEs in the creamer or foamer segments may require a high level of product knowledge. The composition of SDEs (especially the oil/protein ratio) differs for each application.

1558. The process of making SDEs consists of the following steps (926). First, a liquid emulsion mix is produced in a mixing tank. This liquid mix contains fat or oil carbohydrates, water, dry ingredients (sugar and additives) and an emulsifier (notably proteins). It is then pasteurised and homogenised in a homogeniser. At this stage, the mix consists of a continuous water blend in which the oil is finely dispersed as small outlets.

1559. Second the liquid emulsion is pumped to the top of a spray dryer under high pressure and atomised in a hot airstream by means of, for example, a nozzle into the drying chamber. The hot air causes the water to evaporate. The residual dry powder is collected at the bottom of the drying chamber where it is cooled in an external fluid bed by cool air.

1560. Spray drying is carried out in a spray drying tower, the height of which varies between 30 and 40 meters and the diameter of which varies between 10 and 15 meters. There are two main types of spray dryers - conventional/MSD spray dryers and filter mat spray dryers. The main difference between these types of spray dryers is the manner in which the dried powder is processed once it comes out of the drying chamber. In filter mat spray dryers, the cool air flow is constant (whereas it is more turbulent in conventional/MSD spray dryers) which allows for the drying of the emulsion with minimal physical stress. The notifying parties submit that this mild drying of the emulsion is particularly effective for the production of fat concentrates (no oil leaks which require cleaning of the tower so long production runs are possible) and encapsulated nutritional oils (low level of oxidation in the filter mat spray dryers).

1561. As mentioned in recital 1560, six market segments can be identified in the SDE sector (927). Creamers are products in powdered form used to give end-products a creamy texture and/or a white colour. Creamers are used in instant coffee whiteners, instant beverages, dry soup mixes and bakery pre-mixes to produce bakery creams. Different varieties of creamers are possible so that the creamers can be adapted and customized to the wishes of the customers, for example with regard to the oil which is used in the production process (coconut oil, palm oil) or the proportion of protein and fat. Creamers are dry products with good storage stability and a guaranteed shelf life of two years.

1562. Foamers for beverages are used to produce an instant foaming effect when mixed with water in beverages, for example in cappuccinos or instant soups. A foamer is an SDE of the creamer type where gas is added before evaporating the water of the emulsion. By using this technique, gas bubbles are entrapped in the powder and will only be released when water is added. Foamers can also be adapted to customers' wishes by changing the fat content, the nature of the fat or the amount of gas contained in the foamer.

1563. Toppings are products in powdered form that are added to milk, water or another liquid to create stable foam by whipping the mix. They are the aeration device in powdered instant desserts such as chocolate mousse or in bakery applications (aerated fillings such as "crème Chantilly or "crème patissière"). Apart from the ability to aerate, it provides a creamy taste and texture to the final product.

1564. Batter stabilisers are products in powdered form that are added to egg batters for sponge cake, pound cake or muffins to provide good cake structure, taste and stability.

1565. Fat concentrates are mostly used as fat additives in bread, thereby extending the shelf life of bread as well as its texture. It is mostly sold to bakeries which mix it with water and a few additives. Other applications are mixes for cakes or bread that are sold to consumers in supermarkets in order to bake their own cake or bread at home. Unlike other SDEs, fat concentrates are mainly produced on filter mat spray dryers.

1566. Encapsulated nutritional oils (or Omega 3/6 oils) are used in baby food and enhanced nutrition and would have proven benefits for adults and children (brain, neural and eye development). They are produced from algae or fungi, soy, linen seeds or sunflowers (as opposed to the tropical oils like palm or coconut oil used in creamers, foamers, toppings or fat concentrates).

15.2. Relevant product markets

15.2.1. Liquid emulsions are a distinct product market from spray dried emulsions

1567. In the food industry liquid emulsions are also used primarily for bakery applications. For example liquid toppings are emulsions based on vegetable fat packaged in 1 litre bags or in larger bag-in boxes and liquid emulsifiers can be added to egg batters to replace batter stabilisers. However, the notifying parties indicated that if a customer wants to switch from SDEs to liquid emulsions, adaptations of the recipes and the production processes have to be carried out.

1568. A majority of customers and competitors in the market investigation confirmed that liquid emulsions and SDEs are not substitutable928. Liquid emulsions require different processing/packaging equipments which are not available by SDEs customers. For example, one (928) Questionnaire to customers SDEs sent on 17 June 2008, question 5 and questionnaire to competitors SDEs sent on 17 June 2008, question 5.

Customer (929) uses SDEs in a mix with other dry powder based raw materials, which are ultimately re-packed in a final packaging. This process cannot handle liquid raw material and therefore liquid emulsions are not a viable option. Moreover, the shorter shelf life of liquid emulsions makes them unsuitable for powder applications.

1569. Therefore, liquid emulsions and SDES are considered as separate product markets.

15.2.2. Different categories of SDEs identified by the notifying parties constitute distinct product markets

1570. The notifying parties argue that within the market segments for SDE, each product group is produced for a specific application and there is hardly any demand-side substitutability between these product groups (930). The only exception would be toppings and batter stabilisers, which the notifying parties view as part of the same product market since the essential ingredients would be the same (fat based emulsifiers) and the same customers would be targeted. The notifying parties also submit that within each segment, demand-side substitutability is optimal for end-users, whereas producers of an end product can use another creamer or foamer and the final product will remain the same (931).

1571. With regard to supply-side substitutability, the notifying parties indicate that it would be high between creamers and foamers because only a gas injection system, of approximately EUR 100 000 is needed in addition to the spray dryer to switch from the production of creamers to the production of foamers. According to the notifying parties, this is evidenced by the fact that most creamers manufacturers produce foamers as well. Switching from the aforementioned products to toppings and batter stabilisers would only require cleaning of the spray dryer and would therefore be fairly simple.

1572. With regard to fat concentrates, they are produced more efficiently by using a filter mat spray dryer and supply-side substitutability with other products is therefore limited. The same is valid for encapsulated nutritional oils because of the more stringent hygiene requirements bearing on these products which trigger higher cleaning costs of the spray dryer.

1573. The market investigation confirmed the lack of demand-side substitutability between these different product groups - customers and competitors underlined indeed that these products are incorporated into specific recipes (cappuccinos, instant soups and sauces, etc) which meet specific consumers' needs. Therefore SDEs are not substitutable from a demand side point of a view.

1574. As regards supply-side substitutability between creamers, foamers, toppings and batter stabilisers, the market investigation showed that some competitors are only active in the segment of creamers. Competitors to the notifying parties confirmed, however, that the hardware equipment needed to manufacture these products is similar (spray-drying tower) with the

exception of foamers where a gas injection system is needed. Nevertheless, a supplier willing to switch from creamers/foamers to toppings for example would have to acquire significant knowhow with regard to the combination of the main ingredients of a topping powder (vegetable fats, emulsifier, milk proteins and glucose syrup) (932), given that this combination has a strong impact on the performance of the end-product (933). Some competitors, such as German company Cognis are only active in the toppings segment.

1575. The same is valid for a switch from creamers to foamers which requires a specific know-how with respect to the ingredients and the interaction with the finished products. Foamers are considered by competitors as a more complex product to manufacture (934). Moreover some competitors such as Kerry Group and Palsgaard produce only creamers and not foamers whereas others such as Maspex are only active in the foamer segment.

1576. As regards toppings and batter stabilisers, the market investigation did not fully back up the notifying parties' view that these products belong to the same product market. First, toppings and batter stabilisers do not fulfil the same functions (aerate an end product for the former, stabilise a pastry product for the latter). Most of the competitors of the notifying parties manufacture only one product: for example Cognis, Meggle and Premier Foods produce only toppings. This is also the case for Friesland Foods, which has a small position in toppings, but no sales in batter stabilisers. Lastly, while this is true that customers in the bakery sector purchase toppings and batter sectors (but for different purposes), there are some customers outside the bakery sector which purchase toppings mostly for aeration purposes in milk-based beverages or soups, but no batter stabilizers (935).

1577. It is therefore concluded that creamers, foamers and toppings belong to separate product markets. As Campina is not active in fat concentrates or encapsulated nutritional oils and Friesland Foods is not present in the batter stabilisers segment, the precise product market definition for these three products is not addressed.

15.3. Relevant geographic markets

1578. The notifying parties submit that the geographic scope of the markets for SDEs is EEA-wide and possibly worldwide due to the fact that the notifying parties export a substantial amount of SDEs outside the EEA. Prices would also be different between the EEA and other areas. They argue that this is particularly true for encapsulated nutritional oils, since the high value of these products justifies higher transport costs and import duties. The notifying parties also submit that due to the decreasing US dollar value, it is becoming increasingly attractive to ship SDEs in general from the United States.

1579. The market investigation showed that customers from the notifying parties source these products at EEA-level. Worldwide procurement appears to be less widespread since some customers indicated that emulsions sourced from other areas (Asia, USA) are more expensive due to tariffs and import duties. Indeed, the notifying parties confirmed that an import duty between 4.1% and 5.5% and a customs duty (depending on the product composition) apply to SDEs imported from outside the EEA (936). Customers in the market investigation confirmed that these duties, as well as additional costs triggered by overseas procurement, prevent them from sourcing from abroad (937).

1580. Furthermore, some customers indicated that imported SDEs are not fully comparable in terms of quality, raw materials and compliance to their specifications. Since the driving factor for a choice of a SDE supplier is the performance of the SDE in a final application in terms of taste, physical performance (for example foaming capacity for a foamer) and shelf life, customers do not consider sourcing from Asia as a viable alternative for their final applications. For some of these customers, SDEs are a very important input for which proximity with the supplier is a key issue (938). Additional potential problems with regard to stock-keeping and delivery times have also been mentioned (939).

1581. Finally, contrary to what the notifying parties claim it does not appear that companies from outside the EEA are looking to enlarge their presence in the EEA. This is especially true in the market for creamers where the worldwide leader Bay Valley Foods is not active in the EEA and the second worldwide largest player (Kerry Group) has a market share of [5-10]*% in Europe.

1582. It is therefore concluded that the markets for creamers, foamers and, toppings are EEA-wide in scope.

15.4. Competitive assessment

1583. Campina supplies SDEs through its subsidiaries DMV International and Satro. It sells creamers (under the brand name DP), foamers and toppings (under the brand name Aerion) and batter stabilisers (under the brand name Textrion). The portfolio of Friesland Foods-Kievit consists of the following products: creamers sold under the brand names Vana-Blanca, Vana- Crema, Vana-Cerea and Vana-Latta; foamers sold under the brand name Vana-Cappa and toppings sold under the brand name Vana-Monte. The notifying parties' activities do not overlap in the markets for batter stabilisers, fat concentrates and encapsulated nutritional oils and therefore these markets will not be considered further.

15.4.1. Creamers

1584. In the Article 6(1)(c) Decision, it was indicated that during the Phase I investigation critical views were expressed by customers with regard to the effects of the merger on the creamer market. These customers viewed Campina and Friesland Foods as the closest substitutes in the creamer market considering the wide range of creamers they supply, the high quality of their products and the strong R&D support they provide to their customers. They argued that alternative suppliers might be strong for standard products but do not seem to be able to provide customized products which meet their specific requirements in terms of quality and interaction with the end-product.

1585. Furthermore, the Phase I investigation showed that in order to switch suppliers, these customers have to put a supplier approval process in place which is lengthy (1-2 years for some creamers), costly and burdensome. Customers who source from only one supplier have also expressed concerns since they use the rivalry between Campina and Friesland Foods to secure high quality and fair prices.

1586. Table 15-1 shows the notifying parties' shares in the market for creamers in value at EEA Level (940).

Table 15-1: Market shares creamers, EEA - Source: Form CO, Section 7M, Table 17.

1587. According to the notifying parties, the market for creamers in the EEA had a total value of roughly EUR 269 million in 2007. On this market, the transaction would strengthen the leading position of Friesland Foods with a market share of [30-40]*% which has slightly increased since 2005 ([30-40]*%). Main competitors would be German companies DPS-Hochwald ([10-20]*%, stable), Meggle ([10-20]*%, stable), Hochwald ([5-10]*%, stable), US competitor Kerry Group ([5-10]*% stable) and Hochdorf from Switzerland ([5-10]*%, stable). Other smaller competitors would hold market shares of [0-5]*% or below.

1588. Creamers are used in a variety of applications in the food industry. According to sales' data of Friesland Foods' subsidiary Kievit, the main area where creamers are used is "instant beverages" which represent [...]*% of Kievit's creamer sales. The finished products manufactured with creamers in this area are instant cappuccinos, instant coffee chocolate drinks and cappuccinos/chocolate drinks for vending machines where creamer is dosed into the vending machine and sold in combination with coffee. Customers of the notifying parties sell these finished products to the retail segment or integrate creamers in mixes for vending machines ([...]*% of customers are active in vending).

1589. A significant alternative outlet for creamers is savoury creamers which account for slightly less than [...]*% of Kievit's sales. Here the product is used to manufacture dry soups, sauces and pasta sauce dry meals. The other areas (ice creams, bakery, infant formula, blends) account each for less than [...]*% of Kievit's sales (941).

1590. With regard to creamers for instant beverages, the market investigation highlighted the fact that the market will undergo significant changes in the following months. The notifying parties indicated during the Phase II investigation that Campina will exit the instant beverages creamer segment from 1 January 2009 onwards and sell/close down its coffee creamer plant in Prenzlau

(Germany). Campina has stopped the production of these creamers as of 1 September 2008 and these products are currently still for sale until stocks have been depleted (942).

1591. Campina took this step in spring 2007 in the light of the declining sales and poor financial results achieved in the coffee creamer segment. The production of these creamers has been decreasing for several years ([...]* in 2005 to [...]* in 2007) and a negative result of EUR [...]* has been achieved. [...]*.

1592. In the light of this development, customers of the notifying parties in the instant beverages creamer segment will have to search for alternative suppliers, regardless whether the merger proceeds. It is worth noting in that respect that Campina's decision to cut back on the production of creamers for instant beverages applications has no direct link with the transaction since it was taken in Spring 2007 and hence prior to the announcement of a planned merger between Campina and Friesland Foods, which took place in December 2007. The decision to sell/close down the plant in Prenzlau was finally taken in April 2008 following the agreement of DMV's division for consumer products in Prenzlau.

1593. Campina has two subsidiaries active in creamers for instant beverages applications, namely DMV and Satro. The decision to exit the instant beverage creamer segment concerns only DMV. With regard to Satro, the notifying parties confirmed that Satro only supplies instant beverages creamers, together with hot beverages under the brand name Satro Coffee & Co. Satro also integrates non-dairy creamers in cocoa drinks, instant teas and other hot soluble products which are sold to be used in vending machines. These dairy and non-dairy creamers are sourced from several companies, [...]* (943). Therefore it appears that Satro does not supply instant beverages creamers as a stand-alone product (944) to other downstream customers and that the entire overlap between the notifying parties' activities in the instant beverage creamer segment will be eliminated by the removal of DMV from this segment.

1594. As a result, customers in the instant beverages segment sourcing creamers from both notifying parties would have to replace Campina-DMV as a supplier, irrespective of the merger. In that respect, the market investigation showed that several suppliers were considered by customers as potential alternatives in terms of price/quality for instant beverages creamers, provided that their products pass through the supplier approval process mentioned in recital 1585 (945). Meggle, Krüger, Lyempf, Milchwerke, Hochwald, and Mokate were mentioned as credible substitutes by customers.

1595. With regard to savoury creamers customers of the notifying parties use these creamers to manufacture soups, sauces and pasta meals. Most of these creamers are customised to meet specifications of customers. The notifying parties define customised creamers as recipes or product composition tailored to customers request and produced for one customer according to its wishes. In value, [...]*% of creamers sold by Campina-DMV are customised, [...]*% of Satro's turnover is achieved by customised creamers and Friesland Foods achieves [...]*% of its turnover through creamers which are customised to a greater or a lesser extent (946). The main customer in the savoury segment, Unilever, mainly requires customised creamers.

1596. The process of customising creamers is usually initiated or triggered by the customers themselves. Suppliers customise creamers by making adjustments in type or concentration of ingredients (level of protein or type of ingredients). When a customised product is developed, suppliers are usually free to sell this product to other customers which have not initiated the customising process. However, there might also be exclusivity between the customer and the supplier, especially when the customer provides a specific oil or caseinate to be used or requires a specific recipe. In that situation, the dedicated R&D division within the customer's organization cooperates with the supplier and thus develops a high level of knowledge of the creamers required for the final products of the company.

1597. The duration of the customisation process depends of its complexity, the level of cooperation between the supplier and the customer, and the resources available in terms of personnel and equipment for the tests. For the customer, the main criteria to approve the products and the supplier are the comparison with the current approved creamer in actual finished product recipes in terms of taste and physical performances (whitening power, creaminess). For that reason, the validation process can be particularly long. The notifying parties indicated that this process takes between one and 6 months but the market investigation showed that the validation process is likely to last much longer (from 18 to 24 months). (947) However, the switching costs associated with a change of supplier do not appear to be very high: the market investigation indicated that it would amount to EUR 75 000-100 000 per type of creamer.

1598. Given the length and the complexity of the customisation process, customers in the savoury creamer segment are not inclined to switch suppliers on a regular basis. When customers are supplied with products which meet their specific requirements, the decision to switch supplier would trigger a new validation process for the whole of their product base, which these companies are not often willing to undergo. Since the relationships between suppliers of customised products and their customers are often long-term relationships in which both parties know a lot about each other on product development and quality, the decision to switch supplier is not easy to take. The objective of savoury creamer suppliers is to approve additional suppliers to have a broader selection and competitiveness on the market.

1599. In that context, the issue at stake is whether there remains a sufficient number of alternative suppliers on then market who would be able to provide these customised savoury creamers even though they are not yet qualified by customers. The market investigation showed that companies such as Meggle, DP Supply, Kerry and Maspex already provide customised creamers to the savoury segment (948). The fact that some large savoury creamers customers are not currently sourcing from these companies is not linked to a lack of ability to supply them, but rather to the fact that these customers are reluctant to modify their supplier base given the lead time needed to approve an additional supplier. Therefore, it appears that for customers who currently only source from Campina and Friesland Foods and who have not qualified other suppliers, alternative suppliers would remain available and the merger would not lead to a reduction of choice in the savoury creamer segment.

1600. With regard to other potential applications for creamers (infant formula, ice creams or bakery products), no specific concerns have been voiced during the market investigation.

1601. Accordingly, it is concluded that the merger is not likely to lead to a significant impediment of effective competition in the creamer market.

15.4.2. Foamers

1602. With regard to foamers, it was indicated in the 6(1)(c) Decision that Campina is viewed by a majority of customers as the closest competitor to Friesland Foods in the foamer market although some of them stated that it has a limited product range. Other recognized players include Mokate and Meggle although they were ranked somewhat lower in terms of product assortment, product availability and prices.

1603. Table 15-2 shows the notifying parties' shares in the market for creamers in value at EEA Level (949).

Table 15-2: Market shares foamers, EEA - Source: Form CO, Section 7.M, table 20.

1604. According to the notifying parties, the market for foamers in the EEA had a total value of roughly EUR 60 million in 2007. On this market the new entity has a market share of [40-50]*%, which has been stable since 2005, albeit with a limited overlap ([0-5]*%). Main competitors would be the Polish company Mokate ([20-30]*%, slight decrease), Meggle ([10-20]*%, stable), Polish competitor Maspex ([5-10]*%, stable) and Krüger from Germany ([5-10]*%, stable).

1605. In the Phase II investigation, it was investigated whether the current low market share of Campina in the foamer market correctly reflects the competitive pressure that it exercises on Friesland Foods, in light of internal documents from 2005 submitted by the notifying parties where it appeared that Campina had the intention of gaining market shares in the foamer market to the detriment of market leader Friesland Foods.

1606. The market investigation clarified that the low market share of Campina is in reality representative of the limited competitive pressure that it currently exercises on Friesland Foods. It is true that in 2005, Campina decided to focus more on the foamer segment and to become an important supplier of foamers within a few years as part of a renewed focus on aeration (950). However, the notifying parties submit that the results of Campina in the foamer market were far below its ambitions. Campina launched a new foamer product in 2006 but the growth in sales was only moderate (increase in market share from [0-5]* to [0-5]*% between 2006 and 2007) due to, according to the notifying parties, a lack of business intelligence on the foamer market and an inadequate management system (951).

1607. The notifying parties submit that the production of foamers is still difficult for Campina, which has not achieved the significant growth that it expected in this segment. For example in 2008, Campina grew [...]* metric tons in volume which is a small amount compared to a total market of 23 000 metric tons in the EEA with a market growth of over 5% since 2005.

1608. Moreover, the sale of foamers is likely to become even more difficult for Campina since the overwhelming majority of foamers (more than [...]*% for Friesland Foods Kievit, [...]*% for Campina DMV) are used for the same applications, namely instant beverages, and sold often together to the same customers. Since Campina ceased the production and sales of these instant beverages creamers as of 1 September 2008, customers have had to find alternative suppliers for creamers. It can therefore be expected that at least some customers will stop buying foamers from Campina now that Campina has ceased to supply creamers for beverage applications and will rely on suppliers who can offer the whole range of spray-dried emulsions (creamers and foamers) used in the instant beverages applications.

1609. Therefore it is concluded that the merger is not likely to lead to a significant impediment of effective competition in the foamer market.

15.4.3. Toppings

1610. With regard to toppings, it was indicated in the 6(1)(c) Decision that on the basis of the Phase 1 investigation, Campina and Friesland Foods might be particularly close competitors and the merger might remove an alternative source of supply, for some customers in the bakery and ice cream sector.

1611. Table 15-3 shows the market shares of the notifying parties and their competitors in value at EEA level, according to the notifying parties' best estimates (952).

Table 15-3: Market shares toppings, EEA - Source: Answer to the Commission's request for information dated 11 July 2008, page 7.

1612. According to the notifying parties, the market for toppings in the EEA had a total value of roughly EUR 40 million in 2007. On this market, the new entity has a market share of [40- 50]*%, which has been stable since 2005, albeit with a limited overlap ([0-5]*%). Main competitors are German companies Cognis ([30-40]*%, stable) and Meggle ([10-20]*%, stable). All other competitors (Palsgaard, Mokate, Suwelack, Molda and Lactalis) would hold shares of [0-5]*%.

1613. Respondents to the Commission's investigation did not fully confirm the market share estimates submitted by the notifying parties. Whilst sales in value of Meggle and smaller competitors have been corroborated in the market investigation, it appears that sales of Cognis would be higher than the notifying parties' sales post merger and that Cognis would by far remain market leader after the merger.

1614. This limited change in the toppings market structure is confirmed by respondents to the market investigation, especially topping customers who acknowledged that toppings was already the domain of Campina which developed further experience and know-how on this type of SDE. By contrast, customers underlined that Friesland Foods is only active on this market to a very limited extent and it is noted that Friesland Foods has identified the toppings segment as [...]* in its internal documents. No specific competition concerns have been voiced with regard to the toppings market in the market investigation.

1615. It is therefore concluded that the merger is not likely to lead to a significant impediment of effective competition in the toppings market.

15.5. Overall conclusion

1616. Therefore, it is concluded that transaction is not likely to lead to a significant impediment of effective competition in the various spray-dried emulsions markets (creamers, foamers and toppings).

16. LACTOSE

16.1. Introduction

1617. Both Campina and Friesland Foods are active in the production of lactose. With regard to the production of pharmaceutical lactose, Campina is active through its 50% interest in DMV Fonterra Excipients ("DFE"), a joint venture for the production of pharmaceutical excipients. The parent company that holds the remainder 50% interest of DFE is Fonterra from New Zealand. Friesland Foods is active in the production of pharmaceutical lactose through its fully owned subsidiary Domo.

16.2. Relevant product market

16.2.1. Description of the product

1618. Lactose - also known as milk-sugar - is a sugar derived from milk products. Lactose is less sweet than other sugars such as sacharose and glucose. Cow milk contains 4% to 5% lactose. Whey, a by-product of the cheese production, is most commonly used as the major input and starting point of the lactose production.

1619. Lactose is derived by a crystallization process applied to the whey which is followed by washing and drying of the lactose crystals. In principle, the resulting product is an "edible" form of lactose. As pure lactose forms big hard crystals that do not dissolve easily, secondary steps of processing are applied to reach more homogeneous sizes of lactose crystals.

1620. Each of the different production steps creates a number of different qualities of the product. In order to reach a more refined lactose, proteins, minerals, vitamins and other impurities are reduced by secondary processing steps. For example, the process of milling and sieving separates the lactose particles into different size-groupings and the resulting product are referred to according to the (maximum) particle size (for example "Milled lactose with mesh size 150"). Milled and sieved lactose then constitute the starting point for further secondary processing (such as spray drying, roller drying, or agglomeration) which lead to even purer and more homogeneous qualities of lactose.

1621. Lactose principally comprises two main grades: pharmaceutical grade lactose and edible grade lactose. Edible grade lactose is used in fine chemical applications, the manufacture of infant formulae and is added to food and confectionery products. Pharmaceutical grade lactose is used as a pharmaceutical excipient. In a previous case it was concluded that pharmaceutical grade lactose cannot be considered to belong to one single relevant product market for lactose (including also edible grade lactose (953)). The market investigation confirmed this conclusion. (954)

1622. Pharmaceutical lactose is used as an excipient. Pharmaceutical excipients are the non-active ingredients which are included in the drug manufacturing process or are contained in a finished pharmaceutical product dosage form. There are different pharmaceutical excipients that have different chemical composition, basic substance and functionality. In a previous case it has been concluded against a broad market for pharmaceutical excipients (including all excipient based on different basic substances and for all dosage forms). Essentially, there are three different technologies to use lactose as an excipient in a pharmaceutical drug: (i) wet granulation, (ii) direct compression (DC), and (iii) dry powder inhalation (DPI). Wet granulation and direct compression are different methods used in the manufacture of tablets. Whereas the direct compression process enables the pharmaceutical company to combine the lactose without any further processing with the active substance of the drug and produce ("compress") the tablets, the wet granulation process leaves it to the pharmaceutical company to produce in subsequent moistening and drying processes the quality of the lactose excipient needed for the tablet.

1623. Pharmaceutical excipients serve a number of different purposes in a drug formulation. Excipients can be used to transport the active ingredients to the right part of the body, prevent the active ingredients from being released too early, ensure disintegration of the drug, protect the drugs' stability, improve the look and taste of the drug or assist in the identification of a drug product. As a rule, most of the excipients provide several different functionalities for the finished product (the pharmaceutical). As regards the basic substance used, the following substances, sometimes after further processing, may be used as pharmaceutical excipients amongst others: lactose, starch, cellulose, magnesium, stearic acid, gelatine, sucrose, talc or sodium.

1624. There exists a wide variety of different dosage forms of a drug for very different forms of administering the active substance to the body, that is liquids (injections, syrup), semi-solid dosages (ointments, gels), solid dosages (tablets, capsules) or drugs for inhalation. Because of these different methods of administering the active substance to the body, pharmaceutical excipients specifically used in one specific dosage form (for example tablets) cannot be used in another dosage form (injections). Therefore pharmaceutical excipients specifically used in one specific dosage form cannot be considered to be a suitable substitute for those drug formulations which are intended to be used in other dosage forms (for example tablets, capsules).

1625. Pharmaceutical grade lactose is mainly used as filler and binder in tablet dosages and it may also be used in pharmaceuticals in powder sachets. In syrup, lactose is only used as a sweetener. Pharmaceutical grade lactose is not commonly used, if at all, as an excipient in injection pharmaceuticals, ointments and gels.

16.2.2. Pharmaceutical grade lactose and food grade lactose form two distinct separate relevant product markets

16.2.2.1. No demand side substitution between food grade lactose and pharma grade Lactose

1626. The notifying parties have stated that in some regions of the world the enforcement of the rules on pharmaceuticals is less strict, which would allow some customers to use food grade lactose instead of pharmaceutical grade lactose in pharmaceutical applications. One distributor in India explained that food grade lactose is used in pharmaceuticals in India. (955) However, the fact that, according to the notifying parties, in some jurisdictions it may be possible to use for some pharmaceutical applications food grade of lactose (956), does not provide such a competitive constraint on lactose that one single relevant product market for all lactose could be defined.

1627. It was reported to the Commission that refined food grade lactose or standard food grade lactose generally cannot be used to replace the regulated pharmaceutical grades. Only a few countries, such as Bangladesh, Taiwan and Korea allow unregulated lactose for internal country production, but it is not acceptable for most countries that require the use of regulated grades of pharmaceutical lactose. Therefore customers typically require that pharmaceutical grade lactose must be manufactured in a Good Manufacturing Practice environment, according to the IPEC Guidelines. These guidelines are not applicable to food grade lactose. (957) Unregulated lactose can therefore not be seen as belonging to the same relevant product market as pharmaceutical grade lactose.

1628. The majority of customers, especially those in the EEA, the US and Japan, currently using pharmaceutical grade lactose need an approved and certified quality that can only be guaranteed by pharmaceutical grade lactose. Accordingly, virtually all customers currently using pharmaceutical grade lactose would not switch from to edible grade lactose upon a small but significant non transitory price increase.

1629. The response of customers in the Commission's market investigation was quite univocal in that food grade lactose is not usable in pharmaceutical applications. (958) Several reasons were given, such as issues of quality (especially for pharmaceutical products), more stringent testing of pharmaceutical grade lactose, the extra production steps that pharmaceutical lactose has undergone in comparison with food grade lactose (higher purity). Customers mentioned that pharmaceutical lactose goes through a second crystallisation process that food grade lactose does not, resulting in lower protein and mineral content. It was also mentioned that pharmaceutical grade lactose keeps its white colour for a longer period of time, whereas food grade lactose might decolour when stored for longer periods. Also, a customer pointed out that food grade lactose is less processable than pharmaceutical grade lactose and that food grade lactose is not supplied with the supporting documentation required by the regulatory agencies. For some manufacturing processes the usually higher protein content in edible grade lactose may disturb the manufacturing process of a pharmaceutical. A customer mentioned that it is simply forbidden by law in its country to use food grade lactose in pharmaceutical applications.

16.2.2.2. Supply side substitution effect between food grade and pharma grade lactose is not sufficiently immediate and effective

1630. The notifying parties argued in the Form CO that there would be supply side substitutability between food grade lactose and pharmaceutical grade lactose, because at least two of the secondary steps (milling and sieving) are not only applied in pharmaceutical grade lactose, but also in food grade lactose. In the Form CO the Notifying parties stated that "although some investment by a supplier would be needed, the parties respectfully ask the Commission to reconsider this aspect of their previous assessment". The notifying parties are of the opinion that additional milling and sieving is not necessary for obtaining pharmaceutical grade lactose.

1631. As the procedure evolved, the notifying parties changed their position by stating that if one already produces food grade lactose with high purity, one does not need any additional investments. The notifying parties claimed that even if one already produces normal food grade lactose, a refinery unit need to be bought and set up at a cost of EUR 12 million. The notifying parties have, for the purposes of this merger control proceeding, developed a hypothetical business development plan showing the estimated cost for starting producing pharmaceutical grade lactose when one already produces food grade lactose. In the same hypothetical business plan the notifying parties are of the opinion that for a producer that already markets refined food grade, no additional costs are incurred when starting to produce pharmaceutical grade lactose. (959) In their response to the 6(1)(c) Decision the notifying parties stress that supply side substitutability especially exists between food grade lactose that has already been refined and pharmaceutical grade lactose. (960)

1632. Supply side substitution is taken into account for the purpose of defining relevant markets if the supply substitution effect is sufficiently immediate and effective. (961) That is, if "suppliers are able to switch their production to the relevant products and market them in the short term without incurring significant additional costs or risks in response to small and permanent price changes", the resulting "disciplinary impact on the competitive behaviour of the companies involved" is regarded to be equivalent to the restraining effects of demand substitutability.

1633. Moreover, the Relevant Market Notice mentions a "practical example" of the approach to supply-side substitutability when defining product markets: the example concerns the production of paper, which is usually supplied in a range of different qualities (from standard writing paper to high quality paper). While in this case demand substitutability is low as the different qualities of paper cannot be used for all applications, supply-side substitutability is high as paper plants are technically able to manufacture different qualities and adjustments are possible with negligible costs and in a short time-frame. So if there are no particular difficulties in distribution, manufacturers of different kinds of papers actually compete with each other. In such a case of high supply-side substitutability, all those various qualities of paper would be included in the relevant market. (962)

1634. The notifying parties argue that starting from the basic food grade lactose level only requires the purchase of a refinery unit, which would entail an investment of EUR 12 million. The commissioning and operational set-up of such additional facilities would entail considerable lead time. According to the information that the notifying parties have submitted, the supply substitution effect from normal food grade lactose to pharmaceutical grade lactose is not sufficiently immediate and effective.

1635. Moreover, the market investigation did not confirm the notifying parties' opinion on supply side substitutability between food grade lactose and pharmaceutical grade lactose. The market investigation showed that most competitors would need to make a much larger investment if they were to switch food grade capacity in the direction of pharmaceutical lactose capacity.

1636. Pharmaceutical grade lactose - as opposed to edible grade lactose - often requires characteristics that can only be achieved during different production steps that permit the lactose to function during the processing at the customers manufacturing facilities, such as flowability, dissolution and density and particle size requirements. Therefore, the crystallisation process for pharmaceutical grade lactose is more thorough. This process further reduces traces of protein, vitamins and minerals in the pharmaceutical lactose, for example by double crystallization and further cleaning and filtration with carbon. The specifications for edible grade lactose as identified in, for example, the Food Chemical Codex typically have an upper limit of 0.3% for carbon filtration traces (ash). However, typical pharmaceutical values as required by customers in requests for quotations are, according to a respondent during the Phase I market investigation, 0.02%. Pharmaceutical grade lactose containing any trace amounts of protein or too high a value of protein content is not acceptable for the pharmaceutical industry, due to concerns over allergic reactions and risks of decolourisation of the lactose resulting ultimately in discoloured tablets. Moreover, edible grade lactose can and does contain more riboflavin (vitamin B2) to give it a distinct yellow colour. Again, such colour would be unacceptable for (most) pharmaceutical lactose customers as it would indicate impurities in the product. Lastly, the microbial limits for pharmaceutical grade lactose are significantly lower than those for edible grade lactose. Additional hygiene requirements and procedures would have to be adopted to consistently achieve these limits. (963)

1637. The installation of additional equipment to demineralise and decolour edible grade lactose, in addition to the additional milling and sizing of the particles of pharmaceutical lactose, would require a very large capital investment. Moreover, in those instances where the edible lactose facility processes dairy proteins (edible grade lactose) the production facility may need considerable additional capital investments in order to make it suitable for the production of pharmaceutical grade lactose, because of the need to produce a dairy protein allergen free product. (964)

1638. Starting from the more refined food grade lactose, the market investigation showed that supply substitution is not sufficiently immediate and effective. Contrary to the opinion of the notifying parties that there are no costs involved in a switch from production of refined food grade lactose to pharmaceutical lactose, many respondents to the Commission's questionnaire and subsequent probes explained that in their production facilities, many changes and investments would need to be made.

1639. One supplier of food grade lactose, mentioned by the notifying parties as a particularly good candidate for switching production capacity from food grade to pharmaceutical grade, indicates that it could not easily switch production from one grade to another because it would require investments in equipment and specific know how, especially with regard to running the special equipment. The supplier also indicated that it is not practical to switch a plant back and forth between food and pharmaceutical lactose because of the clean up procedure required for pharmaceutical lactose. A company would have to choose one procedure over another. This supplier mentions that the production process of refined grades lactose requires specific production processes, such as pharmaceutical specific purification process, in order to be marketed as pharmaceutical grade lactose. (965) The supplier therefore disagrees with the notifying parties' statement that it would be relatively easy to switch its food grade capacity to pharmaceutical grade capacity.

1640. Another supplier of food grade lactose, that is not active in pharmaceutical lactose, points out that the pharmaceutical grade lactose industry has quite high barriers to entry, in the form of know how, investments and time consuming approval processes. (966) Such approval processes would - in view of regulations - not render any substitution sufficiently immediate and effective. Unregulated lactose can therefore not be seen as belonging to the same relevant product market as pharmaceutical grade lactose. A supplier that purchases high grades of food grade lactose from third parties and then processes it into pharmaceutical grade lactose explains that pharmaceutical grade lactose needs to undergo further production steps. The processing involves several different processing steps according to this supplier, including carbon filtration, refined crystallization, refined washing and centrifugation just to remove the colour specification. Subsequently, the lactose is further processed through drying technology to create functionality characteristics for tabletting, mills and sieves and even agglomeration processes to create the specific mesh profiles for tabletting properties of the lactose. (967) This shows that even starting from a high grade lactose, additional production steps would be required, according to this supplier.

1641. This was confirmed by another supplier of lactose that estimated the changes that its current food grade lactose facilities would need to undergo in order to convert them from food grade lactose to pharma grade lactose. This particular supplier would need to install additional equipment to decolorize and demineralize the edible lactose. Particle sizing equipment would also be required. In addition, because of the need to have a dairy protein (allergen) free product the equipment train would have to be isolated from the remainder of the facility - requiring in one production facility the construction of a separate addition to the facility or a major reconstruction in another production facility. This particular supplier has not fully analyzed such an investment, but a rough estimate is that it could exceed EUR 70 million. (968)

1642. Therefore, the investment and lead time needed would render any supply substitution effect not sufficiently effective and immediate. (969) It is considered that food grade lactose and pharmaceutical grade lactose form two distinct relevant product markets.

16.2.3. Food grade lactose

1643. Within food grade lactose there are three customer groups that use food grade lactose for their products. The first set of customers is the producers of infant nutrition such as infant formulae, follow up milk and grow up milk. The second set of customers is fine chemical producers that produce lactitol and lactulose which are nutrition supplements. Food generic producers which produce desserts, jams, chocolate, confectionary, bakery etc. are the third set of customers. All customers of food grade lactose have their own specific requirements with regard to the composition of food grade lactose. Mostly, according to the notifying parties (970), these requirements are a combination of purity and microbiology. A higher purity is achieved through a refining process. On this basis, three groups could be distinguished within food grade lactose: (i) infant food lactose (ii) fine chemical lactose and (iii) food generic lactose.

1644. The use of infant nutrition products requires, depending on the product and the way the product is prepared, a higher microbiological quality than required for fine chemical or general food stuff applications. This is especially true when the infant nutrition is produced through dry blending of ingredients. In that particular production process there is no further heating step during the production. Infants are by nature more receptive to microbiological contamination than adults. Therefore, in this particular production process of infant food the microbiological quality of lactose is more critical. The microbobiological quality of the food grade lactose for wet processed infant nutrition products that include a heating step is, as a result of the heating, less critical.

1645. On the other hand, for the second set of customers purity is more important than the microbiological quality, since impurities outside a certain bandwidth can disturb the production process at the customer's production facility. The bandwidth for purity requirements for lactulose is not always consistent. Some customers allow lower purities (up to standard food grade lactose according to the Notifying parties). (971)

1646. From a demand side perspective, the lactose for infant food lactose, fine chemicals lactose and food generic may not be interchangeable due to the difference in microbiological quality and purity.

1647. From a supply side perspective, it should be pointed out that some suppliers of lactose would be able to achieve the microbiological quality and purity needed for all three groups of lactose in one production facility. They would be technically able to switch supply between the various forms and would also be able to market the different lactose types to the different customer groups. Campina itself is an example of this capability. However, according to the Form CO "a lactose producer with only one production line can offer only one type of food grade lactose." (972) This in itself points towards a rather limited supply substitution possibility from a technical point of view. However, there is no need to conclude to what extent a supply side effect would be sufficiently immediate and effective within the meaning of Relevant Markets Notice as even in the narrowest alternative (in particular the market for infant food grade lactose where the merged entity would have its largest share on the three narrow alternatives) the merger would not lead to a significant impediment of effective competition.

16.2.4. Pharma grade lactose

16.2.4.1. Pharmaceutical lactose to be distinguished from other excipients with filler / binder properties

1648. In the Form CO, the notifying parties submitted that substitutability exists between different types of filler binder pharmaceutical excipients. (973) The notifying parties in the Form CO explained that when a pharmaceutical has reached the commercial stage there are barriers to switching in the form of redevelopment costs. In that respect the notifying parties agree with the Commission's assessment in case M.4207 Campina/Fonterra Cooperative Group JV. In the Form CO, the notifying parties however submitted that drug manufacturers in the formulation stage take a broad perspective on the possible filler binders and consider many alternatives. (974) The notifying parties also explained that in the generic stage of a drug, drug manufacturers can choose other filler binders. The notifying parties pointed out in their response to the Article 6(1)(c) Decision that the filler binder is chosen for its filler binder functionality and not for its possible additional functionalities. (975)

1649. The notifying parties' statement that all filler binders can perform the filler and binder functionalities is not contested. (976) Also it is not denied that a formulator can (sometimes) formulate the same drug product using multiple sets of excipients and get the same performance. But substitution must not only exist in theory. It should be recalled that market definition is a tool to identify and define the boundaries of competition between firms. The main purpose of market definition is to identify the competitive constraints that the undertakings involved face. Firms are subject to three main forces or competitive constraints: demand side substitutability, supply side substitutability and potential competition. (977) A firm cannot have a significant impact on the prevailing conditions of sale such as prices if customers are in a position to switch easily to available substitute products. According to the notice on relevant markets, the exercise of market definition consists in identifying the effective alternative sources of supply for the customers of the undertakings involved. (978) As in the Commission's previous merger investigation in the same market (979) the current phase I and phase II investigation have not confirmed that customers can switch easily from lactose to another excipient for the following reasons.

16.2.4.1.1. Lactose has a different set of properties

1650. According to the Commission's investigation, customers rely on certain properties when they make their ultimate choice for a certain filler or binder. The technical possibility to replace lactose with another filler or binder in a certain application depends on a number of properties of lactose. Although they bear the same classification (diluents/filler/binder), different excipients have different properties and will behave differently in the production process depending on type of the process and on which other materials are used in the formulation resulting in different properties to the end product. Each excipient has a different set of properties.

1651. The notifying parties submitted two documents that contain opinions on the functionalities of excipients in general and lactose in particular. (980) The opinion of an employee of the French pharmaceutical company, Servier, describes the selection of ingredients for the formulation of an oral dosage drug as a complex process involving several considerations. According to this opinion, several elements are considered in the selection of ingredients: compressibility for tablet formulations, flow ability, particle size distribution (influencing flow, compression, blending and dust problems), potentially moisture content (depending on the active ingredient), bulk density, compatibility with the active ingredients and stability of the final product. Also, the physiological acceptability (toxicity, osmotic effect, taste and feel), cost and availability of the material and regulatory acceptability play a role.

1652. A number of the excipient's properties may therefore determine the choice for using a certain excipient. Respondents to the Commission's questionnaires confirmed that the specific properties of lactose are important when they choose lactose instead of other filler binder excipients. (981) The market investigation showed that lactose has a number of characteristics that other excipients do not have. Excipients such as microcrystalline cellulose ("MCC"), Di-calcium phosphate ("DCP"), starch and silicified microcrystalline cellulose ("SMCC") are not soluble in water according to the notifying parties' internal documents. A customer confirmed, during the Commission's market investigation, that lactose and MCC are indeed different in that lactose is more soluble than MCC. (982) Moreover, lactose is valued by the notifying parties to have superior mixing and segregation values. According to the notifying parties' documents when comparing mixing and segregation aspects lactose is valued "best Gran & SD". According to the notifying parties' document, MCC, DCP, and Mannitol score lower, respectively[...]*. As to compressibility, starch is considered to score poorly. Strain rate sensitivity of lactose is considered to be low as is DCP. Strain rate sensitivity of MCC, and starch are valued high, those of Mannitol and SMCC are unknown. The lubricant sensitivity of starch as compared to lactose is high. Finally, lactose is valued to have a neutral taste, but DCP "can be gritty". Mannitol and starch are also deemed to have a neutral taste. (983) A customer also explained that the compressibility of lactose is different from other excipients such as MCC. (984)

1653. The technical aspects of lactose and several other excipients are reflected in Table 16-1, taken from the notifying parties' internal documents. The table is part of a Campina presentation, during a workshop for sales people held in Singapore on 14 June 2007 (985):

Table 16-1: Technical aspects of lactose and other excipients - Source: Notifying parties.

1654. The selection of an excipient by a formulator is hence predominantly determined by the final functionality or delivery system of the drug application. Pharmaceutical grade lactose is used for its distinct chemical, physical and functionality properties and generally cannot be substituted into every formulation for another type of excipient such as starch poor MCC. According to a market participant this is clearly supported by what transpired as a result of the shortage of lactose during 2007. While prices increased by an estimated 3 to 5 times over the market prices of previous years, pharmaceutical manufacturers did not substitute other excipients that were available on the market at lower prices, such as starch and MCC based excipients. (986)

1655. All of this evidence shows that although different excipients may be considered for use as a filler binder in pharmaceuticals, the unique properties of each different excipient are generally a deciding factor. The demand substitution is thus not at all universal and not sufficiently effective, not even during the early formulation stage.

16.2.4.1.2. Lactose reacts differently to other ingredients

1656. Further elements that customers take into account when choosing for either lactose or other excipients is stability in the end formulation. One excipient may be more stable in a certain composition of a drug than another. In a certain formulation the choice of a drug manufacturer for lactose would also depend on the way in which the lactose behaves together with the active ingredients in the formulation. Sometimes other excipients are less suitable to work together with the active ingredients from the drug. In that case, lactose is the preferred choice in a formulation.

1657. Excipients are also chosen on the basis of their compatibility with the other excipients in a certain drug. Some drug formulations are made using more than one excipient for the same drug. In that case the excipients are being chosen by way of how well they work together in a certain drug. The Commission's market investigation has shown that this is an important motivation in choosing a certain type of excipient - lactose for example - over another type of excipient for certain customers. (987)

1658. The way in which an excipient behaves in general (and lactose in particular) in a certain formulation appears to at least co-determine the choice for the use of a certain excipient. This shows that alternatives might not be equally suitable for all groups of customers and other excipients are not such close substitutes as to exercise a real constraint on market power. Therefore, again, the fact that lactose and other excipients could technically be replaced in some formulations, does not mean that there is demand side substitution.

16.2.4.1.3. Pricing

1659. Notwithstanding the fact that for some formulations it appears to be technically possible at the formulation stage to use another excipient than lactose (988), there are also commercial considerations that customers take into account when they choose lactose over another excipient. Although the cost of excipients is not the key driver related to costs of formulation (as excipients make up only for a small part of the total costs) some respondents have indicated that lactose is one of the main excipients used as a diluent or filler binder due to its relatively low cost. (989)

1660. Respondents to the Commission's questionnaire have indicated that (mostly in early formulation stage) in some cases, it would be technically possible to substitute lactose with a different type of excipient such as MCC or starch if they would reformulate the drug involved. (990) However, the question is not purely whether customers that are sourcing lactose would be able to reformulate the recipes of the drugs and formulate around lactose. For when considering demand side substitution not only the question as to the functionality but also commercial considerations should be taken into account.

1661. The notifying parties have submitted the prices of MCC, starches, Di-calcium phosphate and Mannitol relative to pharmaceutical grade lactose.(991) When looking at the prices it should be borne in mind that the calcium products are in terms of technical supply side substitution possibilities rather remote from lactose. Its functionalities are valued by Friesland Foods [..]* points lower overall if compared to lactose as is reflected in the table below. Still, within the EEA, DCP is almost [..]*% more expensive than pharmaceutical lactose and it is more than [..]*% more expensive in the US.

1662. The Table 16-2 is provided by Friesland Foods in which it compares all filler binders and give numbers ranging form 1 - 5 (poorest to best) to value the properties of all filler binders.(992)

Table 16-2: Comparison of filler binders.

1663. As to Mannitol, this excipient is mostly used for chewable or ODT tablets due to its organeoleptic properties. Mannitol is particularly suitable for delivering drugs to the brain, since it opens the blood brain barrier. It was pointed out to the Commission that Mannitol is so expensive because of the severe shortage of maize, which is caused by the usage of that product for the production of biofuels. (993)

1664. The prices for the various excipients are, according to the notifying parties, as follows:

EEA (994)

Pharma lactose

MCC

Starches

Di-calcium phosphate

Mannitol

Table 16-3: Prices of various excipients, EEA - Source: Form CO.

1665. The question to be answered is whether the notifying parties' customers would switch to readily available substitutes in response to a hypothetical small but non-transitory increase of price. In the EEA and the Americas, the price differences between pharmaceutical lactose and any other excipients are more than 10%. According to the notifying parties only in Asia starches are already more than 20% cheaper than lactose. On the question if a customer would switch from lactose based excipient to any other excipient should the price for lactose bases excipients increase by 5 - 10% (other excipients remaining stable in price) almost all respondents answered no.(995) (996)

16.2.4.1.4. Substituting lactose with another excipient at commercial and generic stage would result in high cost

1666. On top of the price difference in pharmaceutical lactose and other excipients that may play a role at the formulation stage, customers do not switch to another type of excipient after the first formulation stage during the commercial period of a drug, even in case of a small but significant non-transitory price increase. Switching barriers are highest in the commercial stage (branded or generic), because the cost of reformulation is extremely high and sometimes prohibitive. This holds particularly true for prescription or registered drugs, because health authorities require a large amount of testing of the reformulated finished dosage form even when changes are slight, and in addition, approval is not always guaranteed. As such, "The regulatory barriers are the driving force and can substantially offset the benefits of lower cost of goods from using alternative excipient systems." (997) In cases where switching would be technically possible, switching to another type of excipient would be too costly or the procedure would be too lengthy to make it worthwhile. (998) This is mainly due to the additional costs imposed by any change of the formulation of a drug which - among other things - also include additional costs for testing and necessary administrative procedures (approval, registration). (999)

1667. According to an opinion of Finn Britt consulting that was submitted by the notifying Parties (1000), justifying the costs associated with switching from lactose to another excipient would require savings that approach EUR 150 000 per year per product (or groups of products) to offset the costs of extra testing and regulatory application and testing. Finn Brit Consulting continues: 'This is not to say that the cost of all those studies is 150.000 Euro, it is certainly more!" According to this opinion, it is assumed that the cost would only be offset over a period of three years, rendering it easier to justify formulations for innovator products than for generics.

1668. However, during the market investigation both generic and innovator producers have advised to be wary to change formulations. (1001) Since excipients only represent a rather small share of the total cost of the production of a drug pharmaceutical companies are very reluctant to change the excipients chosen in the formulation stage during the commercial stage of a drug. The reason some respondents have indicated that it is not possible to substitute the excipient during the commercial stage is the regulatory approval of drugs. Once a drug is registered it is virtually impossible to change its excipients formulation. (1002) It would require significant cost and time especially with regard to stability data, bioequivalence and change of dossier. (1003) Even before the commercial stage of a drug a switch from lactose to another excipient could delay marketing applications by 12 months or more. (1004) It could constitute considerable process re-engineering with subsequent validation and would hence not be commercially viable.

1669. These barriers and costs associated with switching (1005) are an additional element that prevents considering lactose and other filler binders as belonging to a single relevant product market.

16.2.4.1.5. Conclusion

1670. The market investigation has shown that customers' choice for lactose is based on its properties, the interaction with the other ingredients of a drug, its favourable price and significant barriers and costs associated with switching. (1006)

1671. For the purpose of defining the relevant product market, it must be concluded that the customers do not have an effective alternative source of supply in the producers of MCC, starch, Mannitol and other filler binder excipients. (1007) The fact that products are capable of performing the same function - to a certain extent - does not necessarily mean that they form part of the same relevant product market. A test based on pure functionality would often produce the wrong result. The question is not simply whether, if one product were not available, alternatives would be found which could serve the same purpose. The alternatives are not equally suitable for all groups of customers and are not close enough substitutes as to exercise a real constraint on market power. They do not allow for effective competition. (1008)

1672. It is therefore concluded that the excipients such as starch, Mannitol, MCC are not effective sources of alternative supply for the customers and cannot, therefore, be considered a competitive constraint within the meaning of the Relevant Markets Notice.

16.2.4.2. Pharmaceutical lactose: wet granulation and direct compression

1673. There are a variety of forms of pharmaceutical lactose reflecting the various processes applied to lactose. Examples of primary processes are milling and sieving while more advanced processes consist of roller drying and spray drying. Lactose is referred to by the process immediately applied to it, for example "milled and sieved" lactose or "roller dried lactose". Each of these processes create a number of end products - the process of milling and sieving, for example, separates the lactose particles into different size groupings and the resulting end products are referred to by the particle size, for example "milled lactose with a mesh size of 150".

1674. Direct compression lactose means that the form of lactose must be suitable for the direct compression process. Spray dried lactose, roller dried lactose and agglomerated lactose are all typically suitable for direct compression. All three lactoses are pharmaceutical lactoses but produced with different production steps. Spray dried lactose is for example, contrary to roller dried lactose, dried in a spraying tower. The choice by a drug manufacturer over whether to use wet granulation, direct compression or any other method involved in tablet manufacture depends on a number of factors including whether it suits the active ingredient of the drug in question. Wet granulation is chosen for drugs that cannot be compressed into tablet form (because the active ingredient has a very high or a very low dose for example). Direct compression is generally the preferred method, when the active ingredient allows it, largely because the cost-inuse for the drug manufacturer is lower. This is because making a tablet using the direct compression method involves fewer steps than wet granulation and results in reduced labour easier processing and lower capital cost. However, wet granulation is still the most common form and the majority of tablets are produced using this process, according to the notifying parties. In Europe, Japan and Asia it is estimated that about 80% to 85% of tablets are still produced by wet granulation. Only in the US is there already an equal split between wet granulation and direct compression.

1675. While the notifying parties argue that there is a gradual shift from wet granulated lactose to direct compression lactose - there is limited demand side substitutability between wet granulation lactose and direct compression lactose. Customers who are not equipped for direct compression lactose have to use wet granulation lactose and, for many forms of tablets, wet granulation is not possible (high doses or low doses tablets). This is underlined by the fact that the preferred and cheaper application is direct compression, but the majority of customers are still applying wet granulation.

1676. The notifying parties underlined the fact that supply side substitutability exists between direct compression pharmaceutical grade lactose and wet granulation pharmaceutical grade lactose. To that end, the notifying parties claimed that all suppliers of direct compression lactose can, and do, produce these types of pharmaceutical lactose. This in itself is no reason to conclude that there would be supply side substitution between wet granulation pharmaceutical grade lactose and direct compression pharmaceutical grade lactose within the meaning of the Relevant Markets Notice. Although it is true that the only producers of direct compression pharmaceutical grade lactose, Friesland Foods, Campina, Meggle and KerryForemost, all in addition produce wet granulation pharmaceutical grade lactose, supply side substitution is far from universal. This does not show that producers of wet granulation pharmaceutical grade lactose could sufficiently undergo an immediate and effective switch to the production of direct compression grade lactose.

1677. However, as the transaction would not lead to competition concerns on the market for pharmaceutical lactose nor on any narrower alternative, there is no need to conclude on the supply and demand side substitutability between direct compression pharmaceutical grade lactose and wet granulation pharmaceutical lactose.

16.2.5. DPI (inhalation) grade lactose is a separate relevant product market

16.2.5.1. No demand side substitution

1678. Dry powder inhalation is the specific application of a very fine lactose powder in drug formulations destined to be inhaled by the patient. The notifying parties claimed in the Form CO that lactose for the use in dry powder inhalation is so specific - in particular from demand side perspective - that at first sight it might be considered a distinct product market. In the Form CO, the notifying parties also claimed that all suppliers of pharmaceutical grade lactose in principle offer DPI. (1009)

1679. DPI lactose is significantly more expensive than other pharmaceutical grade of lactose. From a demand side perspective, only DPI lactose can be used for the application in drugs with lactose excipients that are administered through inhalation. The market investigation showed that it is technically impossible to substitute DPI lactose with another type of lactose. Therefore, from a demand side perspective there is no substitutability between DPI pharmaceutical lactose and other types of pharmaceutical lactose. (1010)

1680. The notifying parties agree in part that from a demand side perspective, DPI does not compete with wet granulation or direct compression pharmaceutical lactose.(1011)

1681. On 26 September 2008, the notifying parties submitted that within DPI pharmaceutical grade lactose, a more sophisticated product can be distinguished.

1682. The DPI grade that the Kapuni plant and the Friesland Foods plant produce is much more refined than the other DPI grades lactose. Both plants in which these grades are produced are purpose-built for one particular customer, with the help and know-how of this particular customer. The DPI that can be produced in the Friesland Foods production facility and in the Kapuni facility of Campina is in particular more refined in that higher consistency in the particle size of the lactose can be achieved than the consistency that would be required by most customers. This is due to several extra production steps [...]* Outside the Kapuni plant Campina can not achieve high consistency in DPI, due to the fact that these extra production steps are lacking.

1683. The notifying parties contend that the plant of Friesland Foods produces a much more sophisticated product than the DFE plant, whereas many of the current DFE customers do not require such a sophisticated product. The main difference is, as described above, that the particle size distribution within the product is much more consistent and that the product in that regard is more refined and has a higher quality. The customers for high sophisticated DPI need this kind of DPI because of the application in their end-product. Very often these are special and customised devices for delivering drugs to the lungs where the characteristics of the high sophisticated lactose (i.e. the high consistency) are absolutely vital. High consistency lactose is in general much more expensive and a 5% price increase would not make customers switch to DPI lactoses that have a lower consistency, because they form not an alternative for the drug that the customer is producing or is developing. As a result, from a demand perspective, customers cannot substitute the high consistency DPI lactose with the less sophisticated DPI lactose.

1684. This is further evidenced by the fact that sales to other customers than the particular customer mentioned in recital 1682 by Friesland Foods (from its European production facility) are made at an average price of EUR [...]* and sales to others than this particular customer by Campina (from its European production facilities) are made at an average price of EUR [..]*. This further demonstrates that the lower consistency DPI does not form an alternative for customers that need high consistency DPI. In this particular case, this shows that the sophisticated DPI lactose and more simple DPI lactose are a different proposition.

1685. According to simple computations made by the Commission, prices per customer vary significantly for customers for Friesland Foods and Campina, even excluding the one particular customer. This may be an indication that the price of DPI lactose depends in part on the usage of this input and the extent to which the customers are able to substitute DPI with other inputs or producers. The prices for sales of DPI lactose by Friesland Foods European production facility also appear to be significantly higher in comparison to sales by Campina's European production facility, which indicates that DPI lactose of Friesland Foods is of higher grade than that of Campina.

1686. The notifying parties state that the more sophisticated DPI lactose is produced in a different production process, making use of state-of-the-art equipment [..]* resulting in a product with a very high consistency in particle size distribution. The notifying parties explain that from a demand side perspective the interchangeability between both types is likely to be low.

1687. For the same reasons, supply side substitution is low: the production facilities of Campina [..]* can not achieve the high consistency that is required for the sophisticated lactose comparable to the standards of the Kapuni plant or the Friesland Foods plant. In order to switch at the supply side from less sophisticated DPI lactose to more sophisticated DPI lactose with high consistency, significant investments would need to be made. Apart from the investment, know how would also be required and a long lead time would occur before a DPI facility capable of producing the less sophisticated lactose only would be capable of producing the high sophisticated lactose. This would render any supply effect in this direction not sufficiently immediate and effective. (1012).

1688. Therefore, separate relevant markets should be distinguished for DPI with a very high consistency and less sophisticated DPI lactose with a lower consistency.

16.2.5.2. Supply side substitution effect between (solid) pharmaceutical grade lactose and DPI pharmaceutical grade lactose is not sufficiently immediate and effective

1689. The notifying parties have submitted that from a supply side perspective it is easy to start producing lactose for DPI applications if one already produces lactose for other applications. According to the notifying parties no large investments are needed. The notifying parties have mentioned in the Form CO that customers may be willing to contribute to the investments needed to build DPI production facilities.

1690. However the market investigation showed that substitution from pharmaceutical grades of lactose (in particular wet granulation and direct compression) in the direction of DPI lactose would require considerable lead time and investment (1013), especially small or multiple particle sizes are required for DPI. Not only are the particle sizes different but so are other specifications for DPI lactose. Notably, the microbial limits are more stringent than for solid dose lactose. Therefore more stringent hygienic manufacturing procedures and conditions are necessary than for solid pharmaceutical lactose.

1691. One supplier of pharmaceutical grade lactose introduced some grades of DPI lactose using traditional manufacturing technology but, according to this supplier, the production of DPI lactose that adequately meets the customers' expectations and the specifications proposed by IPEC segregation of production is required. This supplier submits that a separate dedicated production line with capabilities for classification and micronisation of particle sizes is needed as well as a dedicated, nearly aseptic packaging line. (1014) These extra production steps require considerable investments in the production facilities.

1692. The supplier also points out that in addition to manufacturing expansion, investments in applications equipment and R&D (equipment as well as research and analytical scientific capabilities) would be needed. (1015). This is confirmed by another supplier of pharmaceutical grade lactose. (1016) Moreover, from a regulatory point of view, DPI lactose contrary to other pharmaceutical grade lactose (non active ingredient in pharmaceuticals) requires the same (higher) standard of regulatory approval as active ingredients in pharmaceuticals. The regulatory process is therefore more complex than for the rest of pharmaceutical lactose grades. Therefore, regulatory approval and compliance is much more critical. It was also pointed out during the market investigation that the supply substitution would not be effective and immediate since it was assumed that customers would not switch any time soon, due to high switching cost and regulatory approval (1017).

1693. The investment and lead time needed would therefore render any supply substitution effect not sufficiently effective and immediate. (1018).

16.2.5.3. Conclusion

1694. A separate relevant product market should therefore be defined for DPI lactose. Within this market, a distinction must be made between standard DPI lactose and the sophisticated DPI lactose with very high consistency.

16.3. Relevant geographic market

16.3.1. Food grade lactose

1695. The notifying parties have stated that the relevant geographic market for food grade lactose should be defined as being worldwide in scope. The notifying parties explained that there are significant trade flows and that transport cost are relatively limited. The trade flows that the notifying parties have referred to are predominantly between the US and Asia. With regard to the relevant geographic market for food grade lactose the market investigation has indicated that the markets for food grades lactose are likely to be EEA-wide in scope. (1019) Customers answered that they source within the EEA, although recently some US suppliers have started to deliver in the EEA. The market investigation indicated that import duties on food grade lactose would be a constraining factor as to the competitiveness of imported food grade lactose. This effect is likely to be stronger for food grade lactose that is less refined since the added value of such product is lower.

1696. It is, however, not necessary to conclude on the relevant market definition since no competition concerns arise irrespective of relevant geographic market definition in food grade lactose in general.

16.3.2. Pharmaceutical and DPI lactose

1697. The notifying parties argue that the markets for pharmaceutical lactose and DPI lactose are worldwide due to global sourcing by customers, low transport costs and comparable prices. The market investigation has confirmed that the relevant geographic market for pharmaceutical lactose and for DPI pharmaceutical lactose is at least EEA-wide. (1020) However, for pharmaceutical lactose, the market is much more regulated in Europe, USA and Japan than in the rest of the world

1698. The main excipients and lactose suppliers sell at least EEA-wide or world-wide and endusers of excipients often buy on an at least EEA-wide or even world-wide scale. Purchasers that normally source from suppliers located in the EEA generally indicated that they would source from another region if a small but significant non transitory price increase would occur. Furthermore, a majority of purchasers source from different regions in order to be less exposed to the risk of health and safety related supply problems, such as "mad cow"-disease. However, some customers reported that they were wary of importing pharma lactose from outside the EEA because of the additional transport cost and expected regulatory, quality and delivery problems. (1021)

1699. As explained below, the position of the combined entity on the worldwide market, for both pharmaceutical grade lactose and DPI pharmaceutical lactose, would be virtually the same as on an EEA-wide market. The transaction would not significantly impede effective competition on the worldwide markets and EEA-wide markets for pharmaceutical and DPI lactose, irrespectively of the exact definition of the geographic scope of the markets.

16.4. Competitive assessment

16.4.1. The market investigation

1700. In their response to the Statement of Objections, the notifying parties claim that the Commission's market investigation has not been sufficiently thorough.

1701. The Commission's market investigation was very extensive and has also been thorough in relation to the lactose markets. In the market investigation, questionnaires were sent to customers and competitors of food grade and pharmaceutical grade lactose producers alike. The Commission contacted customers and competitors in each phase with specifically targeted questions and has cross-checked the replies of the competitors with the information provided by the notifying parties. The notifying parties have also provided a number of documents that further informed the Commission during its market investigation. It was avoided relying on one particular source of information or on the information of one particular respondent.

1702. In its response to the Article 6(1)(c) Decision the notifying parties stressed that the possibilities of substitution between food grade lactose and pharmaceutical grade lactose would be more important than it had been taken into account in its Decision. The notifying parties also stated that customers could switch more easily between pharmaceutical grade lactose and other filler binders than it had been found in its first phase market investigation. Therefore the second phase market investigation focussed (with regard to the demand side) on the substitution between food grade and pharmaceutical grade lactose and on the substitution between pharmaceutical grade lactose and other filler binders. To that end, a number of individual customers were contacted with targeted follow up questions on these specific questions. The Commission, in its market investigation also informed itself on the basis of internal documentation of the notifying parties and on opinions of participants in the industry that were provided by the notifying parties.

1703. Moreover, the notifying parties stated that the supply side substitution effect between food grade lactose and pharmaceutical lactose was underestimated by the Commission. The notifying parties also stated that entry and expansion possibilities and available capacities of current pharmaceutical lactose suppliers were underestimated by the Commission. Because these elements are highly important for the overall assessment of the effects of the transaction on the markets for lactose the suppliers of pharmaceutical lactose (actual and potential) were targeted with detailed questionnaires on the supply side issues in phase II. These were subsequently followed up with more detailed specific questions in which the prior replies of respondents were tested and cross-checked.

1704. Furthermore, the Commission relied on information from the notifying parties both submitted at the request of the Commission and submitted at the notifying parties own initiatives.

1705. On 22 and 23 November 2008, the Commission received statements from the notifying parties that their extrapolated sales of pharmaceutical lactose would decrease in 2008. This triggered the Commission to check the extrapolated sales data of the main competitors again and to re-check the total market size.

16.4.2. Food grade lactose

1706. The total EEA market is 270 900 Mt according to the notifying parties. The combined output volume of the notifying parties would be as much as [...]* Mt. Although the merger would lead to the merged entity being a clear market leader in the EEA with some distance from other suppliers, the combined output would correspond to a market share of [20-30]*%, according to the Notifying parties.

Table 16-4: Source Form CO, worldwide market shares based on volume in the year 2007

Table 16-5: Source Form CO, EEA market shares based on volume in the year 2007

1707. In the market for food grade lactose a number of large suppliers are active, besides the notifying parties.

1708. Fonterra Co-operative Group Ltd from New Zealand is a very large dairy cooperative with an estimated intake of 14 billion litres of raw milk per year. Fonterra is a significant player on the market for food grade lactose and is according to the Notifying parties estimated to have a production of [...]* Mt of food grade lactose a year.

1709. Glanbia from Ireland processes almost 5 billion litres of raw milk and has a total food grade lactose production of [...]* Mt a year. Glanbia has operations in Ireland, Europe and the US.

1710. Hilmar Cheese Company from the US has a very large integrated cheese and whey protein plant and manufactures a number of dairy products. Hilmar is a large player in the market for food grade lactose with an estimated food grade lactose production of [...]* Mt a year, according to the notifying parties.

1711. Leprino from the US has a number of production facilities and is believed to sell approximately [....]* Mt of food grade lactoses each year. Leprino has also production facilities in Europe. The notifying parties have indicated that Leprino plans to further expand its capacity of food grade lactose to [....]* Mt a year.

1712. Saputo from Canada is also a large dairy company that is active in the production of food grade lactose. Saputo has a mixed product portfolio and specialises in bakery products and cheeses. The notifying parties estimate that Saputo produces [....]* Mt of food grade lactose per year.

1713. Arla Foods Group from Sweden has a large capacity in food grade lactose production with sales of [....]* Mt food grade lactose annually.

1714. Besides those very large suppliers there are also a number of large suppliers of food grade lactose, such as the Lactalis Group from France and Sachsenmilch from Germany each produce [....]* Mt of food grade lactose and a number of smaller suppliers.

1715. The Table 16-6 shows which suppliers of food grade lactose are active in the respective segments of the market.

Table 16-6: Activities of suppliers of food grade lactose in different segments - Source: Form CO, market investigation.

1716. As can be seen from the Table 16-6, the market for food grade lactose is not concentrated. Should the market be defined more narrowly, the number of players in refined food grade lactose would bemore limited, although even there - apart from the merged entity - a number of players would remain active. Moreover, to the extent that from a demand perspective the relevant product market should be defined more narrowly, there would remain also a competitive constraint from the food generic lactose producers. It is for example known that quality requirements of lactulose producers are not always consistent. For instance Fresenius procures food grade lactose with standard specifications for purity and microbiological quality while its direct competitor Solvay has stricter purity specifications. Finally, and most importantly, the share of Friesland Foods in the segment for refined food grade lactose is not significant, whereas at the same time Friesland Foods is not reported to be a driving force of competition in the market.

1717. With regard to the market for infant food lactose, the Commission received during its market investigation a number of critical comments. A number of customers indicated that the merger would likely lead to a loss of competition on the market for infant food grade lactose. Campina and Friesland Foods may be close competitors and the only realistic set of supply alternatives for specific customer groups. It was even indicated that the notifying companies would be the only suppliers able to meet the specific product requirements of a specific customer. It was also indicated that the merger would lead to dominance on this market and would negatively affect prices. Switching to other sources, outside the EEA, would be too expensive, due to import duties

1718. However, on the market for infant food grade lactose, the market share of the notifying parties would not be high, even on an EEA market. Apart form the merged entity a number of other suppliers would remain active to source from. Customers such as Mead Johnson, Abott, Weyth and Nestle would not be dependent on the merged entity as the source where they could purchase infant food grade lactose. The requirements that they would need for their purchases of infant food grade lactose could be sourced from various sources. For this reason the transaction would not give the merged entity the possibility to significantly impede effective competition by increased prices.

1719. Moreover, in the market for food grade lactose in general there would remain a large spare capacity post merger. Should the merged entity decide to reduce output, thereby trying to increase prices, sufficient spare capacity would be available to offset the negative effects for customers. In other words, it would not be a viable for the merged entity to embark on such strategy in view of the size of its competitors and the spare capacity that exists in the market. The market investigation of the Commission confirmed the existence of such spare capacity for food grade lactose.

16.4.2.1. Conclusion

1720. A sufficient number of large suppliers of food grade lactose would remain active in the market, constraining the merged entity on the market for food grade lactose in general. There are a high number of suppliers that customers can source from. According to the market investigation, a number of suppliers would have spare capacity to expand production volume if the merged entity would try to increase its prices by reducing output or by any other means. For these reasons, the concentration is not likely to lead to a significant impediment of effective competition in the food grade lactose in the EEA or worldwide, even if the product market were to be defined more narrowly.

16.4.3. Pharmaceutical grade lactose

1721. The concerns expressed by market players and the arguments put forward by the notifying parties have been carefully assessed and the conclusion is that the proposed transaction would not significantly impede effective competition on the market for the production and sale of pharmaceutical lactose for several reasons.

1722. First, in markets where market prices are decided by output decisions, and more generally in the markets where firms supply products and/or services which are not perfectly substitutable, the incentives to compete are related to the structure of the industry, particularly to the number of firms. The merged entity could have an incentive to reduce output - diverting its whey to other uses such as for example the production of infant food grade lactose or other uses- thereby increasing its prices for pharmaceutical grade lactose.

1723. A number of respondents to the questionnaire indeed expected price increases on the market for pharmaceutical lactose. Other customers and distributors believed that prices would not increase.

1724. The notifying parties argue that even with a high market share in a very concentrated market, they would not have an incentive to increase prices. The notifying parties state that because the margins on the product are high, a loss of a relatively small volume to a competitor would already offset a price increase of 5 to 10 %, thus rendering such a strategy unprofitable. The notifying parties argue that this inability to increase prices is underwritten by a critical loss study that was submitted in the Form CO. (1022) The report argues that a price increase of 5 or 10 % will not be profitable. This is because a price increase of 5% in Europe would lead to a loss of close to [...]* tons or around [5-10]*% of the combined notifying parties' sales in the regulated markets (and a little over [....]* tons or [10-20]*% in case of a 10% price increase). The report concludes that it is unlikely that a post-merger price increase would be sustainable, even though the combined share of the notifying parties on a market for pharmaceutical lactose is relatively high.

1725. This conclusion can of itself not be accepted. Typically there are two steps in performing a critical loss analysis. First, the critical loss amount is estimated for a range of hypothesised price increases. Second, the actual loss that would result for that range of price increases must be estimated. The notifying parties do not attempt to calculate the actual loss, for example by using empirical tools such as estimating demand elasticities, import origin analysis, shock analysis, customer reactions or switching surveys. The report merely states that it is very likely that competitors are perfectly able to accommodate the critical loss in view of the homogeneous nature of the relevant product as well as the fact that many customers have a policy of dual sourcing. However, the notifying parties provide no evidence as to the ability of competitors to expand capacity.

16.4.3.1. Merged entity would have high market shares on worldwide market for pharmaceutical lactose

1726. The worldwide market for pharmaceutical lactose that has a EUR 314 million total market size according to the notifying parties, has few suppliers. Established producers of pharmaceutical lactose in Europe are DMV (a joint venture of Campina and Fonterra), Domo (a fully owned subsidiary of Friesland Foods) and Meggle. Alpavit from Europe also markets some minor volumes of pharmaceutical lactose. Outside Europe Kerry and Foremost are the largest producers of pharmaceutical lactose. The lactose of Foremost is marketed by Kerry and KerryForemost. These entities are referred to as one supplier, when it comes to pharmaceutical lactose. The market investigation indicated that Hilmar also started to bring some volumes of certain grades of pharmaceutical lactose to the market. Hilmar produces only wet granulation types of pharmaceutical lactose according to the notifying parties, but does not have at present the capability to produce popular direct compression types of pharmaceutical lactose. Meggle, Campina, Friesland Foods and KerryForemost each produce all types of pharmaceutical lactose except for DPI inhalation grade.

1727. Meggle is a family owned, German dairy company, with its headquarters in Wasserburg Germany. Meggle supplies various dairy products in a number of Member States and in the rest of the world. Meggle had an output of approximately 170 million litres of dairy products in 2007. Meggle produces pharmaceutical grade lactose and food grade lactose. As regards pharmaceutical lactose, Meggle sells a full portfolio of pharmaceutical grade lactose, according to the notifying parties, including milled, sieved spray dried, granulation and inhalation lactose. According to the notifying parties Meggle currently sells [....]* Mt of pharmaceutical lactose.

1728. KerryForemost from the United States is also active in pharmaceutical grade lactose. Kerry produces pharmaceutical lactose and also markets the pharmaceutical lactose that is produced by Foremost. Kerry exclusively handles the marketing and sales of Foremost. According to the notifying parties, Foremost produces pharmaceutical grade lactose including milled sieved and spray dried. Kerry produces other grades of pharmaceutical lactose, so that together KerryForemost is able to offer milled, sieved, spray dried and roller dried lactoses. KerryForemost according to the notifying parties currently produces [....]* Mt of pharmaceutical lactose.

1729. Some smaller fringe players are also present in the market for pharmaceutical lactose. Two of these are Hilmar and Alpavit. Alpavit currently markets limited volumes of pharmaceutical lactose. Alpavit is a German dairy producer, part of the Hofmeister Champignon Group. Alpavit only produces milled and sieved grades of pharmaceutical lactose in not very large volumes. Hilmar started the production of one specific type of pharmaceutical lactose.

1730. The market investigation did not confirm that suppliers such as Lactalis, Danone, Leprino, Saputo, Glanbia or Arla are active on the market for pharmaceutical lactose. (1023)

1731. Campina has in its plant in Veghel in the Netherlands [...]* Mt capacity for pharmaceutical lactose of which currently [....]* Mt is used for the production of food grade lactose. Campina sold [...]* Mt of pharmaceutical lactose from this plant. Campina also holds a 50% interest in DFE a full function joint venture between Campina and Fonterra from New Zealand. In Norten Hardenberg DFE has a production facility that has a capacity of [...]* Mt of which currently [....]* Mt is sold. It should be mentioned that Norten Hardenberg sources its lactose from the Veghel plant. Friesland Foods has a production plant in the Netherlands, in Borculo. Friesland Foods has a capacity of [...]* Mt for the production of pharmaceutical lactose. [...]* Mt is sold as pharmaceutical lactose and [...]* Mt is sold as food grade lactose.

1732. For pharmaceutical lactose, the leading pharmaceutical producer and number 3 in the market would be merging under the proposed agreement. The proposed transaction would in fact lead to a 4 to 3 merger at a worldwide level, 3 to 2 at EEA level.

1733. The market is prior to the merger concentrated with HHI levels of [2000-3000]* measured by market shares on worldwide level. Campina Friesland Foods, Meggle and KerryForemost together control [80-90]*% of the global market for pharmaceutical lactose, the remainder of the market ([10-20]*%) being served by fringe players (based on 2007 figures). Through the transaction, the leading supplier will further strengthen its leading position having an increased market share of [40-50]*% on a worldwide market for pharmaceutical lactose, based on 2008 figures. Post merger the concentration levels expressed by the HHI index would be above the [3000-4000]*, the delta being more than [500-600]*. However, as is explained below in this particular case the high market shares are not in themselves indicative of the power of the merged entity to increase its prices by a reduction of its output volumes.

1734. The market shares of the merged entity and its competitors in the year 2007 would, according to the Form CO be as follows.

Table 16-7: Market shares pharma lactose 2007, world-wide - Source: Form CO.

1735. The market investigation has however shown that the notifying parties have overestimated the volumes of their main competitors and the total market size. This leads to an underestimation of the notifying parties own market shares.

1736. The market investigation has shown that competitors other than Campina and Friesland Foods have together sold [....]* Mt during 2007. (1024) On that basis, the total market size for 2007 was 113 000 to 118 000 Mt. This is approximately in line with BCC market research report "Excipients in pharmaceuticals" of Norma Corbitt from BCC research from Washington that estimates the total market size for global lactose excipients for 2007 at 118 000 Mt. This would result in a market share based on volumes of [50%-60]*% for the merged entity based on 2007 figures.

1737. On 22 and 23 October 2008, the notifying parties submitted statements that their extrapolated sales of pharmaceutical lactose would decrease in 2008. This triggered the Commission to check again with the main competitors their extrapolated sales data and to re-check the total market size.

1738. Campina submitted that it would supply in 2008 [....]* Mt and Friesland Foods submitted that it would supply [....]* Mt in 2008. The notifying parties argued that they faced declining market shares over a longer period. It is correct that Campina had less sales in 2007 and 2008 than the years before. The shares of Friesland Foods have been rather stable over the period 2004 to 2007, but decreased afterwards.

1739. With regard to the total market size, the Commission assumes a growth for pharmaceutical lactose of approximately 5%. (1025) On the basis of this assumption the market shares of the notifying parties have been decreasing to a level of [40-50]*% over 2008. This is indeed a significant decrease compared to the levels prior to 2008. In this regard it can be concluded that the notifying parties have lost sales to their competitors in 2008.

1740. During 2008, the extrapolated sales volume of main and fringe competitors to Campina and Friesland Foods that the Commission could identify during its market investigation in the market for pharmaceutical lactose was [...]* Mt. The notifying parties claim that there was a supply of [...]* Mt - representing 25% of the total market - of pharmaceutical lactose from other suppliers, but this was not confirmed by the Commission's market investigation and is significantly overstating the presence of other suppliers of pharmaceutical grade lactose in Asia or elsewhere, such as Huamao in China or producers in India. These suppliers have not been identified by the Commission. During the market investigation customers did not report to source from those suppliers.

1741. However, further information received from third parties after the hearing and after the latest and final submission of the notifying parties has shown that competitors of Campina and Friesland Foods have increased their volumes, thereby increasing their market share. (1026)

16.4.3.2. Potential capacity expansion from suppliers currently active in the supply of pharmaceutical grade lactose

1742. For entry to be considered sufficient competitive constraint on the notifying parties, it must be shown to be likely, timely and sufficient to deter or defeat any potential anti-competitive effects of the merger. Relatively high barriers to entry because of know-how, intellectual property rights, access to raw materials and very high investment costs make Greenfield entry unlikely.

1743. Greenfield entry is costly and time consuming according to the notifying parties. The notifying parties have mentioned that Greenfield entry would entail an investment of around EUR [..]* million. Due to the high capital investment it would entail, deciding to pursue a significant increase in pharmaceutical lactose would not be done relatively quickly. The notifying parties have claimed that there is no need for regulatory approval for building a pharmaceutical lactose plant. However, the notifying parties themselves have admitted that regulatory approval and customer validation would be necessary for expanding their own facilities.

1744. With regard to the production of pharmaceutical lactose the notifying parties have submitted that entry is most likely to come from suppliers already active in food lactose since they have production facilities that could be relatively easily converted from the production of food grade lactose to the production of pharmaceutical grade lactose.

1745. Notwithstanding the fact that the market investigation has shown that this supply side substitution effect is not sufficiently effective and immediate in order to lead to the conclusion that food grade and pharmaceutical grade lactose could be considered as one relevant product market, it has been investigated whether such potential competition would likely sufficiently constrain the merged entity.

1746. Suppliers of lactose have indicated for various reasons that entry barriers to producing pharmaceutical lactose are high. Those suppliers pointed out investment cost in production, regulatory rules, high standards for manufacturing, good manufacturing procedures consistent with regulatory requirements, high quality and consistency of products, economies of scale, access to raw material and waste treatment. (1027) The pharmaceutical grade lactose industry has quite high barriers to entry, in the form of know how, investments and time consuming approval. (1028)

1747. The conversion of facilities that currently produce food grade lactose would be considerable. The installation of additional equipment and installations to demineralise and decolour the edible grade lactose and additional particle milling and sizing would require a large capital investment. Also, investment may be needed in order to isolate the pharmaceutical lactose production from edible grade lactose production in order to reduce the risk of protein traces in the pharmaceutical lactose. Producers that are currently active in the production of food grade lactose have substantially explained to the Commission that it is significantly more complicated to start producing pharmaceutical lactose than the notifying parties have submitted. (1029) A supplier of food grade lactose, that was according to the notifying parties particularly well placed to enter the market for pharmaceutical grade lactose because it would already be active in refined food grade lactose, indicated that it simply would not see pharmaceutical lactose as its market, so it has not even studied what kind of investments would be needed, since it will not start producing pharmaceutical lactose, not even after a possible price increase as a result of the merger. (1030)

1748. Current suppliers of edible grade lactose would also face very long lead times as they would need to present their product to the market. Especially in light of regulatory approval procedures for pharmaceutical grade lactose and customer qualification procedures, such introduction of pharmaceutical grade lactose would be lengthy. A supplier of food grade lactose has stated that the approval process for the pharmaceutical industry is quite time consuming. According to this supplier, a general approval as supplier as well as an approval for individual drugs would be required. This supplier estimates that these processes would take 3 to 5 years. This is well beyond the two year time horizon that is normally used in merger assessment. (1031)

1749. However, existing producers of pharmaceutical lactose could increase their volumes thereby preventing the merged entity to increase its prices. During 2008, competitors of the merged entity appear to have supplied more, thereby reducing the sales volume of - mainly - Campina.

1750. The figures of 2008 have shown that competitors are able to expand production volumes, for example by further debottlenecking their facilities. Moreover one supplier has indicated that it could divert a very significant production volume from the production of food grade lactose to the production of pharmaceutical grade lactose. (1032) Another producer has indicated that it has several Mt spare capacity for the production of pharmaceutical lactose.

1751. This production volume, together with further expansion by competing suppliers could constrain the merged entity, thereby preventing it form increasing prices, by decreasing output. This holds true both for direct compression and wet granulation types of lactose.

16.4.3.3. Customers have possibilities of switching supplier

1752. For some customers, switching to another supplier is difficult during the life cycle of a drug. Some customers indicate that post product launch (commercial production) of a pharmaceutical product it is difficult to switch outside the qualified sources due to the amount of validation and production process investigation required. One customer indicates it would be easier to switch during the formulation stage of a drug's lifecycle, but it could cause regulatory submissions to slip, so would not be desirable for this particular customer. (1033)

1753. The market investigation in a previous case with regard to pharmaceutical lactose (M.4207 Campina/Fonterra Cooperative Group JV) has shown that Friesland Foods and Campina are close competitors in this market. A number of customers in this investigation have confirmed that Friesland Foods and Campina are close competitors. (1034) However, the market investigation did not confirm that Campina and Friesland Foods are the closest competitors. Competitive and decisive criteria for customers are quality, availability and capacity, technical ability, know-how and support of the supplier, range of products and price. According to customers, Campina and Friesland Foods are close competitors when taking those criteria into account. A large number of respondents to the market investigation indicated that Campina and Friesland Foods are especially close competitors. Especially on range of lactose products, quality and support Campina and Friesland Foods seem to be close competitors. Other competitors follow with some distance. Some customers indicated that Meggle is a close competitor to both Friesland Foods and Campina. The notifying parties have submitted that Meggle is the closest competitor to both Campina and Friesland Foods. Kerry Foremost is also a recognized player although some customers rank Kerry Foremost lower in terms of quality, product innovation and support.

1754. Many customers apply dual sourcing or multi sourcing with two or more qualified suppliers. Some customers have one or two suppliers as a back up in addition to their dual sourcing strategy. The reduction in potential suppliers would reduce the number of alternative supply sources with one, but would not eliminate the options for purchasers of pharmaceutical lactose. Friesland Foods, Campina, Meggle and KerryForemost all a range of pharmaceutical lactose products, including wet granulation and direct compression grades. Hilmar and Alpavit also produce pharmaceutical wet granulation types of pharmaceutical lactose.

1755. Hilmar Cheese offers pharmaceutical grade lactose since it has entered recently. Hilmar presented to a Canadian distributor (1035) already in 2004 its plans to invest in the production of spray dried lactose and anhydrous lactose and it is currently producing a specific grade of pharmaceutical lactose (200 mesh USP/EP/JP/NF). The notifying parties have submitted that customers approve new suppliers of all excipients regularly. Such approvals may take time and come at a certain cost, but they are part of the normal business expenditures of pharmaceutical companies. The evidence on switching that was submitted by the notifying parties upon request of the Commission on 31st of October 2008 shows that customers do switch. This is reflected in Table 16-8 that reflect a list to whom Friesland Foods and Campina respectively lost sales. (1036)

[...]*

Table 16-8: Switching behaviour of notifying parties' customers - Source: Submission of the notifying parties.

1756. The 2008 data that were submitted by the notifying parties and as combined with 2008 sales volumes of its main competitors, both Meggle and KerryForemost and the fringe competitors Hilmar and Alpavit, show that volumes can shift. Therefore it is also shown that a number of customers can switch. Customers would therefore be protected in case the merged entity would attempt to increase prices.

1757. Therefore, because of the possible capacity expansion and the possibility of customers to switch, the merged entity will likely be constrained by its competitors and would not be able to reduce output and increase prices.

16.4.3.4. Conclusion

1758. It can therefore be concluded that the concentration will not lead to a significant impediment of effective competition as regards pharmaceutical grade lactose, be it in the EEA or worldwide.

16.4.4. DPI Pharmaceutical grade lactose

1759. The concerns expressed by market players and the arguments put forward by the notifying parties have been carefully assessed and the conclusion is that the proposed transaction would not significantly impede effective competition on the market for the production and sale of pharmaceutical lactose for the reasons set forth below.

1760. Pre-merger, the market for DPI pharmaceutical lactose is already concentrated as there are currently only three active players. Campina and Friesland Foods are both particularly strong in DPI pharmaceutical lactose in Europe and on the worldwide market. In the market for DPI pharmaceutical lactose Campina has a market share of approximately [40-50]*% and Friesland Foods has a market share of approximately [50-60]*%. (1037)

1761. However, a separate relevant market should be defined for DPI lactose with high consistency that is used by one particular customer and less sophisticated DPI with a lower consistency.

1762. Campina and Friesland Foods are not close competitors in DPI lactose, not only because the market test indicated that Friesland Foods is perceived by some customers of having superior quality and product innovation skills. Although one customer indicated that the level of functionality of the DPI lactose can only be offered by the notifying parties and by no other producer. (1038)

1763. The notifying parties underlined on 26 September 2008 that in their opinion there is no effective competition between Friesland Foods and Campina in the field of DPI lactose and that hence the proposed merger cannot create a significant impediment of effective competition. Indeed, it can be concluded that, because the Kapuini plant is only producing to one particular customer, no changes in the market structure appear.

1764. The notifying parties have stated that Friesland Foods is supplying [...]*% of its sales to one customer. Campina is also supplying from a particular plant to this particular customer from its plant in Kapuni. The market investigation confirmed that customers of high consistency DPI can currently purchase DPI lactose only from Friesland Foods, since Campina in its Kapuni plant only produces for one particular customer. Post merger there would be no change for these customers, because these customers do not benefit from current competition between Friesland Foods and Campina in the sophisticated high consistency DPI lactose because Campina (through its Kapuni plant) has an exclusive agreement with the aforementioned customer. Therefore for these customers the transaction will not bring about any changes with regard to competition in the market for high consistency DPI.

1765. At the same time, according to the notifying parties, Friesland Foods is not supplying the lower consistency type of DPI that Campina is producing outside its Kapuni plant. Outside this particular plant, Campina has, according to the notifying parties, a diverse portfolio of customers for its sales of DPI. Campina also points out that within this portfolio the three biggest customers are not sourcing from Friesland Foods. Therefore with regard to the less sophisticated DPI lactose, there is hardly any overlap. As for the group of customers purchasing [...]*% of the DPI lactose of Friesland Foods, it should be noted that Friesland Foods was not seen as an alternative by customers due to the fact that the Friesland Foods production facility was dominated by the aforementioned customer. [...]* For these reasons, Friesland Foods has not been able to develop its DPI business and can therefore not be seen as a competitive constraint to Campina.

1766. With regard to the particular customer that sources its sophisticated high consistency DPI pharmaceutical grade lactose from both notifying parties, it should be noted that it has long term contracts and supports the production facilities actively. The notifying parties have submitted evidence that the customer helped developing the plants that produce DPI.

1767. The notifying parties have argued that the merged entity would be as dependent on this customer as vice versa. According to the notifying parties the merged entity could not increase its prices vis-a-vis this customer. The customer itself has stated that developing the Kapuni plant was aimed at securing supplies and was not aimed at increasing competition for its DPI product. The increased competition was a by-effect. In that respect this particular customer is protected by a long term contract. The Commission also considers that this particular product is developed in cooperation with the customer and that the merged entity would need to rely on this particular customer. This particular customer has stated that it would rather prefer the risk of a price increase as a result of the merger than the divestiture of the DPI business to a third party. This customer is therefore protected by its long term relationship and interdependency of the customer and suppliers in question.

1768. On the basis of these elements, it can be concluded that the proposed concentration would not significantly impede effective competition on the markets for DPI lactose. In light of this conclusion, it is not necessary to consider the efficiency defence invoked by the notifying parties.

16.4.5. Overall Conclusion

1769. It can therefore be concluded that the concentration will not lead to a significant impediment of effective competition as regards pharmaceutical lactose and DPI lactose (nor in the market for sophisticated nor in the market for less sophisticated DPI lactose) be it within the EEA or worldwide. For these reasons developed, the concentration is also not likely to lead to a significant impediment of effective competition in the field of food grade lactose.

17. COMMITMENTS PRESENTED BY THE NOTIFYING PARTIES

17.1. Introduction

1770. In order to remove the competition concerns identified which arise from the concentration, Campina and Friesland Foods have proposed commitments under Article 8(2) of the EC Merger Regulation. The first set of commitments was submitted on 28 October 2008, complemented on 5 November 2008 with a view to obtaining clearance of the operation from the Commission. The remedy package ("the first commitment proposal") consists of divestment businesses in fresh dairy products, cheese, long-life dairy drinks and access to raw milk.

1771. Subsequently the Commission market tested the commitments ("The first market test"). The results of the first market test showed that significant improvements were needed. As a consequence, the notifying parties submitted on 19 November an improved commitments package ("the second commitment proposal"), which addressed weaknesses identified in the first commitment proposal concerning the fresh dairy divestment business, the cheese divestment business and the long-life dairy drinks package.

1772. However, the Commission continued to have concerns that the lack of access to raw milk would create a significant impediment of effective competition on the downstream markets for fresh basic dairy products and Dutch-type cheese in the Netherlands in general and would result in a lack of viability for the downstream divestment businesses in particular. Therefore the Commission market tested the second commitment proposal with respect to access to raw milk for the divestment businesses. The market testing of the second package ("the second market test") confirmed that improvements were needed in this respect.

1773. Subsequently, on 27 November 2008 the notifying parties submitted a final commitments proposal ("the third commitment proposal"). This proposal includes:

(a) the divestment of the entire fresh dairy business of Friesland Foods in the Netherlands covering the products fresh milk, fresh buttermilk, plain yoghurt, value added yoghurts and quark, fresh custard, porridge, fresh flavoured dairy drinks, fresh cream and organic fresh basic dairy products.(hereinafter "the Fresh Divestment Business").

(b) an exclusive, renewable 5 year license to use the Friesche Vlag brand name in the Netherlands for the current Friesland Foods fresh dairy product portfolio, followed by a perpetual black out period (1039).

(c) the ownership of Campina's Melkunie brand and the ownership of all Friesche Vlag subbrand names and all brands that are specific to the fresh dairy business of Friesland Foods (with the exception of the Friesche Vlag brand itself) are included in the proposal.

(d) the divestment of Campina's Bleskensgraaf production facility and the carve out of a sales team and other employees for R&D, planning and logistics and general support from the sales organisation of the merged entity (hereinafter "the Cheese Divestment Business"). (e) for long-life dairy drinks, the divestiture of Campina's brand in the chocolate flavoured segment Choco Choco and the divestiture of Campina's fruit-flavoured brand Yogho Yogho.

(f) three elements aiming at ensuring access to raw milk for downstream competitors, including the divestment businesses : transitional supply agreements which would allow purchasers of the divestment businesses to source raw milk from the notifying parties during a transitory period under competitive conditions; drawing rights for a maximum volume of 1.2 billion kg of raw milk to plants of dairy companies (purchasers of the divestment businesses in the Netherlands and other dairy companies active in the same downstream markets); and reduction of the exit barriers from the new cooperative by means of financial incentives for farmers deciding to leave the merged entity.

17.2. Fresh Dairy

17.2.1. Competition concerns identified

1774. As explained above, the proposed merger would significantly impede effective competition in the markets for (i) fresh milk, (ii) fresh buttermilk and (iii) plain yoghurt in the Netherlands (see section 7.1.3.4); on the market for value added yoghurt and quark in OOH in the Netherlands (see section 10.3.3); on the market for branded non-health fresh flavoured dairy drinks in the Netherlands separated according to the distribution channel in retail and OOH (see section 11.2.3.6); and on (i) the market for fresh custard in the Netherlands and (ii) the market for porridge in the Netherlands (see section 12.3.2.3).

17.2.2. Commitments offered

1775. The Fresh Divestment Business consists of all fresh dairy activities of Friesland Foods (hereinafter "FF Fresh") in the Netherlands. According to the notifying parties, it is a stand alone operating unit of Friesland Foods responsible for the production, sales and marketing in the Netherlands of nonorganic (i) fresh milk, (ii) fresh buttermilk, (iii) fresh custard (vla), (iv) fresh plain yoghurt, (v) fresh dairy desserts, (vi) fresh flavoured dairy drinks, (vii) value added yoghurts and quark and (viii) fresh liquid cream which are sold under the Friesche Vlag brand and under private label. In addition, it sells and markets all (i) organic fresh milk, (ii) organic fresh basic yoghurt, (iii) organic fresh custard (vla) and (iv) organic fresh buttermilk which is sold under the Friesche Vlag brand (together referred to as the "FF Fresh product portfolio").

1776. The Divestment Business currently employs approximately 513 staff (full time equivalents, "FTEs"). The majority is permanent staff (363 FTEs), most of whom are in operations (318 FTEs), with the remainder in commerce and functional support areas. All commercial and functional support staff are situated next to the production facility and distribution centre in Nijkerk.

1777. The tangible assets include the ownership of the production facility and the distribution centre at Nijkerk as well as the dedicated (commercial and administrative) organisation of the FF Fresh. The Fresh Divestment Business will include the whole site and the distribution centre of FF Fresh, including ownership of the real estate on which the buildings are located as well as the buildings themselves. The Nijkerk plant consists of a mainstream and a specialty products facility. Overall the plant produces [...]* litres of fresh dairy products per year and has a capacity of [...]* litres of fresh dairy products per year.

1778. The Nijkerk production facility has a total of 18 packaging lines, which produce 190 store keeping units ("SKUs"), divided into 146 branded and 44 PL products. Of the 18 packaging lines, FF Fresh owns 11 and leases 7 lines. 8 packaging lines are dedicated to mainstream products and 10 lines are dedicated to specialty products.

1779. The Fresh Divestment Business will include an exclusive, once renewable, royalty-free five year license to use the Friesche Vlag brand name in the Netherlands for the FF fresh portfolio. After the license period the notifying parties will no longer use the Friesche Vlag brand for these products in the Netherlands. The licence will be granted for five years but may be prolonged for another five year period at the request of the purchaser.

1780. Furthermore, the Fresh Divestment Business will include a perpetual royalty-free license for the use of the Weidemelk logo, under the condition that the logo is used for products that fulfil the criteria of the Weidemelk foundation.

1781. The Fresh Divestment Business will include the full ownership of all Friesche Vlag subbrand names and all brands that are specific to the products of FF Fresh (with the exception of the Friesche Vlag brand itself). Furthermore the proposal also includes the divestment of Campina's brand Melkunie at the option of the purchaser. The purchaser of the Fresh Divestment Business will be able to continue using all Friesche Vlag sub-brands even after having re-branded the FF Fresh product portfolio from Friesche Vlag to, for instance, Melkunie.

1782. The Fresh Divestment Business will include the transfer of ownership of recipes, production and process know-how and related patents for all products currently contained in the FF Fresh product portfolio.

1783. Friesland Foods commits to transfer to the purchaser of the Fresh Divestment Business all licenses, permits and authorisations which have been issued for the operation of the Fresh Divestment Business, to the extent permitted by the competent authority.

1784. The Fresh Divestment Business will also include eight year supply agreement for residual cream at the option of the purchaser and a one year supply agreement for organic fresh dairy products.

17.2.3. Assessment of the commitments offered

1785. Under the EC Merger Regulation, the Commission has power to accept commitments that are deemed capable of rendering the concentration compatible with the common market so that they will prevent a significant impediment of effective competition. As indicated in point 9 of the Commission notice on remedies acceptable under Council Regulation (EC) NO 139-2004 and under Commission Regulation (EC) No 802-2004 ("the Remedies Notice") (1040), the commitments have to eliminate the competition concerns entirely and have to be comprehensive and effective from all points of view

1786. In particular in the case of a divestiture, it is important that the divested activities consist of a viable business which, if operated by a suitable purchaser, can compete effectively with the merged entity on a lasting basis.

1787. The Commission market tested the commitments consulting customers and competitors on the commitments for the Fresh Divestment Business. The first market test revealed overall positive results. However, some respondents considered the originally foreseen licensing agreement of three years followed by a black out period of three years as a problematic factor to be taken into account when assessing the viability of the remedy proposal. Others considered that the product portfolio of the Fresh Divestment Business would not be broad enough to allow sufficient flexibility in the production of fresh products to be competitive. Finally, several concerns were raised with respect to the necessary access to Dutch raw milk. (1041)

17.2.3.1. Effectiveness

1788. The divestiture of Friesland Foods' fresh business to a suitable purchaser will entirely eliminate the downstream overlap brought about by the proposed merger with respect to the Dutch market for (i) fresh milk, (ii) fresh buttermilk, (iii) plain yoghurt, (iv) custard, (v) porridge (vi) value added yoghurt and quark (OOH) and (vii) fresh flavoured dairy drinks (in sections 17.2.3 and 17.2.4, all the products mentioned in (i) to (vii) are commonly referred to as the "fresh dairy markets").

1789. Looking at the downstream fresh dairy markets, the remedy will restore the market structure which existed pre-merger. The suitable purchaser of the Fresh Divestment Business will have an initial market share between [20-30]*% and [50-60]*% depending on the relevant product market and will be at least the second largest player in the respective markets, and all possible sub-segmentations like private label or branded, retail or OOH.

1790. The Fresh Divestment Business will be a competitive force. It can be expected that it will adapt to the specific demand of retailers and OOH wholesalers in the Netherlands and will at least be able to defend Friesland Foods' current market shares, provided that it can secure sufficient volumes of raw milk supply at prices and terms that allow the divestment business to compete effectively downstream. Provisions in the first and second commitment proposal, in particular the production facility with a capacity allowing large volumes to be supplied, the transfer of all current contracts, the licence of the Friesche Vlag brand, the ownership of all subtrade marks, the product portfolio and the complementary supply of raw milk will ensure that the suitable purchaser can compete successfully with FrieslandCampina on the Dutch market.

1791. For these reasons, the divestiture of the Fresh Divestment Business will be sufficiently effective to remedy the competition concerns in the fresh dairy markets in the Netherlands provided that access to Dutch raw milk is ensured.

17.2.3.2. Independence, viability, and competiveness of the divested business

1792. The first market test of the commitment package for fresh dairy products revealed that a suitable purchaser will be able to operate the Fresh Divestment Business as an independent, viable and competitive force on the Dutch fresh dairy markets provided certain conditions, in particular that access to Dutch raw milk at competitive prices and on competitive conditions and in sufficient volumes, are fulfilled. This is especially important since only with raw milk supply at competitive terms and conditions would the Fresh Divestment Business be able to compete effectively on the markets for fresh basic dairy products downstream. Without a raw milk supply at conditions that would allow the divestment business to compete effectively, not only would the business not be viable in itself, but effective competition would not be restored either. For the same reason a sufficient volume of raw milk supply should be secured, so that it allows the Fresh Divestment Business to grow without being dependent on the merged entity.

1793. This has been confirmed by the majority of customers and competitors who considered the Fresh Divestment Business to be a viable and stand alone business with an attractive product portfolio, which retailers indicated would allow them to continue sourcing from it. (1042)

17.2.3.2.1. The licence agreement for Friesche Vlag

1794. The commitments do not foresee a divestiture of the Friesche Vlag brand; instead an exclusive, royalty-free licence with the purpose of a re-branding is foreseen. This arrangement has been criticised by respondents in the market test, which requested at least an extended licence and black out period allowing for re-branding and ensuring the viability and competitiveness of the Fresh Divested Business.

1795. The second and third commitments proposals extend the licence period from three to five years, followed by a perpetual black out period. Moreover, the five year licence period could - at the request of the purchaser, - be prolonged by a further five years.

1796. The Commission in general takes a critical view towards re-branding commitments and rather prefers brand divestitures. Based on its previous experience re-branding may fail. In this particular case, however, the remedy package also includes a production and distribution facility as well as the Melkunie brand, which was used for many years in the Dutch fresh dairy market and still appears to be known by many consumers in the Netherlands. Under these circumstances and also taking into account the fact that the notifying parties will not be allowed to use the Friesche Vlag brand in the fresh dairy markets due to the perpetual black out period the commitment can be expected to allow the purchaser of the Fresh Divestment Business to successfully re-brand its products. (1043)

17.2.3.2.2. Product portfolio and flexibility to supply

1797. Respondents to the first market test viewed the ability to use raw milk outside the fresh dairy markets as an important condition for the viability and competitiveness of the Fresh Divestment Business. A broader product portfolio would in this respect allow the purchaser to react in a flexible manner to fluctuations in demand because of promotions or seasonal demand. Without a product like cheese or milk powder the Fresh Divestment Business would not be able to get rid of excess raw milk or re-channel additional raw milk to the fresh business if necessary.

1798. In reply to these concerns, the notifying parties strengthened the supply conditions applicable under the raw milk transitional supply agreement and the Dutch Milk Fund (see 17.2.4) for the Fresh Divestment Business providing for more flexibility with respect to volume and adjustments of volume.

1799. Moreover, the purchaser of the Fresh Divestment Business has been given the right of first refusal for the Cheese Divestment Business, which allows it to enlarge its product portfolio and enhance its flexibility.

17.2.3.2.3. Supply and sales agreements

1800. As mentioned in section 17-1, the commitments offered by the notifying parties to restore effective competition on the Dutch fresh dairy markets also include several supply and sales agreements to be concluded at the option of the purchaser with the notifying parties. In particular, the purchaser has the option to buy raw milk from the notifying parties (see in detail section 17.4) and to sell residual cream for a period of eight years (in proportion to the same quantities as the merged entity supplies raw milk to the Nijkerk production).

1801. The market test on the proposed sales agreement for residual cream was on balance positive. Although some respondents mentioned that it might be difficult for the Fresh Divestment Business to find other outlets for cream and that the supply agreement might lead to a dependency of the Fresh Divestment Business upon the notifying parties, most respondents consider that (i) the price determined on the basis of the fat price which is an element of the guaranteed price for raw milk minus any costs for additional transport and handling would be an arm's length sales price for cream in the Netherlands and even above current market prices; that (ii) the Fresh Divestment Business could find easily other outlet channels for this cream; and that (iii) such a sales agreement would not influence the independence of the Fresh Divestment Business. In particular fresh dairy producers in the Netherlands made clear in their responses to the market test that the Fresh Divestment Business could easily find other outlets for residual cream or could set up its own cream facility and that the independence of the Fresh Divestment Business would not be influenced by such a sales agreement.

1802. The notifying parties explained in this context that the residual cream currently produced by the Fresh Divestment Business could be transported over a distance of more than 300 km and submitted that several alternative outlets other than the facilities of the notifying parties are located within this area. This further confirms that the Fresh Divestment Business can be expected to find alternative outlets for its residual cream.

17.2.3.2.4. Criteria to be fulfilled by the purchase of the Fresh Divestment Business

1803. Respondents to the market test widely agreed that a suitable purchaser would be an existing dairy company with experience in the fresh dairy business and possibly complementary products like cheese or milk powder.

17.2.4. Conclusion

1804. On the basis of these elements, it can be concluded that the commitment for fresh dairy products, in combination with the commitment for raw milk discussed in section 17.4, is sufficient to remedy the competition concerns identified in the fresh dairy markets in the Netherlands.

17.3. Cheese

17.3.1. Competition concerns identified

1805. The proposed merger would lead to a significant impediment of effective competition (i) on the markets for the sale of Dutch-type cheese to specialised cheese wholesalers in the Netherlands (or narrower segmentations into nature, Gouda, and 15-day-old cheese); and (ii) on the markets for the sale of Dutch-type cheese to modern types of retail in the Netherlands (or narrower segmentations into nature and Gouda) (see section 8.4.4).

17.3.2. Commitments offered

1806. As explained in section 17.1, the notifying parties propose to divest Campina's Bleskensgraaf production facility and to carve out an experienced sales team and other employees for R&D, planning and logistics and general support from the sales organisation of the merged entity (together the "Cheese Divestment Business") in order to restore effective competition on the markets for the sale of Dutch-type cheese to specialised cheese wholesalers and modern types of retail in the Netherlands.

1807. The Cheese Divestment Business comprises the whole of the Bleskensgraaf production facility (including the land and premises which are wholly owned by Campina), the recipes for the production of Gouda cheese currently produced by the Bleskensgraaf plant (as well as recipes for Maasdam and Edam if required) and all necessary licences, permits and authorisations to operate the plant.

1808. The Cheese Divestment Business further comprises the current personnel of the Bleskensgraaf plant (46 fixed and 10 temporary employees). The employees to be carved out from the merged entity's organisation are (i) a senior sales manager with sufficient experience to make sales to specialised cheese wholesalers and modern types of retail and supporting staff, (ii) a sales and operations planner and supporting staff, support related to development and quality assurance, a senior controller and supporting staff as well as support staff on administration and stand alone operation.

1809. The Cheese Divestment Business has no ripening and packaging facilities on site, but is using outsourced (internal and external) services for these activities. It does not include customer contracts and customer records which are not plant specific. The Dutch-type cheese produced at Bleskensgraaf is so far marketed centrally by the Campina sales organisation.

1810. The Bleskensgraaf facility is currently dedicated to the production of Gouda nature cheese (both wheel shaped and rectangular shaped), but it is also suitable for the production of Edam and Maasdam nature cheese. In 2007, the Bleskensgraaf production facility achieved a production of approximately [...]* tonnes Dutch-type cheese. Taking into account certain conditions in the current configuration it has the capacity to achieve a higher output of approximately [...]* tonnes. The Bleskensgraaf facility does not produce rindless Dutch-type cheese.

1811. The commitments offered by the notifying parties to restore effective competition on the Dutch-type cheese markets also include several supply and sales agreements to be concluded at the option of the purchaser with the notifying parties: (i) a transitional supply agreement for the supply of raw milk (see in detail section 17.4.1); (ii) a supply agreement of product and nonproduct related goods required for the production of the current volumes of the product portfolio of the Bleskensgraaf facility with a maximum duration of 1 year; (iii) a sales agreement for residual cream for a maximum period of 8 years; (iv) a sales agreement for residual whey for a maximum period of 8 years; and (v) an agreement to ripen cheese produced by the Bleskensgraaf facility at third parties facilities under the current agreements between the notifying parties and third parties and/or to pre-pack the cheese in the notifying parties' own facilities at market conditions.

17.3.3. Assessment of the commitments offered

17.3.3.1. Effectiveness

1812. The divestiture of the Cheese Divestment Business will remove a very significant part of the overlap brought about by the concentration in the Dutch-type cheese markets in the Netherlands (including narrower segmentations) (1044).

1813. On the market for the sale of nature Dutch-type cheese to specialised cheese wholesalers in the Netherlands, the activities of the notifying parties overlap by a volume of [...]* tonnes. On the market for the sale of nature Dutch-type cheese to modern types of retail in the Netherlands, the activities of the notifying parties overlap by a volume of [...]* tonnes. Adding together the overlaps in both markets would represent a "total" overlap of [...]* tonnes.(1045)

1814. As explained below, the notifying parties submit that the Bleskensgraaf plant has a maximum output of [...]* tonnes of nature Dutch-type cheese. This output is as such suitable to remove either (i) almost the entire overlap on the market for sales to specialised cheese wholesalers as such ([...]* tonnes); or (ii) the entire overlap on the market for sales to modern types of retail as such ([...]* tonnes); or (iii) approximately two thirds of the "total" overlap on these two markets ([...]* tonnes).

1815. In assessing the effects of the remedy proposed, account should be taken of the fact that specialised cheese wholesalers sell Dutch-type cheese to modern types of retail (see section 8.2.4.1) and that both markets are therefore linked. For instance, assuming that the entire production of the Bleskensgraaf plant will be sold to specialised cheese wholesalers (thereby removing almost the entire overlap on that market), it can also be reasonably assumed that a significant part of this output (equivalent to more than 70% of the overlap at the level of modern types of retail) will be passed on by those wholesalers to the level of modern types of retail. (1046)

1816. It follows that the divestiture of the Bleskensgraaf plant will remove a very significant part of the overlap between the notifying parties in the markets for nature Dutch-type cheese.

1817. This divestiture will either create a new player on the Dutch cheese markets in the Netherlands or will significantly expand the nature cheese capabilities of the existing players on the Dutch markets. In both scenarios, an additional competitive force in the nature Dutch-type cheese market is created which is able to supply a significant volume of nature Dutch-type cheese to specialised cheese wholesalers and/or modern types of retail. Once bought by a suitable buyer, the Cheese Divestment Business can be expected to constrain the merged entity and compensate for the loss of competitive pressure that Friesland Foods and Campina exerted on each other before the merger.

1818. For these reasons, the divestiture of the Cheese Divestment Business will be sufficiently effective to remedy the competition concerns both in the markets for the sale of Dutch-type cheese to specialised cheese wholesalers (including narrower segmentations) and in the markets for the sale of Dutch-type cheese to modern types of retail (including narrower segmentations), provided that access to Dutch raw milk is ensured.

17.3.3.2. Independence, viability, and competiveness of the Divested Business

1819. Respondents to the market test widely agreed that a suitable purchaser should be able to operate the Cheese Divestment Business as an independent, viable and competitive force on the Dutch-type cheese markets in the Netherlands. (1047)

17.3.3.2.1. Viability and Competitiveness of the Divestment Business

1820. A clear majority of respondents consider that the Cheese Divestment Business could become an alternative supplier for both 15-day-old nature cheese and mature nature cheese. (1048)

1821. Furthermore, no significant concerns were received as regards the proposed carving-out of personnel or as regards the fact that no customer contracts are transferred to the Cheese Divestment Business. Respondents also widely agreed that it would not be important for the Bleskensgraaf plant to have its own maturing capacity as several maturing companies would be located in the vicinity. (1049)

1822. While the general transfer of know-how appears to be sufficiently guaranteed by the fact that the Cheese Divestment Business comprises the current personnel of the Bleskensgraaf facility, some respondents underlined, however, the necessity of a transfer of recipes. (1050) The notifying parties offered in their improved remedy package to transfer the recipes for the production of Gouda cheese currently produced by the Cheese Divestment Business (as well as recipes for Maasdam and Edam if required) to the purchaser of the Cheese Divestment Business.

1823. Some respondents has doubts about the technical infrastructure of the Bleskensgraaf plant and its cost-efficiency to produce bulk nature cheese at large scale. (1051) The notifying parties have, however, provided internal data showing that the Bleskensgraaf facility is among Campina's most cost efficient cheese production facilities. More particularly, the plant was established in 1990 and a number of expansions and upgrades have taken place in the course of the years. Most recent investments include: (i) an investment in a new production line for rectangular cheese in 2006 (approximately EUR [...]*); (ii) the automation of the factory cheese ripening facility for 15-day-old cheese in 2008 (approximately EUR [...]*), and (iii) an upgrade of the wastewater treatment plant in 2008 (approximately EUR [...]*). The Bleskensgraaf production facility is therefore an upgraded, cost-efficient plant and no major replacement investments will be necessary within the next five years. The competitiveness of the Bleskensgraaf production facility is further demonstrated by a comparison of conversion costs of the Bleskensgraaf plant and other Campina factories.

1824. This submission of the notifying parties is consistent with the comments made in the course of the market test by several respondents that the Bleskensgraaf plant is a state of the art plant and that it has been renovated in the recent past. (1052)

1825. The cost-efficiency and the ability of the Bleskensgraaf plant to produce at large scale are further improved by the right of first refusal of the acquirer of the Fresh Divestment Business (see section 17.2 above) which will allow the acquirer of both businesses a higher scale of economies.

1826. Furthermore, the notifying parties have stressed in this context that the Bleskensgraaf production facility already has the capacity to achieve a higher output (i.e. approximately [...]* tonnes) and that it would be able to extend its capacity by approximately [...]* tonnes with limited investments and by further approximately [...]* tonnes with additional investments within a period of three years. The Bleskensgraaf plant would therefore be able to expand its current output of approximately [...]* tonnes by more than 20% to approximately [...]* tonnes.

1827. As to the capacity of the Bleskensgraaf plant for expansion the market test of the commitments pointed to the need of the Cheese Divestment Business to have access to raw milk (see section 17-4 below). In particular the access to Dutch raw milk at competitive prices and conditions and in sufficient volumes is necessary. This is especially important since only with raw milk supply on competitive terms and conditions would the Cheese Divestment Business be able to compete effectively on the markets for Dutch-type cheese in the Netherlands. Without a raw milk supply on conditions that would allow that business to compete effectively, not only would the business not be viable in itself, but effective competition would not be restored either. For the same reason a sufficient volume of raw milk supply should be secured, so as to allow the Cheese Divestment Business to grow without being dependent on the merged entity.

1828. It can be concluded that the Cheese Divestment Business is currently a competitive supplier on the Dutch market for Dutch-type cheese, that it can be run as a viable and stand alone business provided access to raw milk is ensured, and that it would allow a new entrant (or an existing player) in the Dutch cheese market to compete effectively with the merged entity in the markets for Dutch-type cheese.

17.3.3.2.2. Supply and sales agreements

1829. As mentioned above in section 17.1 the commitments offered by the notifying parties to restore effective competition on the Dutch-type cheese markets also include several supply and sales agreements to be concluded at the option of the purchaser with the notifying parties. Access to raw milk will be discussed in section 17.4. The sales agreements would allow the acquirer of the Cheese Divestment Business to sell residual cream and whey to the notifying parties for a period of eight years.

1830. The market test on the proposed sales agreement for residual cream was on balance positive. Although some respondents mentioned that it would be difficult for the Cheese Divestment Business to find other outlets for cream and that the supply agreement would lead to a dependency of the Cheese Divestment Business upon the notifying parties, most respondents consider that (i) the price determined on the basis of the fat price which is an element of the guaranteed price for raw milk minus additional costs would be an arm's length sales price for cream in the Netherlands; that (ii) the Cheese Divestment Business could easily find other outlet channels for this cream; and that (iii) such a sales agreements would not influence the independence of the Cheese Divestment Business. (1053)

1831. The market test on the proposed sales agreement for whey revealed that almost all respondents found that a cost calculation on the basis of the whey price quotation published by the Dutch Dairy Board would constitute an arm's length price for whey. Concerns were expressed as to the possibility of the Cheese Divestment Business to find other outlet channels for whey (in particular since whey could not be transported over long distances) and that the Cheese Divestment Business would be dependent on the notifying parties. (1054)

1832. It has been clarified by the notifying parties that the liquid whey currently produced by the Bleskensgraaf plant is a concentrated product which can be transported over a distance of more than 300 km. Several alternative outlets other than the facilities of the notifying parties are located within this area. Additional investments of approximately [...]* EUR would moreover enable the Bleskensgraaf plant to further concentrate the whey, thereby allowing it to be transported over distances of 800 to 1000 km.

1833. Against that background, it can be concluded that the Cheese Divestment Business will find alternative outlets for its residual whey, thus allowing it several options for the sale of whey.

17.3.3.2.3. Criteria to be fulfilled by the purchase of then Cheese Divestment Business

1834. A suitable purchaser would be an existing dairy company with experience in the cheese business. This was also confirmed by respondents to the market test. (1055)

17.3.4. Conclusion

1835. On the basis of these elements, it can be concluded that the commitment for cheese, in combination with the commitment for raw milk discussed in section 17.4, is sufficient to remedy the competition concerns identified in the Dutch-type cheese markets.

17.4. Access to raw milk for the divestment businesses in fresh dairy products and cheese and other competitors active in these markets.

1836. As explained above, the proposed merger would lead to a significant impediment of effective competition in the market for procurement of conventional raw milk by bringing together the two main purchasers of raw milk in the Netherlands. The market power that they would have on downstream markets would enable them to raise additional profits and therefore pay higher prices to farmers. Consequently, the merged entity would be in a position to attract more farmers and maintain and/or strengthen its farmers' base. This situation would increase barriers to entry and/or to expansion on the primary downstream dairy markets where Dutch raw milk is needed to compete effectively.

1837. These competition concerns in the market for procurement of conventional raw milk can therefore only be solved through viable commitments in the downstream markets. However, as will be explained in recital 1838, a viable commitment in these downstream markets requires also that access to raw milk for these businesses is ensured on a long-lasting basis. Therefore, specific remedies with respect to access to raw milk are necessary to eliminate significant impediments of effective competition in the downstream and the upstream markets.

1838. The viability of the divestment businesses in fresh dairy products and cheese and the ability of purchasers to compete effectively with the notifying parties on these markets require that access to the main input material, raw milk, is ensured on a permanent basis for these divestment businesses. Given the scarcity of conventional raw milk available in the Netherlands (1056) and the fact that the notifying parties would control post-merger [70-80]*% of the supplies of Dutch raw milk which is necessary to manufacture fresh basic dairy products and cheese, it is necessary that the notifying parties ensure that purchasers of the divestment businesses are guaranteed access to raw milk under conditions which allow them to compete on a lasting basis against the notifying parties on the downstream markets. Without a raw milk supply on conditions that allow the divestment businesses to compete effectively, not only would these businesses not be viable in itself, but effective competition would not be restored either. For these reasons a sufficient volume of raw milk supply should be secured, so as to allow the Fresh and Cheese Divestment Businesses to grow.

1839. Further, purchasers of the Divestment businesses in fresh dairy products and cheese should not be limited in their capacity to develop these businesses by a lack of raw milk. Other competitors active in fresh dairy and cheese markets have also mentioned during the market investigation that access to raw milk could be a problem for their current competitive position and future development since it would be difficult to convince member-farmers of FrieslandCampina to redirect their supply to these players under similar conditions. The fact that [70-80]*% of the raw milk produced in the Netherlands is locked in with the notifying parties is also considered to be a curb for future growth of alternative suppliers of fresh dairy products and cheese. The restoration of conditions of effective competition in fresh dairy products and cheese is highly dependent on access to a significant volume of raw milk under competitive conditions.

1840. As indicated in point 32 of the Remedies Notice, in order to be considered to be standalone, the divestment businesses in fresh dairy products and cheese should operate independently of the notifying parties as regards the supply of input materials or other forms of cooperation other than during a transitory period. Whilst acknowledging that the raw milk supplies are necessary to maintain the full economic viability and competitiveness of the divestment businesses, the notifying parties have consistently indicated that they were not in a position to terminate the membership of member farmers or to "force" them to enter a supply agreement with the acquirers of the divestment businesses. Since farmers are free entrepreneurs, the notifying parties claim that they cannot unilaterally reduce their member base other than in cases linked to the quality of the raw milk supplied. This would go, according to the notifying parties, against the basic functioning principles of a cooperative.

1841. The notifying parties have provided with respect to access to raw milk a set of commitments consisting in the following elements: (a) transitional supply agreements which would allow purchasers of the divestment businesses to source raw milk from the notifying parties during a transitory period on competitive conditions; (b) drawing rights for a maximum volume of 1.2 billion kg to plants of dairy companies (purchasers of the divestment businesses in the Netherlands and other dairy companies active in the same downstream markets) through the mediation of an independent nonprofit organization, the Dutch Milk Fund, ("DMF") which will act as an agent between FrieslandCampina and potential users of these drawing rights. This remedy has been devised in order to free up sufficient volumes of raw milk supply for the downstream business and growth thereof so as to allow sourcing on the longer term and to remove barriers to expansion for other competitors with respect to access to raw milk; (c) reduction of the exit barriers from the new cooperative FrieslandCampina by means of financial incentives for farmers deciding to leave the merged entity ("the incentive scheme"), in order to ensure a more structural sourcing of raw milk independently of the merged entity.

1842. These three elements are further described and assessed below.

17.4.1. Transitional supply agreements

17.4.1.1. Description of the transitional supply agreements

1843. The merged entity undertakes to conclude with purchasers of the divestment businesses in fresh basic dairy products and cheese supply agreements for all the raw milk required to enable the purchasers to utilize the current maximum processing capacity of the Nijkerk facility ([...]* kg) and the Bleskensgraaf facility ([...]* kg). This supply agreement for a volume corresponding to the maximum capacity of the plants will be limited to two years, in order to allow enough time to the purchasers of both divestment businesses to find alternative sources for their raw milk supply, including the DMF scheme. The supply agreements stipulate that, should the Dutch Milk Fund not become operational, at the option of the purchasers, the duration of the two year supply agreement may be extended until 1 January 2017.

1844. The price for the raw milk under the supply agreements will be the monthly "guaranteed price" applicable in the month of delivery, plus a fee for logistical and other costs. The guaranteed price will be subject to a rebate of 1% for the period until six months after the Dutch Milk Fund described in section 17.4.2 has become operational, so at the latest until 31 December 2009. The "guaranteed milk price" corresponds to the pay out prices that the merged entity would offer to its member farmers irrespective of its results. In determining the "guaranteed price", milk prices (pay out prices on annual basis) in Germany (ZMP), Denmark (Arla), The Netherlands (Cono, DOC and Bel Leerdammer) and Belgium (Milcobel) are taken into account. These milk prices are weighted on the basis of the quantity of processed milk in the entire country in question.

1845. All costs that arise from the logistics and transport (collection at farms and delivery. to Nijkerk and Bleskensgraaf as well as quality control, quality assurance and administration) will be calculated in advance by the notifying parties and charged on an objective, transparent and verifiable basis. Those costs will be the same as the average costs charged by FrieslandCampina internally. If he wishes so, the buyer of the raw milk can make his own arrangements for collecting it. In that case no additional cost will be charged.

1846. With respect to planning, the volume needed from year to year will be communicated by the purchaser to FrieslandCampina six months before the beginning of the calendar year concerned. With respect to fresh dairy products, the annual forecast may be modified afterwards to accommodate changes in expected output of fresh dairy products but shall be made definitive at the latest by 1 April of the calendar year concerned (1057). This annual forecast will be supplied by FrieslandCampina in equal weekly volumes provided that orders may be adjusted by a maximum of plus or minus 20% of this weekly volume.

17.4.1.2. Assessment of the transitional supply agreements

1847. The supply agreements as foreseen in the third commitment proposal of 27 November would solve issues raised in the first market test with respect to the supply agreements as annexed to the first commitment proposal.

1848. In particular, with respect to the guaranteed price several competitors in the first market test explained that the price would be too high leading to a competitive disadvantage for the fresh and cheese divestment businesses comparing it with other pay out prices in the Community or with the spot price in the Netherlands (1058). The majority of customers (1059) also saw this guaranteed price as too high since it would be based on companies active in different areas and with a wider product portfolio or more profitable products. Moreover, the guaranteed price appears to be the market price for stable supplies of raw milk; any cooperative or private purchaser that pays less is unlikely to attract farmers and secure supplies in the medium term. The notifying parties have addressed this point by lowering the guaranteed milk price for the supply agreements by 1% for the first year in the third commitment proposal. Moreover, the purchaser would also have the possibility to transfer its supply arrangements to the Dutch Milk Fund (see section 17.4.2) where the price is also the guaranteed milk price minus 1% for the first five years, or of entering into supply contracts with farmers having left FrieslandCampina through the incentive scheme described in section 17.4.3.

1849. Concerning the absolute volume provided for the fresh and cheese divested businesses, several dairy companies stressed in the first market test that growth opportunities should be foreseen and incorporated into the volume (1060). The volume originally provided for both divestment businesses (which corresponded to the volume of raw milk processed by Nijkerk and Bleskensgraaf plants in 2007) was therefore considered as insufficient. The latter view was widely shared by customers who considered the restriction in volume as a way of preventing the achievement of economies of scale for the divestment businesses(1061). The notifying parties have addressed this point by increasing in the second commitment proposal submitted on 19 November the annual volume of raw milk available for the divestment businesses up to the current maximum capacity of the plants. That should enable the purchasers of the fresh and cheese divestment businesses to achieve economies of scale stemming from utilization at full capacity of these plants and to allow for sufficient growth at least for the duration of the supply agreements which is transitory.

1850. Finally, the notifying parties have inserted in the supply agreements a clause with respect to the resolution of disputes through an arbitration process. This is particularly important with respect to all the additional costs likely to be charged by FrieslandCampina, which should be monitored in such a way that potential disagreements are solved rapidly to ensure that the competitiveness of the fresh and cheese divestment businesses is not hindered by a lengthy resolution process.

1851. In conclusion with respect to the transitional supply agreements, the modifications made by the notifying parties in the second and third commitment proposal are sufficient to ensure the viability and competitiveness of the Divestment Businesses in fresh dairy products and cheese for a transitional period of two years.

17.4.2. The Dutch Milk Fund (DMF)

17.4.2.1. Description of the commitment

1852. In the third commitment proposal of 27 November 2008, the notifying parties submitted a remedy with respect to access to raw milk for the divestment businesses after the two-year transitory period and for other current and potential competitors in the fresh dairy and cheese markets. The supply agreements foreseen with respect to fresh basic dairy and cheese commitments are meant to be replaced by supply arrangements through this remedy, which involves the Dutch Milk Fund.

1853. This remedy is devised as follows. The notifying parties will grant drawing rights for a maximum volume of 1.2 billion kg of raw milk to plants of dairy companies in the Netherlands that produce fresh basic dairy products, Dutch type nature cheese or one of these products in combination with other dairy products ("DPC plants"). These drawing rights will be granted through the mediation of an independent non-profit organization DMF (in the form of a "stichting" under Dutch law) which will act as an agent between FrieslandCampina and potential users of these drawing rights.

1854. The price for raw milk will correspond to the guaranteed price that the merged entity will offer to its member farmers minus 1% for the first five years from the date on which this remedy becomes effective (which corresponds to the date that DMF becomes operational) and thereafter will correspond to the aforesaid guaranteed milk price without reduction.

1855. With respect to volume, the volume of 1.2 billion kg has been determined in order to ensure that the fresh and cheese divestment businesses will be in a position to compete effectively against the merged entity. It will first cover the maximum capacity of both plants of the divested businesses (in total [...]*kg). The remaining [...]*kg available will allow purchasers of the divestment businesses to expand their business during the period of functioning of the Dutch Milk Fund. This period will last until the volume made available by DMF is reduced to zero as a result of application of the incentive to leave the merged entity.

1856. DMF will be incorporated before 1 July 2009 and its costs of incorporation will be borne by FrieslandCampina. The board of DMF will consist of 3 qualified members currently not active in the dairy sector and independent and unrelated to the notifying parties, to be appointed by the Minister of Agriculture or the Minister of Economic Affairs in the Netherlands.

1857. Contracts are for a fixed volume, such volume to be spread evenly over the contract term on a weekly basis plus or minus 5% taking into account seasonal fluctuations in raw milk production at normal commercial conditions. The minimum volume per DPC plant per month is 2 million kg based on full truckloads. The minimum order term depends on the volumes involved (3 months between 2 and 4 million kg/month, 4 months for volumes between 4 and 6 million kg/ month, 6 months for volumes over 6 million kg/month).

1858. All costs that arise from the logistics and transport (collection at farms and delivery at plants as well as quality control, quality assurance and administration) will be calculated in advance by the notifying parties and charged on an objective, transparent and verifiable basis. They will be the same as the costs charged by FrieslandCampina internally. The allocation of costs will be supervised by DMF and in case of a dispute, DMF may give binding instructions to FrieslandCampina. However, if a buyer decides to arrange collection and transportation itself, he would be free to do so (subject to consultation of FrieslandCampina and prior approval of the farmers concerned) but should engage to collect for at least three years without interruption. In this case, the buyer will not be charged by FrieslandCampina any mark-up for collection or transportation.

1859. DMF will monitor compliance by FrieslandCampina with the conditions of this remedy on an ongoing basis and may give binding instructions to FrieslandCampina in case of noncompliance. FrieslandCampina and DMF will agree upon an effective arbitration clause in connection with disputes and complaints ensuring quick dispute resolution (and providing for continuity of supply of raw milk pending the arbitration procedure).

1860. The fresh divestment business and the cheese divestment businesses will have preferential drawing rights as set out in the improved commitments (up to the volume representing the total production capacity of those businesses). However, FrieslandCampina is not obliged to supply raw milk to the divestment businesses in excess of the volume that should be made available to them under this remedy. In case preferential drawing rights exercised by the two divestment business exceed the aforementioned volume, the available volume will be allocated on a pro rata basis. Otherwise raw milk is allocated through DMF on a first come, first serve basis.

1861. Finally, DMF will remain in operation until the volume of raw milk to be made available by FrieslandCampina through DMF has been reduced to zero following the application of the incentive to leave (see section 17.4.3). The volume of raw milk available under the DMF commitment will be reduced every year by the volume of raw milk for which applications have been made under the incentives to leave.

1862. Moreover, the Commission will conduct a review at the request of FrieslandCampina following the abolition of the milk quota regime. The Commission will decide to withdraw the remedies with respect to access to raw milk if it is satisfied that sufficient Dutch raw milk is available to DPC plants as defined in recital 1841.

17.4.2.2. Assessment of the commitments offered

1863. The commitments as described in the third commitment proposal of 27 November 2008 would allow purchasers of the divestment businesses and other DPC plants to source raw milk from the notifying parties under conditions which allow them to compete with the notifying parties in the downstream markets. They furthermore address several issues which have been raised in the first and the second market tests with respect to access to raw milk.

1864. In particular in the first market test, many critical views (1062) have been expressed with respect to the guaranteed price which was understood to be one of the highest prices in North- West Europe since companies included in the benchmark are those trying to pay their farmers as much as possible. However, no clear views have been expressed on what would be a "fair" market price with respect to raw milk in the second market test (1063), given in particular the volumes at stake which are much larger than the volumes usually available in the spot market. Moreover, the guaranteed price appears to be the market price for stable supplies of raw milk; any cooperative or private purchaser that pays less is unlikely to attract farmers and secure supplies in the medium term. The notifying parties have addressed this point by lowering the guaranteed milk price by 1% during the first five years in the third commitment proposal submitted on 27 November 2008. Moreover, the purchaser would also have the possibility to finalize supply contracts with farmers having left FrieslandCampina through the incentive to leave scheme.

1865. With respect to planning and flexibility, respondents to the second market test unanimously put forward (1064) that a greater flexibility is needed with respect to the duration of the contract, a possible deviation from the volume required and the minimum order term. This is particularly necessary in the case of fresh dairy manufacture where products are produced overnight on daily orders from customers with strong fluctuations and no other outlets to divert raw milk are available. Therefore the purchaser must be able to vary his raw milk intake on short notice. This issue has been addressed by allowing purchasers to vary their weekly forecast by plus or minus 5% and by fine-tuning the minimum order term with respect to the quantities ordered.

1866. With respect to volume, the volume of 1.2 billion kg has been unanimously welcomed by respondents to the first and second market tests as allowing purchasers of the divestment businesses and other DPC plants to compete on a lasting basis against the notifying parties on the downstream markets. It would first cover the maximum capacity of both plants of the divested businesses (in total [...]* kg). The market test has confirmed that the remaining [...]* kg available would allow purchasers of the Divestment businesses to expand their business during the period of functioning of the Dutch Milk fund. It should also enable alternative players in the fresh dairy and cheese markets to source raw milk on competitive conditions to support further growth, thereby further contributing to the restoration of effective competition in the downstream markets for fresh basic dairy products and cheese in the Netherlands.

1867. With respect to duration of the arrangements involving the Dutch Milk Fund, respondents to the second market test indicated that the expiry date should depend when the newcomers have been able to constitute their own supply platform (1065). This concern is addressed by the fact that arrangements involving the Dutch Milk Fund are no longer limited in time and will continue to function until a number of farmers whose production of raw milk corresponds to the volume of raw milk made available through the Dutch Milk Fund have left FrieslandCampina.

1868. Finally, the preferential drawing rights granted to the purchasers of the divestment businesses for a volume equivalent to the current production capacity of both plants are necessary to ensure that these divestment businesses would be able to compete with the notifying parties on a lasting basis.

1869. In conclusion, the DMF remedy would allow the divestment businesses and other undertakings in the same markets to compete effectively against the notifying parties in the downstream markets.

17.4.3. Reduction of the exit barriers of the merged entity

17.4.3.1. Description of the commitment

1870. The notifying parties have also undertaken to reduce the exit barriers for members of the merged entity by granting to existing members of FrieslandCampina in the Netherlands a financial incentive to leave FrieslandCampina (the "Start Up Payment") and to enter into a supply arrangement of any kind with a certain minimum duration with a buyer of raw milk in the Netherlands. The amount of the Start Up Payment is 5 EUR per 100 kg raw milk delivered in the year immediately preceding the year in which the application for the Start Up Payment is made. 1871. Any member of FrieslandCampina may apply for the Start Up Payment by giving three months written notice to FrieslandCampina, provided it becomes a supplier to any buyer of Raw Milk in the Netherlands ("Buyer") for a period of three years. After the three months notice period the applicant farmer may start supplying a Buyer forthwith and his membership terminates at the end of the calendar year following the three month notice period. The Start Up Payment will be financed by FrieslandCampina and granted by DMF.

1872. Whenever a member terminates his membership with FrieslandCampina, the raw milk volume for which the Start Up Payment has been paid is withdrawn from the volume that should be made available by FrieslandCampina through DMF.

1873. Members that have exited FrieslandCampina may rejoin FrieslandCampina as a member. However, if a member re-accedes to FrieslandCampina within three years from the date he ceases supplying raw milk to FrieslandCampina or definitely stops supplying raw milk altogether, he will have to repay the exit fee on a pro rata basis. Moreover, the raw milk volume for which the Start Up Payment was paid will be added to the volume that should be made available by FrieslandCampina through DMF from the date the re-accession becomes effective. After three years from exit FrieslandCampina may apply the then applicable entrance fee to such former member that may wish to re-accede to FrieslandCampina. FrieslandCampina is not to offer terms to former members more attractive than those offered to new members applicable at the relevant time.

17.4.3.2. Assessment of the commitment offered

1874. The granting of an incentive to leave to farmers of the new entity has been welcomed by several respondents to the second market test. These incentives would indeed provide access for the purchasers of the fresh and cheese divestment businesses to the original sources of milk (farmers). These purchasers should be in a position to contact members of the cooperative societies and to offer alternative terms and conditions. This commitment provides for a more structural solution which enables the purchasers of the divestment businesses to secure their own supply base.

1875. Since the downstream divestiture of the downstream businesses is aimed at restoring effective competition on the downstream markets indefinitely, long term supply, independent of the merged entity should be secured. By granting an exit payment thereby reducing the barriers to exit for farmers who are members of the cooperative the long term supply of raw milk for the downstream divestment businesses may be secured.

1876. No clear views have been expressed in the second market test in respect of the amount of the incentives. Respondents rather underlined that the incentive should function so as to ensure that a sufficient number of farmers leave FrieslandCampina in order to ensure access to a free supply base. The financial incentive should be sufficient to stimulate a significant number of farmers, representing a significant volume of raw milk, to leave FrieslandCampina as members.

1877. The main issue which has been highlighted during the second market test is that leaving framers should have the possibility to rejoin Friesland/Campina. Otherwise, the prospect of not being able to return will function as an additional deterrent to leave and make the incentives less interesting. Also the fact that farmers can apply for financial incentives only until 1 January 2012 in the remedy proposal submitted on 19 November 2008 was considered unsuitable to ensure the success of the incentive scheme.

1878. The notifying parties have addressed these concerns by removing the termination date of 1 January 2012 to apply for the Start Up Payment and by allowing member farmers to rejoin provided that they pay back the Start Up Payment on a pro rata basis if they wish to rejoin within three years or the entrance fee if they wish to rejoin after three years. Members who cease supplying raw milk within three years after having been granted the Start Up Payment will also have to pay back this Start Up Payment on a pro rata basis.

1879. In the light of the above, it is concluded that the incentive scheme as proposed in the third commitment proposal of 27 November 2008 allows purchasers of the divestment businesses and current and potential competitors on downstream markets to secure adequate supplies on a lasting basis and that it therefore contributes to restoring effective competition in the markets concerned.

17.4.4. Overall conclusion with respect to access to raw milk

1880. It is concluded that the commitments put forward by the notifying parties with regard to access to raw milk (supply agreements, milk fund and incentive scheme) are sufficient to ensure, together with the commitments concerning fresh dairy products and cheese, that effective competition is restored in the markets for fresh dairy products and cheese in the Netherlands.

17.5. Long-life flavoured dairy drinks

17.5.1. Competition concerns identified

1881. As explained above, the proposed merger would significantly impede effective competition on the market for branded long-life chocolate-flavoured dairy drinks in the Netherlands, the market for branded long-life fruit-flavoured dairy drinks in the Netherlands, the market for branded long-life chocolate-flavoured dairy drinks in Belgium, the market for branded long-life fruit-flavoured dairy drinks in Belgium, the market for branded and private label longlife chocolate-flavoured dairy drinks in the Netherlands, Belgium and Germany and the market for branded and private label long-life fruit-flavoured dairy drinks in the Netherlands, Belgium and Germany, these three Member States being a substantial part of the common market, regardless of whether these markets need to be further segmented according to the distribution channel (see section 11.3.3.4).

17.5.2. Description of the commitments offered

1882. With respect to long life dairy drinks, the commitments offered consist of the divestiture of two brands covering both products markets of the LLDDs sector: Campina's chocolateflavoured brand Choco Choco and Campina's fruit-flavoured brand Yogho Yogho.

1883. The purchaser of the Yogho Yogho brand will have to grant an exclusive perpetual, royalty-free irrevocable license for the use of the Yogho Yogho brand outside the Netherlands and Belgium. The presence of the Yogho Yogho brand outside the Netherlands and Belgium is limited.

1884. The divestiture of Choco Choco and Yogho Yogho also includes full access to recipes and formulations, copyrights to packaging and customer records in the form of data on customers, all customer lists with contact details, overdue list with details per customers and credit terms.

1885. At the option of the purchaser, the divestiture of those two brands will include temporary two year cost-plus sourcing agreements and manufacturing technical assistance.

17.5.3. Assessment of the commitments offered

17.5.3.1. Effectiveness

1886. Choco Choco is Campina's main brand in the chocolate flavoured LLDD market. Currently the brand Choco Choco is used for two types of chocolate flavoured LLDD, a classic version and a low-fat version. The classic version is sold as "Choco Classic" and the low-fat version as "Choco Choco". The notifying parties submit that Choco Choco has generated in 2007 a turnover of EUR [...]* in Belgium (market share of [0-5]*% in the overall LLDDs market and [5-10]*% in the chocolate-flavoured LLDDs market) and EUR [...]* in the Netherlands (market share of [0-5]*% in the overall LLDDs market and [0-5]*% in the fruit-flavoured LLDDs market). This divestiture would therefore remove a substantial part of the overlap in chocolateflavoured dairy drinks in the relevant markets. It would also provide the purchaser with an Abrand of Campina that competes directly with Friesland Foods brand Chocomel.

1887. Yogho Yogho is Campina's main brand in the fruit flavoured LLDD market. Under the brand name Yogho Yogho, fruit flavoured LLDDs are supplied with different flavours and different packaging. Currently Yogho Yogho is supplied in six flavours: strawberry, raspberry, pear, peach, yellow fruit and red fruit. In addition to the 1 litre gable tops, Yogho Yogho is also available in 500 ml plastic bottles and 200 ml cartons. The notifying parties submit that Yogho Yogho has generated in 2007 a turnover of EUR [...]* in Belgium (market share of [0-5]*% in the overall LLDDs market and [0-5]*% in the fruit-flavoured LLDDs market) and EUR [...]* in the Netherlands (market share of [0-5]*% in the overall LLDDs market and [10-20]*% in the fruit-flavoured LLDDs market. This divestiture would therefore remove a substantial part of the overlap in fruit-flavoured dairy drinks in the relevant markets. It would also provide the purchaser with an A-brand of Campina that competes directly with Friesland Foods brand Fristi.

1888. The divestment of brands is sufficient in these particular markets to ensure that the commitments offered would eliminate the competition concerns. Brands are the main assets to be competitive in these strongly brand-oriented markets and given the limited volumes at stake, additional capacity is much easier to find or add than in the fresh dairy markets.

1889. For these reasons, the divestment of these two brands will be sufficiently effective to remedy the competition concerns in the long-life dairy drinks markets.

17.5.3.2. Independence, viability and competitiveness

1890. The market test confirmed that the brands Choco Choco and Yogho Yogho would constitute an attractive portfolio in the LLDD market since a purchaser would not have to bear the costs and lead time to build a new brand. However, a majority of respondents to the first market test also indicated that these brands should be divested given that the first commitment proposal consists in a divestment of Choco Choco and a perpetual license of Yogho Yogho. A divestment of Yogho Yogho would enable the purchaser to invest in order to ensure further development of the brand or a widening of the assortment of Yogho Yogho (1066). A licence would not form the basis of further investment and the feasibility of a perpetual license of the brand has been put into question during the market test. As already mentioned, upon the results of the market test, the notifying parties agreed to divest Yogho Yogho in the second commitment proposal submitted on 19 November 2008.

1891. Entry in the market for the purchaser of Choco Choco and the Yogho Yogho brands will be facilitated by the proposed sourcing agreement with the notifying parties. This will allow the potential purchaser to be immediately present in the Belgian and Dutch markets without any production disruption. Finally, the buyer will benefit from Campina's technical assistance to operate the migration of the production to its production lines. It has, however, been highlighted in the first market test (1067) that the sourcing agreement should be longer than one year, which was the period originally proposed by the notifying parties, given that especially Yogho Yogho is available in different flavours and packaging. Therefore a longer period is necessary for the purchaser to set up its own production lines or find available capacities to manufacture the product. Accordingly the notifying parties agreed in the second commitment proposal to extend the supply agreement for a period of two years.

1892. It is therefore concluded that the divested brands constitute an independent, viable and competitive business allowing the purchaser to compete effectively and on a lasting basis with the new entity on the chocolate-flavoured dairy drinks and the fruit-flavoured dairy drinks markets and therefore enabling effective competition on the markets concerned to be restored.

17.5.4. Conclusion on the commitments on long life dairy drinks

1893. On the basis of these elements, it can be concluded that the commitments submitted are sufficient to remove the competition concerns identified in the markets for chocolate-flavoured dairy drinks and fruit-flavoured dairy drinks.

17.6. Conclusion on the commitments

1894. In light of the above, it is concluded that the commitments, as submitted on 27 November 2008, ensure that the proposed merger would not significantly impede effective competition in the markets in which this Decision have identified competition concerns.

18. CONDITIONS AND OBLIGATIONS

1895. Under the first sentence of the second subparagraph of Article 8(2) of the Merger Regulation, the Commission may attach to its decision conditions and obligations intended to ensure that the undertakings concerned comply with the commitments they have entered into with the Commission, with a view to rendering the concentration compatible with the common market.

1896. The fulfilment of measures that result in a structural change of the market is a condition, whereas the implementing steps which are necessary to achieve this result are generally obligations on the notifying parties. Where a condition is not fulfilled, the Commission's Decision declaring the concentration compatible with the common market no longer stands. Where the undertakings concerned commit a breach of an obligation, the Commission may revoke the clearance Decision in accordance with Article 8(6) of the Merger Regulation.

1897. In accordance with this basic distinction, the Decision in this case is conditioned on the full compliance with Section B of the Commitments submitted by the notifying parties on 27 November 2008, whereas sections A, C, D, E and F of the Commitments are obligations within the meaning of Article 8 (2) of the Merger Regulation. The full text of the Commitments is annexed to this Decision and forms an integral part thereof.

19. OVERALL CONCLUSION

1898. For these reasons, the notified operation, as modified, should be declared compatible with the common market and with the EEA Agreement pursuant to Article 2(2) of the Merger Regulation, subject to compliance with the Commitments annexed to this Decision.

Has adopted this decision:

Article 1

The proposed concentration whereby the cooperatives Zuivelcoöperatie Campina U.A. and Zuivelcoöperatie Friesland Foods U.A. merge within the meaning of Article 3(1)(a) of Regulation (EC) No 139-2004 is hereby declared compatible with the common market and the functioning of the EEA Agreement.

Article 2

Article 1 is subject to full compliance with the commitments set out in section B of the Annex.

Article 3

Article 1 is subject to full compliance with the obligations set out in sections A, C, D, E and F of the Annex.

Article 4

This Decision is addressed to:

Zuivelcoöperatie Friesland Foods U.A. Blankenstein 142 7943 PE Meppel Netherlands

Zuivelcoöperatie Campina U.A. Hogeweg 9 5301 LB Zaltbommel Netherlands

ANNEX 1

ESTIMATION OF DEMAND ELASTICITIES

1. INTRODUCTION

1. How consumers view goods as substitutable determines the extent of competition between them. One way to summarise the extent of competition between products using measures called the 'own and cross price elasticities of demand'.

a) The own elasticity of demand measures the responsiveness of a product's demand to its own price (the own price elasticity is formally defined as the percentage change in demand for the product that would result from a 1% increase in the product's price).

b) The cross elasticity of demand measures the responsiveness of demand for one product, say product A, with respect to the price of a second product, say product B (the cross elasticity of demand for product A with respect to product B's price is formally defined as the percentage change in the demand for product A that would result from a 1% change in product B's price).

2. The larger the cross elasticity of demand between two products, the closer the two products are as substitutes in the eyes of consumers. Using standard econometric techniques, one can estimate a statistical relationship between the quantity of the good purchased and the price and promotional activities and other determinants of demand (such as the size of the market). This allows estimating the "matrix" of demand elasticities for a set of product categories or in some cases for individual brands.

3. In the context of this case, the parties have provided a number of data sets for a number of different product markets of 'aggregate-level retail scanner data'. Each data set provides information on price and quantity aggregated over individual consumers within a specified set of geographic areas or supermarket chains. Originally, this data has been collected by specialized firms (IRI and ACNielsen) from a sample of supermarkets by checkout scanning systems. These systems record information on every item that passes over the scanner as consumers make their purchases. IRI and Nielsen then processed the data and developed estimates of value sales, unit sales, and other variables by geographic area (e.g., subnational regions), time period (e.g., week), channel (e.g., supermarkets), and UPC code (e.g. 1 liter of Campina semi-skimmed milk). IRI and Nielsen then sell the resulting data to the product manufacturers, who then used it for market research purposes. Campina and Friesland both purchase on a regular basis scanner data from IRI and ACNielsen. For the purposes of estimating demand elasticities in the relevant markets of interest, in particular in the Netherlands, IRI data only has been used.

4. It should be stressed that an econometric analysis of scanner data is not only useful in the context of merger assessment. It is noteworthy that others, such as marketing professionals, also undertake similar analyses. For example, manufacturers of consumer products estimate systems of demand equations to help them determine optimal prices for their products. Clearly, scanner data, drawn from consumers' actual purchases, provides a wealth of information that can be used to describe and analyze consumer demand.

5. This Annex provides a more detailed presentation of the Commission's demand estimation results. The demand system estimation is based on the Almost Ideal Demand System (Deaton and Muellbauer, 1980). This model and its extensions have been extensively used in the empirical literature, as explained in more detail below.

6. In our exploratory analysis we have considered a number of alternative econometric specifications as well as alternative estimators. We report all results generated which can be considered useful in terms of providing evidence regarding the possible effects of the merger, or lack thereof.

7. There are many crucial choices to be made in constructing, estimating, and simulating econometric models. These choices can only be evaluated in the broader context of the specific case. With econometric and related qualitative materials, the whole is certainly greater than the sum of the parts.

2. PAST EMPIRICAL EVIDENCE ON ELASTICITIES FOR DAIRY PRODUCTS

8. The aggregate demand for dairy products is commonly considered as price inelastic. This, in part, is because dairy products are for human consumption and do not have many substitutes. Moreover, because the production is regulated by quota (the milk quota system has been prolonged until 2014), any change in the aggregate demand for milk can condition its price as quantity adjustment is restricted by the production quota. On the whole, results from different economic models suggest that, assuming a perfect transmission of price changes from upstream producers to downstream consumers, a 1% change in the aggregate domestic demand for milk causes a 3 to 4% change in milk price (1).

9. Due to the high sensitivity of milk price to demand, farmers' revenues strongly depend on the increase of aggregate demand for milk. For example, the INRA-Wageningen Consortium Study (2002) on dairy policy scenarios has shown that if the demand growth rate is 0.5% a year rather than 0.75% a year (as has been observed in the past), ten years later, the farm milk price is 5% to 7% lower and the producer surplus is decreased by 2 billion €.

10. If demand for dairy products is relatively inelastic at the aggregate level, this is not true at the product level because dairy products may compete among themselves. In addition, the changes in demand for the different dairy products are heterogeneous.

11. Previous studies have addressed the sensitivity of demand for dairy products with respect to prices and income in EU countries (2). Estimates of demand elasticity are often calculated using the Almost Ideal (AI) demand system (Deaton and Muellbauer, 1980) or extensions of its linearised version such as the Quadratic Almost Ideal demand system (Banks et al., 1997).

12. Table 1 presents a synthesis of the elasticities estimated in a number of recent studies, reviewed by Bouamra-Mechemache et al. (Nov, 2007) while Tables 2 and 3 present more detailed information on the estimates of price elasticity and income elasticity for each study.

<table>

13. As a rule, the demand for dairy products is rather inelastic as most of the studies report price elasticity lower than 1 (in absolute terms). According to these results, the demand for butter is the least elastic and the demand for fresh dairy products and cheese are the most elastic among dairy products. However, as Bouamra-Mechemache et al. (Nov, 2007) report, results vary significantly from one study to another and it is difficult to define the source of variation with any precision (methodology, period, type of data and type of elasticity that is computed). There are also country differences. 14. Note that there is only one study for the Netherlands by Burrell and Jongeneel (1999). It appears that the study relies on annual data spanning from 1973 to 1996. Unfortunately, the study is not recent and therefore has limited relevance for this case.

<table>

14. Note that there is only one study for the Netherlands by Burrell and Jongeneel (1999). It appears that the study relies on annual data spanning from 1973 to 1996. Unfortunately, the study is not recent and therefore has limited relevance for this case.

3. DATA MANAGEMENT AND AGGREGATION ISSUES

3.1. Data Description

15. In the IRI database every product is very precisely identified by a series of detailed information. The typical structure of the product dimension is the following:

(i) type of product: milk, yoghurt, vla, etc. (ii) segment: fresh or long lasting (iii) sub group: flavoured or not flavoured (iv) producer (or major brand) information: Campina, Danone, private label, etc. (v) brand information: danone activia, becel pro active, etc. (vi) vendor information: Muller NL, Campina NL, etc. (vii) number of packages sold in a box: 1CT, 6CT, etc. (viii) package size information: 1L, 500ML, etc. (ix) flavour: natural, chocolate, fruit, etc.

16. A given product, described by the previously presented information, is identified in the dataset by a specific code, called ean.

17. For every single ean, the level of sales (in terms of units, volumes and value) is reported. Moreover a variable also records the share of sales of each product which have been sold under promotion (3). Using these variables it is possible to identify which products have been sold under promotion and also the exact type of promotion. There are four different types of promotions which can be identified:

(a) Feature and Display (b) Feature Only (c) Display Only (d) Price Reduction Only

18. A number of other variables, based on previous aggregations, are present in the dataset, including "base sales" and "incremental sales", which relate to the expected level of sales in the absence of promotions of any kind and the increase in sales derived from promotional efforts, respectively. A handbook describing all the variables present in the dataset, their meaning and the way in which they have been computed has been provided by the parties (4).

19. The dataset also has a geographical dimension. Observations are collected at regional level, at supermarket level and at infoscan level. The first dimension contains all the products sold in each one of the five geographical regions in which the Netherlands is divided. The second dimension reports all the products sold in every supermarket and the last one aggregates over all supermarkets.

20. A simple graphical analysis has been implemented in order to compare geographical dimensions. In particular volume sales have been averaged across all products within a dimension and then the trends over time have been compared. The result shows that trends are exactly the same across all dimensions. However, while the absolute level of average volume sales are exactly the same in each period for infoscan and region, the levels observed in correspondence of supermarket are constantly much higher (almost doubled). After consulting with the parties and their economic advisors it has been clarified that (5): "The supermarket fields from IRI include data at various levels of aggregation. These are explained in the IRI usage description guide on page 72 (reproduced below for convenience). - The total "Infoscan supermarkets" field in the IRI data includes the total for the listed supermarkets. - The highest level of grouping includes Albert Heijn, Schuitema, Laurus and Superunie. - Except for Albert Heijn, each of the other supermarkets consists of groups of other supermarkets. E.g. Schuitema includes the C1000 group, and Laurus includes three supermarkets. Depending on the type of analysis that you want to do you might have more interest in using the groupings at the higher levels of aggregation, or at lower ones. The only possible disadvantage of using lower levels is that the sum of the parts will not always add up to the total. The reason for this is that a number of supermarkets do not allow their data to be released at the supermarket level. From what I remember when looking at the data this applies to a few of the supermarkets in the Superunie group. There might be a few others in the lower levels of groupings"

21. In our analysis we have chosen to aggregate at the level of the highest supermarket grouping. Hence, all the observations related to lower levels of aggregation have been deleted prior to analysis.

3.2. Data Management

22. The previously described information is available for every type of every product. Our analysis, however, is concentrated on the four fresh basic dairy products, that is: MILK, YOGHURT, VLA and BUTTERMILK as well as CHOCOLATE MILK, FLAVOURED MILKS, DRINKYOGHURT and QUARK. Only these products have been kept into the dataset. The dataset resulting by the elimination of the non-relevant products has then been divided into eight datasets, one for every product.

23. Each one of the product-specific dataset has then further been divided into three smaller datasets, on the basis of the geographic dimension. This further disaggregation allows analysing every product with information at region, supermarket or infoscan level. Every data management operation implemented since that operation onwards has been repeated on each of the 24 datasets.

24. After evaluating each of these data sets and in light of the priorities identified, QUARK has been dropped and the flavoured dairy drinks data sets has been separated according to "duration" and aggregated into two data sets: FRESH and LONG-LIFE FLAVOURED MILKS.

25. The next step of the adjustment of the data consisted in extracting all the relevant information summarized in a single variable called "description". After this operation, three new variables have been generated, one referring to package size, one to the number of packages in a box and the last one containing the information about flavour. Note that the information contained into the last variable is coded, so a new variable has been added to the dataset containing the meaning of the codes.

3.3. Analysis of the Data

26. The first step of the analysis consisted in some data cleaning in correspondence of the variable referring to the producer (major brand). Intervention was in particular required in correspondence of two observations: "private label" and "no brand".

27. In order to be able to distinguish between the different private labels, each one of them has been associated to its correspondent supermarket. If, for example, the variable "major_label", which identifies the producer, shows the value "private label" and we observe that the particular product is sold in an Albert Heijn supermarket, then the observation will become "private label Albert Heijn".

28. When, instead, the variable "major_label" assumed value "no brand", it has been tried to retrieve this information by the variable "vendor" (6). Whenever the link between the two variables was not clear, the observation was still classified as "no brand".

29. The second step of the analysis consisted in observing the distribution of package size, durability (i.e. whether the product is fresh or long lasting) and private_label (a variable assuming value 1 if the product is private_label and 0 if it is branded), across all observations but also across producers and supermarkets. The aim of this step is understanding which possible aggregations can be made across products. Also the distribution of prices has been analysed.

30. Notice that the distribution of packages and prices has been computed also taking into consideration product characteristics and durability. This could help to have information at a more detailed level (7).

31. As a third step, four dummy variables have been generated, each one assuming value 1 if the product was sold under that particular promotion (and 0 otherwise). A fifth dummy variable was created in order to asses whether a given product was sold under any promotion or not. A sixth variable, called "special pack", should assume value 1 if the product is sold under special package and 0 otherwise (8). It controls if the product is sold under Special pack, a main promotion category that could include the other categories previously described. This dummy, however, is always equal to zero, leading to conclude that this type of promotion is not present in our dataset.

32. The last step concerns the identification of flavours. A variable has been created which tells if a product is natural, if it has chocolate, vanilla, fruit or other flavour. If a product is fruit flavoured, then some dummy variables have been generated, informing about the specific type of fruit.

3.4. Data Aggregation

33. The IRI data is already aggregated across time (weekly), and is aggregated across retailers, for all of the Netherlands. In addition, it is necessary to undertake additional aggregations, to make the estimation tractable. For example, estimating demand for each individual size or variation of a given consumer product is generally not practical, and attempting to do so would often lead to imprecise parameter estimates. Aggregating the data requires us to make assumptions. These assumptions are described in this section since aggregation may affect the parameter estimates (9).

3.4.1. Aggregation across time

34. We have access to weekly scanner data from supermarkets. Such level of aggregation appears appropriate for the purposes of estimating demand elasticities in this case. Weekly data has two advantages over monthly or quarterly. First, because supermarkets tend to change their prices weekly (promotions typically last one or two weeks), weekly data most accurately relate consumer prices to their corresponding purchases. Second, the use of weekly data offers more observations, which increase the precision of the elasticity estimates.

35. However, when relating downstream demand by end-customers to upstream demand by retailers it may be necessary to consider that unlike retailers, dairy products producers or suppliers may not change their wholesale prices on a weekly basis. In addition, using weekly or even monthly data may overestimate elasticities if consumers often buy large quantities of items, which are on sale and take them into household inventories. In that case, the elasticities being measured are short run purchasing elasticities not the consumption elasticities, which tend to be more relevant for antitrust analysis. Consumer inventor behavior could lead to incorrect conclusions about the effects of a merger. One implication of consumer inventorying behavior is that a week with larger than normal demand (e.g., due to a sale) may be followed weeks with smaller than normal demand (as consumers deplete the inventories rather than purchasing at the full price). In this respect the inferences drawn from our results are likely to be more favorable for the parties and thus correspond to a conservative approach.

36. Nonetheless, there is little evidence in the data suggesting that such inventory effects are very strong with respect to the products considered in the present case. Most dairy products, in particular fresh dairy, are perishable and have only limited shelf life. Indeed, for items like fresh milk there is evidence that households make weekly purchases, in accordance to the household needs, which tend to be stable over time. In any case, for our purposes it is sufficient to consider the estimated elasticities as an upper bound for the true long-run demand elasticities.

3.4.2. Aggregation across Product Sizes and Varieties

37. For a given category, e.g., yoghurt, the number of individual products can be very large because each brand (e.g., Campina) might have many varieties e.g. different flavors. In addition yoghurts are sold in different package sizes, with the price per unit of weight or volume generally declining with package size. Specifying a demand system to account for all of individual products is not realistic. As a practical matter, when estimating a demand system, we are required to make some aggregation choices to minimize the number of parameters to be estimated. In some markets, aggregation across package sizes is not very important because most sales are made through one package size, e.g. sliced bread. However, in other markets significant volume is sold through multiple package sizes, e.g., soft-drinks. In the case of fresh milk, for example, most sales are in bottles or packs of 1 litre, 1.5 or 2 liters.

38. The degree of aggregation should be undertaken is determined by practical considerations and desire not to distort the econometric estimates. A good way to proceed is to test the effect of using different levels of aggregation within a range dictated by the practical considerations.

39. Consistent with some previous work on private labels (e.g., Slade 1995), in some specifications we also aggregate private label brands together. In such case, the private label share is the sum of all private label brands in the ith market, jth category. Private label price is the volume-weighted average price of all private labels in the ith market, jth category.

3.4.3. Aggregation at retailer-chain level

40. Scanner data used in estimating demand curves sometimes combines price and quantity data across retailers within a broad geographic area. This is the case in the "regional" data set provided by IRI. However, because retailers charge different prices, the price and quantity data aggregated across retailers do not represent the prices faced by any consumer.

41. A further problem arises because of non-price competition that varies across retailers within a market at a point in time. If a given chain charges approximately the same prices and offers similar promotions within a given area, chain-specific data should accurately reflect the price and non-price attributes facing consumers in a given week.

42. Retailer-chain data, as contained in the "supermarket" data set can reduce the problem of measurement error, provided that non-price competition, as can be expected, varies more across retailer chains than within retailer chains.. However, it is important to point out that the demand curve facing competing retailer-chains in a region is more elastic than the aggregate demand curve for the entire market. For example, assume there is a highly perishable product for which consumer demands are virtually inelastic (e.g., milk), and that some consumers are loyal to shopping at their favorite store, while others shop at the store with the lowest milk price. Assume further that supermarket chains compete for the non-loyal customers by offering milk at discounted prices. In the data, when a chain happens to have the lowest price for milk in a given week it will experience a surge in sales (because it captures some of the non-loyal consumers), while the total quantity of milk sold in the market is unchanged. Thus, the demand elasticity for milk estimated using chain data would be positive, while the market level elasticity (correctly measured using market level data) is zero, by assumption.

43. Note this can be relevant when using elasticity matrix to delineate markets. In such case, low cross price elasticities across categories are indicative of separate product market but not decisive in the absence of other qualitative evidence. The reason is that a small increase in the price of one product A may cause sufficient consumers to switch to another product B to render it unprofitable for the manufacturer of A to impose a unilateral price increase. However, the reverse need not be true. Indeed, asking whether one product is in the same market as another focuses only on the competitive significance of individual substitutes rather than on the collective competitive significance of all substitutes. For example when there are many product brands, cross elasticities between any pair of products may be quite small. Nevertheless, it may be the case that no individual brand has any significant market power because a small increase in its price would induce substitution to many other brands, with each of them gaining only a small fraction of the switching customers. However, also note that this is less of a concern when considering the cross price elasticities across aggregate categories of brands

44. The own price elasticity of demand of each category and not the cross price elasticity of demands are the critical parameter. The power to control price requires a low own-price elasticity of demand. Evidently if the demand for fresh milk is quite elastic with the respect to the price of fresh milk (the "own-price"), there is no significant ability to profitably raise price, since price increases will result in a large reductions in the sales of fresh milk (10).

45. In interpreting the own price elasticities for the purposes of market delineation it can be observed that under the simplest textbook case of linear demand and constant marginal cost a monopolist would set output at exactly half the level of a competitive industry. With nonlinear demand and constant marginal cost or linear demand and increasing marginal cost, a monopolist would select an output greater than half that of the competitive industry but not substantially greater (11). In general a monopolist is likely to select a level of output at least 30 per cent less than a competitive industry. Now suppose that a significant price increase is likely to lead to a drop in sales of less than 30 per cent. Such a fall in sales is likely to be insufficient to render the price increase unprofitable, since, in general, a profit maximising monopolist would be likely to set output at least 30 per cent below the competitive industry level. This implies that demand is sufficiently inelastic for a hypothetical monopolist to increase price by more than 5 per cent and no further products need to be added to define the relevant market. "Thus, a showing of inelastic demand is a sufficient (but not necessary) condition for demonstrating a relevant antitrust market." (see Kaserman and Zeisel, 1996, p. 679) (12). In the case that the demand elasticities have been estimated based on retailer-chain data it should be taken into account that the estimates provide an upper bound of the true own price-elasticity, thus a finding of a narrow market based on an inelastic demand estimate is conservative.

46. If the analyst is trying to infer the market level demand for a product and consumers are likely to change retailers to purchase that product given small changes in that product's price, it is preferable to use market level data aggregated by region. However, if consumers are unlikely to change retailers to purchase a product at a promoted price, then data aggregated within a chain is always preferable.

4. ECONOMETRIC METHODOLOGY

4.1. Specifying the Functional Form

47. To estimate demand, we must choose the mathematical formula (functional form) that expresses the demand relationship. The statistical estimation, itself, does not "choose" the functional form. In theory, there is a "true" functional form that generates the purchase data, and of course it is important that the functional form chosen for estimation approximates the "true" functional form. Capps, Church, and Love (2003) stipulate that a demand system for use in merger simulations must possess two characteristics: i. For each brand, the cross-price elasticities of the demand system are estimated from the data. A flexible functional form leaves the own and cross price elasticities of demand free to be estimated from the data. A nonflexible form, on the other hand, may impose restrictions on the demand elasticities, which can lead to biased results.; and ii. The elasticities are not constant, but vary as prices change, typically rising as price increases (Crooke et al., 1999).

48. In general, when choosing an econometric specification, a tradeoff exists between the flexibility of the specification to reflect the characteristics of the observed data and the statistical precision of the elasticity estimates. A less flexible specification generally has fewer parameters to estimate and thus may lead to more precise elasticity estimates. On the other hand, being less flexible, the specification may fail to fit the data well, which could induce bias into the elasticity estimates. In other words, the specification may fail to capture important characteristics of the data (13). Moreover, "flexibility" is often expensive: the number of price elasticities (own and cross) that must be estimated is equal to the square of the number of products considered. Even with several years of data, it could become difficult to estimate with acceptable precision the demand elasticities for a large number of products (14).

49. Under the economic theory of consumer choice, a demand system must satisfy three properties: Slutsky symmetry, homogeneity of degree zero in prices and total expenditure, and adding up: ?? Slutsky symmetry requires that the compensated cross price derivative of product A with respect to product B equals the compensated cross price derivative of product B with respect to product A. ?? Homogeneity of degree zero in prices and expenditure requires that demand for all products be unchanged if the prices of the products and total expenditure all increase by the same percentage. ?? Finally, adding up requires that the sum of expenditures on the individual products equals total expenditure.

50. Some demand specifications allow these properties to be easily imposed and tested, while other specifications do not. Generally, one would want to impose the restrictions implied by these properties because certain calculations of interest (e.g., consumer welfare calculations) would not be valid if the demand system did not satisfy the properties of consumer demand. On the other hand, empirical demand studies have often found that the properties of consumer demand are rejected by statistical tests. Thus, the ability to both impose and test the properties of consumer demand is valuable property for a demand system specification.

51. A second theoretical consideration relates to whether the demand system specification can be obtained by aggregation over individual consumers. A demand system and its associated properties are derived at the level of the individual utility-maximizing consumer. The question is whether the demand system and its properties transfer over to the aggregate-level data that is obtained by aggregating over individual consumers. In that case, the aggregatelevel demand can be treated as the demand of a 'representative consumer' and the estimated demand system should exhibit the appropriate properties.

4.1.1. Alternative specifications

52. As a matter of practice, assessing the validity of the functional form chosen for estimation is done through various statistical tests and testing alternative functional forms. There are two basic demand models commonly used in industrial organization models: ?? The representative consumer model and the address or location based model. The representative consumer model can be estimated by specifying the appropriate demand system (AIDS, Q-AIDS, Translog). ?? Address or location based models are usually estimated with logit, nested logit, or random coefficient discrete choice models. Discrete choice models solve the dimensionality problem. If one is estimating models of brand interactions, then the number of parameters to be estimated increases exponentially with the number of brands examined.

53. Discrete choice models do not appear entirely appropriate in this case because they implicitly assume consumers purchase only a single unit. This seems an especially limiting assumption given that retailers frequently conduct promotions, and these promotions create increases in the quantity purchased.

54. In contrast, the representative consumer model has strong theoretical underpinnings. However, this model is limited by the assumption of the representative consumer. That is, they do not allow for heterogeneous consumers. Aggregation can be a problem in these models, and unlike discrete choice models, there is an exponential expansion of the parameters to be estimated when the number of brands modeled increases.

55. Possible functional forms include the linear demand system and the log-log (i.e., constant elasticity) demand function.

4.1.2. Log-log functional form

56. A log-log demand system takes its name from the fact that the log of a product's quantity is related to the logs of the prices of all the products as well as the log of category expenditure. It is estimated by regressing the natural logarithms of the quantity variables on the natural logarithms of the price and demand-shifting variables. As a result the regression coefficients are the elasticities - no further computations are required. Specifically, under the log-log specification, the demand equation for product i is: 1 log log log( ) N i ij j j Q a ß Y ? p = = + +S (1)

57. Where Qi is quantity of the product i and Y is category expenditure, pj is the price of product j and a, ß and ?ij, are the parameters to be estimated. The log-log demand system is flexible in that it can approximate any demand system at a given set of prices. It is also relatively easy to estimate.

58. One important disadvantage of the log-log demand system is that it cannot guarantee that the parameters have the "right" signs. Indeed, the "adding-up" restriction of demand theory (i.e., the requirement that expenditure shares sum to unity) cannot be satisfied by the constant elasticity demand system. It is preferable to allow demand elasticities to vary as prices and quantities vary (i.e. demand becomes more elastic as one moves up the demand curve). Thus, although the log-log system might approximate a general demand system at the point of approximation, it may fail to approximate it well as one moves away from the point of approximation. Moreover, imposing the restrictions of consumer theory is not straightforward. In addition, the log-log system as applied to aggregate-level data cannot be obtained through aggregation over individuals.

4.1.3. The Almost Ideal Demand System (AIDS)

59. These disadvantages have led many researchers to use alternative functional forms in merger analysis. In particular, the Almost Ideal Demand System (popularly termed the "AIDS" model), first proposed by Deaton and Muellbauer (1980), is advocated by many as a basis for empirical merger analysis (15) and is well documented in the literature (16). Importantly, it has been extensively used in empirical work (17). In the AIDS model, expenditure shares of each product are regressed on the logarithms of the prices of the different goods and the log of total expenditure (deflated by a price index). In this section, we follow Hausman, Leonard, and Zona (1994) as well as Capps, Church, and Love (2003) and others who have used the AIDS specification.

60. The primary disadvantage is that it requires the estimation of a large number of parameters, and it does not guarantee (at least without further restrictions) that cross-elasticities have the "right" signs. Although AIDS allows elasticities to adjust as equilibrium prices and quantities vary, it restricts somewhat the way these adjustments take place. Consequently, the predictions of models estimated using the AIDS specification do not vary greatly from those estimated under an assumption of constant elasticities.

61. However, the AIDS model has many desirable attributes:

(a) It is an arbitrary first order approximation to any demand system. This result implies that even if the true underlying demand system is not AIDS, AIDS will nevertheless provide a reasonably accurate approximation at any set of prices not too far from the point of approximation. (b) it satisfies the axioms of choice in consumer theory (c) individual behavior can be aggregated to consistently estimate demand parameters from market level data (d) it is easy to estimate.

62. The downside to flexibility is the large number of parameters that need to be estimated. Even after imposing Slutsky symmetry and homogeneity of degree zero as described below, estimation of the most parsimonious flexible function form demand system (e.g., AIDS) with N products, will generally require the estimation of at least N2+3N-4/2 parameters. For example, a system with 10 products would have 63 parameters.

4.1.3.1.AIDS Specification

63. The AIDS demand model was introduced by Deaton and Muelbauer (1980). Instead of deriving the demand system directly from the first-order conditions of a representative consumer's utility maximization program they start from an underlying expenditure function: E(u,p), which represents the minimum amount of expenditures that is necessary to reach a particular utility level u given a vector of prices p. That expenditure or cost function is the result of a minimization program subject to the constraint that the utility function is set to a specific level.

64. The standard household utility maximization problem is:

V(p, M) = Maxq {U(q): p' q = M} (2)

with its associated dual expenditure minimization problem:

q E(p, u) = Min {p'q: U(q) = u} (3)

where q = (q1, ..., qN) is (Nx1) vector of goods, p = (p1, ..., pN) is a (Nx1) vector of prices for q, M denotes expenditure on the N goods, U(q) is the household direct utility function, and u is a reference utility level. The solution to (X) gives the Marshallian demand functions g(p, M), while the solution to (X) gives the Hicksian demand functions h(p, u) (18).

65. Deaton and Muellbauer (1980) apply this basic framework to derive the AIDS demand function. But they specify the Marshallian demand functions in terms of budget shares and they select a functional form E(u; p) in such a way that the demand function remains flexible and satisfies utility maximisation. More specifically, following Deaton and Muellbauer (1980b), assume that the expenditure function E(p, u) takes the general form: E(p, u) = exp[a(p) + u b(p)] where the function a(p) and b(p) are linearly homogeneous and represents the cost of subsistence and bliss respectively:

Where:

( ) 'ln 1ln( )' ln( ) 2 a p =d +a p+ p G p (4)

where 1 ( ,... ) ' N a = a a is a (N´1) vector and: 11 1 1 ... ... ... ... ... N N NN ? ? ? ? ? ? G =?? ?? ? ? ? ? is a symmetric (NxN) matrix and, 1 ( ) exp ln( ) N i i i b p ß p = = ??? ??? S (5)

66. Differentiating the log of expenditure function with respect to ln(p) generates the AIDS specification. Let i w denote the expenditure share of brand i, where i i i w p q M = and pi is the price of brand i, qi is the quantity of brand i demanded, and M is total expenditure on the group of brands.

67. Assume there are N brands to be included in the demand system, indexed by i=1,...,N. A product's revenue share is equal to the revenue generated by the product, divided by the total revenue generated by all products included in the demand system. Thus AIDS can be interpreted as a Marshallian demand system where the revenue or expenditure share is a function of total expenditures and prices. The revenue share of product i is then specified as: 1 ln ln( / ) n i i ij j i j w a ? p ß M P = = +S + (6) where: 0 1 1 1 ln( ) ln 1 ln ln 2 n i i ij i j i i j P a a p ? p p = = = = +S + SS (7)

and where wi is the revenue share of brand i, M is the total revenue over all the brands in the demand system, P is an overall price index for the products, pj, j=1,...,N are the prices of the brands, and ai, ßi, ?ij, are the parameters to be estimated.

68. Thus under AIDS, the revenue share of product i is the result of three terms. The first term is a constant ai that differs across brands. Everything else being equal, some brands will have higher shares as a result of consumer preferences. The second term is based on the real expenditure devoted to the category (M/P). The revenue share of product i increases as the total real expenditure on the category increases, if ßi the coefficient on the log(M/P) term is positive, and conversely if negative. The third term is based on the prices of the various products.

69. Although it is not necessary to have a well-behaved demand system (and often not observed in practice) it is helpful in understanding equation (6) to suppose that the own-price coefficient ?ii is negative and that the cross price coefficients ?ij are positive. In that case the share of brand i increases when its own price decreases or when the price of another brand increases

70. It is relatively straightforward to test whether the AIDS demand form satisfies the axioms of consumer demand theory (constrained utility maximization). The adding-up and homogeneity properties are the results of a linear budget constraint (19). In particular, the adding up property implies the following parameter restrictions for all i: 1; 0; 0; i i ij i i i Sa = Sß = S? = (8)

71. When consumers face a linear budget constraint, the adding-up conditions simply states that consumers are always utility-maximising and as a result the budget constraint always binds. The non-satiation axiom implies that consumers, facing a price increase in one good, will rearrange their optimal consumption basket and every euro will be spent. Adding-up therefore requires that the sum of the budget share equals one, 1 i i Sw = . This condition is satisfied when conditions (8) hold.

72. Homegeneity of the demand function simply states that there is no "money illusion". If there were a general price increase so that each relative price remains unchanged, then the composition of the optimal consumption basket would not be altered. In other terms, the demand function (Marshallian or Hicksian) is homogeneous of degree zero in prices. When the condition, 0 kj k S? = , holds, which is also part of the adding-up condition, then the demand function satisfies the homogeneity condition. A proportional change in price will not alter consumer purchases, and the budget share of good i remains unchanged.

73. Finally, the last condition imposed on the AIDS demand system satisfies Slutsky symmetry: ? ij = ? ji . This condition implies that the demand function is consistent with constrained utility maximisation.

74. The parameters in the AIDS demand equation indicates whether good i is a luxury good. That is, when expenditures rises so does the share of budget devoted to good i (ß > 0 ). Alternatively, when ß < 0 , the data suggest that good i is a necessary expenditure.

75. The above parameter restrictions can be imposed during estimation. Alternatively, the restrictions can be tested using standard statistical methods after estimation of the AIDS model.

4.1.3.2.LA(AIDS)

76. The non-linear nature of P means that in practice when an AIDS system is estimated, Stone's index (P*) is often used instead. The Stone index is: * 1 ln ln n j j j P w p = = S . If prices are highly collinear then P may be approximately proportional to P* such that P_ FP* . Then an approximation to (7) is given by: * ln ln( /*) i i ij j i j w=a +S? p+ß M P (9) Where: * log i a =a -ß F.

77. This equation is also known as the linearised AIDS demand model (LA/AIDS), which is an approximation of the original model. Equation (9) can be estimated using scanner data. Substituting P for the Stone price index leads to: 1 1 ln( ) ln( ) ln( ) n n i i ij j i i j j j j w a ? p ß M ß w p = = ? ? = + + - ? ? ? ? S S (10) Which is a system on N equations in N unknowns.

78. The reduced form expenditure shares are given by: 1 1 ln( ) ln( ) 1 ln( ) N N i j j i j j j i j i i N i i i x p x p w p ß ß ß ? ? = ? ? ? ? ? + ?- ? ? = ? ? ? ? + S S S (11)

Where: 1 ln( ) ln( ) N i i ij j i j x a ? p ß M = = +S + (12)

79. The Stone price index is a weighted average of the individual brand prices that uses the brands' revenue shares as weights. Typically, to avoid inducing endogeneity in the log price index, a fixed weight version of this index is used where the weights equal the average of the revenue shares over the entire time period. Furthermore the weights are typically allowed to differ across cities/regions in order to reflect differences in consumer preferences across geographies.

80. The use of the Stone index means that there is a number of alternative formulas for the own and cross price uncompensated or Marshallian elasticities.

4.1.3.3.Elasticity estimates in AIDS

3.1.1.1.1. Elasticity formulas

81. As shown by Green and Alton (1990) A general definition of the uncompensated elasticities of demand from the AIDS and LA/AIDS is: ln ln ln ln ln ln i i ij ij ij ij i i j j j d Q d w d P w d p d p d p ? d d ? ß = =- + =- +??? - ??? ?? ?? (13) where these elasticities refer to allocations within the group holding constant total group expenditures (M) and all other prices (pk, k = j), ij is the Kronecker delta and for the LA/AIDS we use P* from (3) instead of P from (2).

82. The differences in the literature can be represented in terms of different expressions for the elasticity of the Stone price index with respect to the jth price (i.e., ln ln j d Pd p ). These differences carry over directly into the computation of compensated elasticities.

83. Because the AIDS formulation approximates a Marshallian demand system, uncompensated price elasticities can be directly computed from the parameter estimates. The uncompensated price elasticities are generally derived as follows. The budget share of good i is i i i w p q M = . Taking the logarithms of both sides and differentiating with respect to the log of the price of brand j, and assuming constant expenditure, the cross-price elasticity of good i with respect to good j is then given by the general formulation ln 1 ln ln i i ij ij j i i e q p p ? d ? ? ? = =- + ? ? (14) Where ij d is the where is the Kronecker product delta ( 1 ij d = , if i=j; 0 ij d = if i ? j ).

84. The literature uses various specifications for price elasticities that mainly depend on how log log j P p ? ? is specified. The elasticity of demand resulting from using the true AIDS specified in (1) is given by: 1 ln n AIDS ij i ij ij j mj l i i m e p w w ? ß d a ? = = - + - ?? + ?? ? ? S (15)

85. The use of Stone's index means that there are a number of possible alternative formulas for the uncompensated own- and cross-price elasticities. One alternative is to use those of the AIDS model, a second alternative commonly used is an approximation that assumes expenditure shares remain constant, and a third alternative is to use the elasticities of the LA/AIDS model.

86. The issue of the "correct" elasticity to use when the LA/AIDS model is estimated has been considered in Green and Alston (1990), Alston, Fost, and Green (1994), and Buse (1994). The common approximation is method (iii) in Green and Alston (1990) and it holds exactly when expenditure shares are constant: dln(Pi)/dln(pi)=wi. The LA/AIDS elasticities when this assumption is made are: PS ij j ij ij i i i w e w w ? d ß ? ? = - + - ? ? ? ? (16)

87. Green and Alston (1990) advocate to use a more complex formulation (20) for the LA-AIDS: LA/ AIDS ij i( ln ( )) ij ij j m m mj mj i i e w w p e w w ? ß = -d+ - + S +d (17)

88. The issue of the correct elasticity to use when the LA/AIDS model is estimated has been considered in Green and Alston (1990), Alston, Foster, and Green (1994), and Buse (1994). As Buse notes the choice depends on whether it is the AIDS or LA/AIDS demand specification that is thought to be the correct model. If the correct model is LA/AIDS then the LA/AIDS elasticities should be used. However, the LA/AIDS models is not an integrable demand system and is typically used not because it represents the outcome of consumers' choices, but because it is not subject to the same convergence problems associated with estimating the AIDS model. It is thus an approximation, so the choice of which elasticity formula to use depends, presumably, on which formula using the LA/AIDS coefficient estimates yields the most accurate estimate of the AIDS elasticities.

89. The Monte Carlo results of Alston, Foster and Green (1994) suggest that both the Green and Alston approximation and the LA/AIDS elasticities provide reasonable estimates of the true AIDS elasticities when used in conjunction with LA/AIDS estimates. Buse (1994) argues that the common approximation gives the best estimates. He also argues that the AIDS elasticity formulas are preferable to the common approximates when the LA/AIDS estimates of the intercepts are adjusted appropriately. The adjustment requires information on subsistence consumption levels; information not likely to be readily available in most merger cases.

90. For the purposes of this statement of objections we report the matrix of elasticity estimates computed on the basis of Green and Alston approximation (equation 16).

3.1.1.1.2. Recovering compensated elasticities

91. Based on the uncompensated elasticities it is then easy to recover the compensated elasticities. From the Slutsky equation: ( , ) ( , ) ( , ) i i i j j j h u p g q p q g q p p q p ? ? ? = + ? ? ? (18) 92. We can rewrite the latter equation as follows: * ij ij i j e =e +ew where * ij e is the compensated cross-price elasticity and 1 i i i e w ß = + is the total expenditure elasticity. Unlike Marshallian price elasticities that contain both substitution and income effect, the compensated price elasticities only measure substitution. As a result, for unrelated goods we will always have * 0 ik e = while 0 ik e ? because of the income effect.

4.2. Issues in estimation

4.2.1. Wrong signs

93. In an AIDS system, the estimated cross-elasticities are not guaranteed to be positive and some may be negative, especially when the number of products is large. Negative crosselasticities can be a cause for concern because they are counterintuitive and because they can lead to odd results in consumer welfare calculations or merger simulations.

94. The first question to ask is whether the two products in question might, in fact, be complements rather than substitutes, in which case the true cross-elasticities would be negative. If the products should be substitutes, the next question is whether the estimates are statistically significantly different from zero if not, the negative estimated cross-elasticities should be of no concern unless they unduly affect subsequent calculations of interest, e.g., merger simulations. In that case, the cross-elasticity can be constrained to zero; however, one must proceed carefully once restrictions from economic theory have been imposed because these properties link the elasticities together.

95. If one or more cross-elasticities are estimated to be negative and statistically significantly different from zero, then the appropriate response depends upon the number of negative estimated cross-elasticities relative to the total number of estimated cross-elasticities. If the number of products is large, so that many cross-elasticities have bee estimated, it would not be surprising to find some negative statistically significant cross-elasticities. However, a relatively large number of negative cross-elasticity estimates would suggest a problem with the data or the model specification. The appropriate response would be to examine the data for errors and try different model specifications (e.g., add other variables to the specification or implement a different flexible functional form).

4.2.2. Endogeneity

96. Estimation of the LA/AIDS demand system could raise concerns regarding identification since prices are included as independent regressors. Specifically, there may exist some factors unobserved to the econometrician that affect both consumer demand and the pricesetting of firms. For example, if stores change prices during the period of data reporting in response to "unexpected" changes in the volume of purchases, we would say that prices are "endogenous," as opposed to a situation in which stores set their prices in advance of the data period and do not change prices during the data period, despite occasional unexpected changes in the volume of sales.

97. In that case, the prices appearing on the right-hand-sides of the AIDS system would be correlated with the error terms of these equations. Ordinary least squares or its variants (e.g., seemingly unrelated regressions, with cross-equation restrictions) would then be biased and inconsistent.

98. In general, the solution to the simultaneity problem is to employ an 'instrumental variables' technique. An instrumental variables technique involves finding variables (the instruments) that are correlated with the endogenous variables (in this case, prices), but not correlated with the error terms. Loosely speaking, one replaces the endogenous variables with the instruments and the simultaneity problem disappears (since the instruments are not correlated with the error terms, as were the endogenous variables). The question in implementing an instrumental variables technique is where to obtain the necessary instruments.

99. Good candidates for instrumental variables would be the prices of inputs used in the production of the product, since they are likely to be correlated with the supply curve (at retail) but not disturbances that shift the demand curve. Such data series are generally termed "costs shifters." In the context of demand estimation with scanner data, however, it is usually difficult to find enough cost shifters, since the analyst would need as many different costs shifters are there are prices in the demand system. To be useful instruments, such variables would have to be measured with a great degree of frequency and specificity (i.e., separately for the individual manufacturers). Cost variables measured monthly and at a national level would not ultimately be very helpful in estimating a demand equation based on the prices of N individual brands, measured weekly in a large number of cities. While plant-specific variable cost data for each manufacturer would be more helpful, having access to such data is rare.

100. Hausman, Leonard, and Zona (1994) have suggested two alternatives. The first is that if scanner data are available from different geographic markets, then price data from one geographic region can be used as instruments for other regions. This approach does not appear particularly adequate in the context of the present case since the necessary assumption that demand shocks across regions or retailer-chains (depending on the panel variable used) are independent is likely not met (21). The second consideration relates to the fact that prices may in fact not be endogenous, but instead are set by retailers prior to consumers making their purchase decisions. This assumption is consistent with the nature of scanner data: retailers set their price and at that price their supply is perfectly elastic.

101. Indeed, when using high frequency data it is often argued that prices are set before consumers make their purchase decision, and retailer supply is perfectly elastic. In other words, regardless of quantity demanded prices are ultimately determined by costs factors during the measurement period. As a result, prices are not correlated with the disturbance term of the demand equation. This is a widely accepted argument in the literature. For example, Capps et al. (2003) who assume that weekly scanner data implies that prices are not endogeneous. Again how realistic is this assumption depends on the specificity of each situation. This implies that endogeneity is unlikely to be an issue with weekly data (22).

102. We thus follow the literature on this point. We assume that, for the most part retailer chains do not adjust weekly prices to equilibrate demand and supply. Since shelf space tends to be sufficient supply curves are approximately flat on any given week. As a result prices can be treated as econometrically predetermined, thus obviating the need for instrumental variables estimation procedures (23).

103. Besides price and expenditure endogeneity, there be can two other possible sources of inconsistency and asymptotic bias in parameter estimates, they are: errors in variables, and omission of relevant variables. The IRI-Infoscan data used in our empirical analysis is directly collected from supermarkets scanners. Such scanner data are of high quality and reliability. Hence, we do not think that errors in variables are a serious problem in our analysis. Omitted variables, on the other hand, can be a potential source of problem in any econometric analysis. The fixed effects estimator controls for differences across the panel variable (in this case supermarket chains) which do not vary over time. The only relevant time-varying variable the omission of which could be potentially problematic are nonseasonal national advertising campaigns or other non-price promotions. Unfortunately we have no data concerning such promotions but we note that there appears to be a significant degree of seasonality on promotional campaigns the impact of which would be largely captured by seasonal dummies.

5. ECONOMETRIC MODEL AND RESULTS

5.1. Empirical Specification included in the Statement of Objections

104. With any theory and empirical analysis, the transition from theoretical model to empirical specification entails the need for careful variable selection. It is important to choose the empirical model carefully, using the appropriate hypothesis tests to guide final model specification. A key is specifying the set of variables that are consistent with the theory and a set of control variables that remove extraneous variation.

105. Empirical implementation of the AIDS is often possible when retail scanner data are available on brands for a number of cities and time periods. For example, in this case, the IRI dataset region includes data on a large number of products across 4 supermarket chains, in 5 regions for 156 weeks.

106. According to several authors (e.g. Hausman and Leonard, 2004) it is best to cast a wide net when choosing the products to include in the demand system because the purpose of estimating the demand system is to determine the extent of competition between products. It is better to be overly inclusive, letting the data decide the extent to which products compete. This is the approach followed in this report where in some instances, for example, Long Life Flavored or Fresh Flavoured milk we have included private labels in the estimation even if the retained market delineation based on complementary qualitative evidence extended only to branded products. We assume that data is available on N brands (indexed by i) for M supermarkets (indexed by m ) and T time periods (indexed by t).

107. The basic AIDS revenue share equation needs to be modified to account for the fact that a product's revenue share might differ across time and chains for reasons other than differences in price and expenditure. For example, consumer preferences for the product might grow over time or be seasonal. In addition, the typical consumer of one chain might have a greater preference for the product than typical consumers of other chains. To account for time-invariant differences in demographics or preferences across chains, separate constants are needed for each chain in the specification. To account for changes in demographics or preferences over time, time trend variables and seasonal variables in the specification are also included.

108. The baseline specification adopted for the purposes of this analysis, is the following: ln 1 1 1 2 , mt i t i t q m imt j imt ij jmt j w = S? p + ß g +t trend +t trend + c + c +e (19) where w and ln p are, as before, the expenditure shares and log prices, respectively. g is the real segment-specific expenditure, trend1 and trend2 are time trends before and after April 2007, respectively; cq and cm are the quarter and panel fixed effects, respectively, and e is the error term. The products are indexed by i, the panels by m and the weeks by t. The adding up conditions implies a singular variance-covariance matrix for the disturbances and this is handled in the conventional manner by deleting the nth equation.

109. The linear system of equations can be estimated by ordinary least squares (OLS). A more efficient method is the SUR (Seemingly Unrelated Regression) estimator which applies feasible generalized least squares (FGLS) to the system. That amounts to taking into account the correlation between the error terms of the different equations. The SUR estimator uses the variance-covariance matrix of residuals from the OLS estimated equations as the weighting matrix in the FGLS estimation. The SUR method can be iterated by applying FGLS using the previous step's residual variance-covariance matrix as a weighting matrix. When iteration is carried out until convergence of the parameter vector, the estimates are Maximum Likelihood estimates. (24)

110. The system can also be estimated by instrumental variables techniques if some of the right hand side variables are suspected to be endogenous. The two-stage least squares (2SLS) method in its first stage runs OLS regressions of the potentially endogenous regressors on the set of instruments and exogenous variables. In the second stage, the endogenous variables in the original system are replaced by their fitted values from the first stage. The OLS parameter estimates of this modified system are the 2SLS estimates of the original system's parameters. Three-stage least squares (3SLS) amounts to applying FGLS to the system based on the 2SLS residual variance-covariance matrix as weighting matrix. 3SLS can be iterated though this is not guaranteed to improve efficiency of the estimator over non-iterated 3SLS results.

111. Several possibilities arise to exploit the panel structure of the data. The two most important are the fixed effects (FE) and first difference (FD) estimators. (25) In the FE framework, the fixed effects cm are estimated by dummy variables. In the FD framework, all variables are first differenced and this transformed system is estimated:

mt i t i t q imt j imt ij jmt j ?w = S? ? ln p + ß ?g +t 1 ?trend1 +t 1 ?trend2 + ?c + ?e (20)

112. Here, ? denotes first difference (i.e., -1 ? = - t t t x x x ). Note that the panel fixed effects cm drop out as they are not time varying. Both the FE and the FD estimators can be implemented using the above mentioned methods (OLS, SUR, 2SLS or 3SLS).

113. As mentioned in Wooldridge (2002, p284-285), the choice between FD and FE hinges on the assumptions about the error term. In particular, if the FE estimator is more efficient if there is no reason to believe that the error term is serially correlated. On the other hand, the FD estimator is more efficient when the error term follows a random walk. Indeed Wooldridge points out that "[i]n many cases the truth is likely to lie somewhere in between".

114. In the context of our case, on the aggregate, there are not many reasons to expect to have residual autocorrelation. One possible source of autocorrelation is non-price related promotions extending over several weeks. In the data, we observe negligible volume of sales under non-price promotions, such as display and/or feature advertisement. The demand for branded products, such as Friesland and Campina, can be affected by national advertising campaigns even in those events when the advertised product belongs to different category. For instance, advertising for Friesland's Chocomel or Campina's Fristi could influence demand for Friesland's fresh milk. However, the IRI data does not contain information on national advertising campaigns. This motivates estimating both specifications.

115. The FD specification has some advantages over FE when instrumental variables estimation is applied. If any endogeneity is caused by significant contemporaneous correlation between some of the right hand side variables and the error term then lagged levels of the exogenous variables can potentially be used as instruments. Moreover, in the FD framework lagged differences and levels of the endogenous variables can also be used as instruments. (26)

116. Both, the panel data and the time series models are estimated in three different ways: (i) fixed effects model by iterated SUR; (ii) first differenced model by iterated SUR; and (iii) first differenced model by iterated 3SLS.

117. In the instrumental variables context the (log) prices on the right hand side are treated as potentially endogenous. The instruments include (i) lagged differences of each price series; (ii) lagged levels of each price series (from the second lag); (iii) lagged differences of the other right hand side variables; and (iv) contamporaneous and lagged levels of the other right hand side variables.

118. Eight AIDS systems are specified, six of them on the bottom level and two on the top level. On the bottom level there are four systems of basic fresh dairy products (milk, buttermilk, natural yoghurt and vla), one for fresh flavored products and one for long-life flavored products. In the case of the basic bottom level models, brands include an aggregate Campina, aggregate Friesland and aggregate or disaggregate private label brands. In the case of the fresh flavored model, brands include Mona, Optimel, other Campina, Friesland and the private label brands. The long-life bottom level model includes Bonomel, Yogho, Chocomel, Fristi and private labels. Other brands have been dropped in all bottom level specifications as they had negligible shares.

119. The top level model for basic fresh dairy products includes the aggregate segments of milk, yoghurt, buttermilk and vla. The flavored top level specification models the choice between fresh health, fresh regular, long-life chocolate and long-life fruit segments.

5.2. Empirical Results included in the Statement of Objections

120. Estimation results are presented in the Appendices. Appendix 1, 2 and 3 contain the parameter estimates. The corresponding elasticities are in Appendix 4, 5 and 6. Appendix 1 and 4 presents FE (or level) SUR, Appendix 2 and5 FD SUR and Appendix 3 and 6 FD 3SLS estimates/elasticities. In each of these Appendices, there are eight choice models as listed above (milk, buttermilk, natural yoghurt, vla, fresh flavored, long-life flavored, basic dairy and flavored segments). All of the eight models, with the exception of buttermilk, fresh flavored and the two top-level specifications, are estimated both with aggregated and disaggregated private labels specifications. Finally, Appendix 7 presents the results of the Hansen and Hausman tests for the instrumental variables models.

121. In general, the estimated models provide a good fit of the data as measured by the Rsquares. The estimated coefficients in almost all models are jointly significant and the important price parameters are precisely measured.

122. The Hansen test accepts the models' instrument set. The implication is that the issue of potential endogeneity of prices is handled. The Hausman tests compare the instrumental and non-instrumental variables estimates. In most cases the test does not find statistically significant difference. This implies that the non-instrumental variables estimates are unbiased and are preferred to the instrumental variables estimates as the former are more efficient.

123. Note that the FD models of milk, buttermilk and natural yoghurt which provide a surprisingly poor fit. Because of this reason, the results from these specifications will not be considered for the purposes of the analysis.

124. Note that only in the case of the FD framework were instrumental variables estimations carried out. However, it is still possible to infer about the validity of the FE models. Absent endogeneity problems, the SUR estimates of FD and FE specifications should not differ systematically. A consequence of this is that if a SUR-3SLS FD model pair passes both the Hansen and Hausman test then the FE SUR estimates are also valid.

125. The elasticity estimates provide the most important insights for the purposes of the analysis. The names of the rows are the same as those of the columns and the ordering is also the same. In each elasticity table, the diagonal elements are the own price elasticites. That is the percent change in the demand for the row's (column's) product as a response to a 1 percent increase in its price. The off-diagonal elements are the cross-price elasticities. The cross-price elasticity shows the percent change in the demand for the row's product as a response to a 1 percent increase in the column product's price.

126. The following patterns or "stylized facts" emerge from the elasticity structures.

a) Fresh milk. The parties tend to be each other's closest competitors. This is revealed especially by the disaggregate PL specification.

b) Natural yoghurt. The parties tend to be each other's closest competitors. Again, this is revealed especially by the disaggregate PL specification.

c) Vla. The parties' products are each other's closest competitors as they tend to have their highest cross-elasticites with each other. The next best substitutes are the high range private label products but the substitution is significantly less strong between these PL products and the parties' products than between the parties' products themselves.

d) Fresh flavored. The parties' products are each other's closest competitors and the PL products provide a less strong competitive constraint on them. The parties' products have their lowest cross correlations with the PL brand (in each row of a given Campina of Friesland brand the lowest number tends to be in the PL columns).

e) Long-life flavored. The parties' products are each other's closest competitors and the PL products provide a less strong competitive constraint on them. The parties' products tend to have their lowest cross correlations with the PL brand. Chocomel seem to be the strongest competitive constraint on each product. Yogho and Fristi tend to be especially constrained by Chocomel. The chocolate flavored PL brands are closer substitutes of the parties' products than the fruit flavored PL brands.

f) Top level estimates: Basic dairy. Own price elasticites are low (almost always below 1). Cross-price elasticites are usually weak. The inelastic demand and low cross-substitution suggest that the segments are separated markets. The lowest cross-elasticites are in the column of buttermilk. Vla has the highest own price elasticities which suggest that this product category is less basic than the others and has some outside substitute.

g) Top level estimates: Flavored drinks. The fresh health segment has the lowest own price elasticites and is the least constrained by the other segments. It can be considered as a separate market. The long-life chocolate and long-life fruit segments have the highest own price elasticites and provide strong competitive constraints on each other.

h) Buttermilk. There is no specification that simultaneously provides (i) a reasonable fit of the left hand side variables and (ii) economically meaningful elasticity matrix (in particular, negative own-price elasticities and mostly positive cross-price elasticities). For the purposes of the present case, the econometric results can be considered inconclusive.

5.3. Response of the Parties to the Statement of Objections

127. In their response to the Statement of Objections the parties include a submission by RBB (annex 4.1) intended to rebut the Commission's findings concerning market definition as well as the competitive effects of the merger in the segment of Long-Life Dairy Drinks (LLDD). Annex I attached to RBB's submission also provides a criticism of the econometric analysis included in the Statement of Objections. The emphasis is placed on the use of the AIDS model described above to delineate the boundaries of the market in the market for LLDD, and in particular, the question of whether chocolate flavored and fruit flavored drinks are part of the same market. Nonetheless a number of methodological criticisms are raised which, according to RBB, would also apply more generally.

128. For the reasons explained below we regard as well founded some of the criticisms put forward by RBB concerning the appropriateness of the AIDS specification to assess the boundaries of the market in the LLDD market. This leads us to reject the econometric results we presented in Statement of Objections for this market. This means that, for the purposes of this decision, the elasticity matrixes presented in the Statement of Objections in the section for LLDD are given no weight in the Commission's assessment of the competitive effects of the merger in this market.

129. RBB also claims that the econometric methodology chosen by the Commission is generally flawed and thus the results concerning other Fresh product markets should also be dismissed. With respect to markets other than Long-Life Dairy Drinks we find that the concerns raised by RBB are either invalid or where necessary can be appropriately corrected. However, the corrected models do not necessarily lead to similar qualitative results as those presented in the Statement of Objections for the Fresh basic product markets.

130. In this section we report and assess the criticisms raised by RBB. We first discuss the criticisms which apply generally to the use of the AIDS model to delineate the boundaries of the market. Next we evaluate the criticisms which apply more specifically to the LLDD market. Finally we discuss a number of criticisms which are more general in nature and which, according to RBB, affect the econometric results for all product markets.

131. Further, in the section that follows we provide extensions to the econometric model presented in the SO. These extensions resolve the valid concerns with the econometric results raised by RBB.

5.3.1. Criticisms concerning the use the of AIDS model to delineate markets

132. RBB argues the AIDS demand system can shed no light on the definition of the relevant market in the context of the proposed merger, for the following reason: "In the context of the AIDS model it is therefore necessary to make assumptions that have an appreciable impact on the result. This property is the main reason why AIDS models are often not useful to inform market definition. Specifically, the need to restrict substitution patterns a priori as well as the need to aggregate information as a result of the modelling choice is problematic"

133. RBB refers to two problems which it labels (i) "restrictive substitution patterns" and (ii) "restrictive aggregation of products". The exposition of the arguments is somewhat vague, moreover RBB fails to recognise that arguments are in fact intimately related and are not independent. On balance, however, we believe that the concern raised by RBB has merit.

5.3.1.1."Restrictive substitution patterns"

134. With respect to the claim that the AIDS model is inappropriate because it imposes "restrictive substitution patterns" it appears that RBB is simply pointing out that the choice of what product aggregates to include or exclude in a top level demand specification for the purposes of establishing whether such product aggregates belong to the same market or not is arbitrary. Moreover, adding or excluding certain product aggregates can lead to different conclusions concerning product market definition. In particular RBB claims that "the extraordinary structure that is required to impose substitution patterns makes the AIDS specification put forward by the Commission inappropriate for market definition as it requires to a significant extent to know the answer one seeks to estimate beforehand".

135. This criticism emerges from a misunderstanding of the purpose, and consequently the relevance of the two "top level estimations" put forward by the Commission, one for Fresh Basic Dairy and one for Flavored Drinks. It is recognised that the selection of segments to include in either of these specification is made "a priori". But this "a priori" selection of product aggregates is far from arbitrary. It is based on a qualitative assessment of which product aggregates are potentially part of the same market.

136. It is not claimed in the SO that the purpose is to perfectly delineate the boundaries of the market for all dairy products and thus determine in the context of a full model which product categories belong with each other and which are separate from the rest. In fact such claim would be absurd. As it is well known this is simply not possible since market definition, particularly when following the SSNIP methodology, is sensitive to the first candidate market of choice, as well as the market which is considered the next best substitute (if this is at all possible (27)). For example it may not be possible to impose a SSNIP on market A as many customers would substitute to B. This would lead to aggregating A and B. However, it may be possible to impose a SSNIP on B alone, keeping A and B as separate markets. The starting candidate market is thus determined on a case by case basis depending on the question of interest. For example one would often start with a narrow candidate market where the merging parties significantly overlap.

137. In fact, the two top level demand estimation included in the SO are context specific. Their only purpose is to assess the claims made by the parties that: (i) in the case of fresh basic dairy, Fresh milk, Fresh yoghurt, Vla and Buttermilk are part of the same market. (ii) in the case of LLDD, LL chocolate flavoured drinks and LL fruit flavoured drinks belong to separate markets.

138. In the case of fresh basic dairy the results confirm the qualitative evidence as well as the descriptive analysis of the data that the four fresh basic dairy products can not be considered to form part of the same market

139. In the case of LLDD, note that, as mentioned above, the goal is not delineate the boundaries of these market but rather to establish whether or not the claim made by the parties that LL chocolate flavoured drinks and LL fruit flavoured drinks belong to separate markets has merit. (28) This is done on the basis of comparisons across the product categories "included" in the model, under three maintained assumptions: (i) that none of the excluded product categories (that is non-flavoured dairy drinks or non-dairy drinks) are part of any candidate market involving one or more of the included products. (ii) the Fresh Health segment forms a separate market from Other fresh flavoured products (iii) fresh and long-life flavoured products also do not form part of the same product market

140. With respect to the second and third qualitative assumptions on Flavoured Dairy products, it has been established on the basis of descriptive analysis as well as qualitative evidence that the Fresh Health segment forms a separate market from Other Fresh Flavoured products. It has also been established that fresh and long-life flavoured products do not form part of the same product market. Concerning the first assumption RBB argues that "it is unclear on which basis the Commission has apparently concluded that fresh dairy drinks are the most likely next best substitutes to LLDD, and could hence provide for a benchmark but other drinks are not. For the parties, fruit juices and other still soft drinks are equal or even better candidates for inclusion in a broader relevant market than fresh dairy drinks, implying that the Commission's benchmark may be wrong".

141. First, it would appear that Fresh Flavoured dairy drinks share a greater number of commonalities with LLDD than fruit juices or still soft drinks. In this respect it would appear reasonable to consider them as relatively closer substitutes. In any event, precisely because they are considered as close but still part of separate market their inclusion in the model is justified as they provide a useful benchmark. Indeed the econometric model does not purport to shed light which product categories are the next best alternative for LLDD. The econometric model is intended only to provide a first indication of whether LL chocolate and LL fruit flavoured drinks provide (on the aggregate) an alternative to each other from the perspective of the end customer, relative to other product categories included in the model, which, a priori, it has been established that belong to separate markets. If, as claimed by RBB, fruit juices were a relatively close substitute to LL choco and LL fruit flavoured drinks, such that it would belong to the same market this would not provide a useful benchmark for comparison. In any event the parties have provided no qualitative, or for that matter, quantitative evidence suggesting that fruit juices or soda drinks are the next best alternative to LLDD.

142. Second, as shown in Table 10-7 in the SO, reproduced below, the own price elasticity of Fresh Health and Fresh Other is -0.6 and -1.5 respectively. Using these as a benchmark it can be seen that that LL Choco and LL Fruit have much higher own price elasticities (-2.4 and - 2.2 respectively). This suggests that under the maintained assumptions, customers are more price sensitive or have more alternative as regards these products than as regards Fresh flavoured products. When assessing the cross price elasticities it can also be observed that, relative to Fresh flavoured drinks (health sector or otherwise) end consumers are much more prepared to consider Long Life Choco as an alternative to Long Life Fruit and vice-versa. This does not exclude that consumers might also be prepared to other alternatives not included in the model, but it suggests that consumers perceive Long-Life Flavoured as closer substitutes to each other than to Fresh Flavoured products.

<table>

143. In conclusion, RBB's criticism that one has to impose a particular substitution patterns on the AIDS model "which requires to a significant extent to know the answer one seeks to estimate beforehand" is misplaced. It is of course the case that "a priori" but justifiable decisions are made concerning which product categories are good candidates to belong to the same product market. This knowledge comes from a qualitative assessment of the products in question. But the objective of the exercise is much narrower than RBB cares to admit. The only questions that we seek to answer is whether compared with and relative to other product categories which are (a priori) believed to be part of separate markets, LL Choco Drinks and LL Fruit flavoured drinks can be considered to be part of one market or not.

144. According to RBB this is also the case irrespective of whether one is estimating a full multi-stage budgeting system or simply trying to determine whether a hand-picked number of products belong or not to the same market. More concretely RBB argues that "Especially in the context of market definition, multi-level budgeting is undesirable as it requires a very strong prior about how products should be grouped. Even in order to test if the segment structure is appropriate requires a priori knowledge about the closeness of products, a quantity that should be determined by the model and not feed into the model as input".

145. In more general terms, it is important to emphasize that the AIDS model itself does not restrict substitution patterns. In fact, its flexibility, in terms of the implied elasticity matrix, is the main advantage of the AIDS specification over the more restrictive logit, nested logit or log-log models. This general observation on flexibility holds for both top and bottom level AIDS models. The implication is that the substitution patterns in a given top level model, for instance basic dairy, are not restricted. In this respect, the criticism of RBB applies only to the choice of segments/brands included in a given specification. As explained above, these choices can be driven by other qualitative evidence.

5.3.1.2."Restrictive aggregation of products"

146. RBB also criticises that the AIDS model reduces the number of products by aggregating brands and products of different sizes into product categories to ensure tractability. RBB suggests that a "characteristics framework" would have been more appropriate. RBB explains that in a characteristics framework one does not estimate substitution between products but substitution between characteristics that define products. These models that have been widely used over the last two decades exploit differences in product characteristics to estimate substitution patterns. Essentially RBB seems to prefer discrete choice models as a way to solve the dimensionality problem (see section 4.1.1)

147. However, as mentioned in section 4.1.1 discrete choice models do not appear entirely appropriate in this case because they implicitly assume consumers purchase only a single unit. In reality concerning dairy products consumers tend to purchase more than one unit in proportion to the size of their household as well as the number of visits the make to the supermarket each month and the expiry date of the product. RBB argues that there exist econometric techniques to correct for this difficulty and therefore this is not a valid reason to dismiss the use of discrete choice models in this case.

148. A further, more practical reason is that the available data set does not include sufficient information concerning product characteristics to implement a discrete choice model. For example information on important characteristics such as ingredients, positioning in the shelf, level of stock, type of label, expiry date, packaging type etc, are not available. In fact, other than flavour and packaging size the data set contains no key characteristics to distinguish between one product/brand and another. It is noteworthy that RBB, despite arguing that a discrete choice model would have been more desirable did not provide any submissions following this methodology.

149. RBB fails to explore that the estimation of a discrete choice model involving multiple choice (i.e., a framework in which the consumer can chose to buy more than one product at the same time) would require individual level data or detailed measures on the distributions of individual level variables. An example is provided by Hendel (1999). Hence, mainly due to the lack of available data, the Commission's choice of a non-discrete choice model framework is validated. It is also interesting to note that these multiple discrete choice models are less standardized, less widespread and considerably more complicated than simple discrete choice models.

150. RBB claims that "to artificially assume that pack size does not matter and therefore considering prices per litre is therefore unsubstantiated and likely to lead to misleading results". Further they argue that "even if one were to agree that pack size does not matter and consumers view products with different serving sizes as perfect substitutes - a view that we do not subscribe to - aggregation is still problematic due to mix effects. As Hosken et al. (2002) point out"

151. First, the Commission does not assume that pack size does not matter. In fact the Statement of Objections includes, for each product market for which data was available, a descriptive assessment of the relative shares according to package size. It is shown that concerning fresh basic dairy the large majority of sales are in one or two litre bottles or in the case of yoghurt and vla in one liter gable top. The product mix in terms of package size is in fact relatively constant (i) across supermarkets, (ii) across brands and (iii) over time, as RBB itself can attest. Thus, with respect to fresh basic dairy, aggregation by package size would appear unproblematic. Concerning Long-Life drinks there is more variety in package sizes and it is possible that the results would be more sensitive to aggregation. In any case there are more persuasive reasons to attach no weight to the econometrics results concerning LL Dairy Drinks than this one, as explained below.

152. RBB also claims that the results are sensitive to slight changes in the level of aggregation over time. They argue that if one considers the results based on monthly data or even quarterly data the elasticities differ significantly from the Commission's estimates. Typically, aggregating to monthly data results in estimated elasticities deviating from those based on weekly data by about one third and aggregating to quarterly data results in a change of almost two thirds. RBB claims that this is clear evidence that the estimated elasticities are heavily reliant on the method of aggregation used. This, however, is not surprising, since these sets of elasticities are not the same. Elasticities computed on the basis of weekly data are shortterm elasticities, whereas elasticities computed on the basis of monthly or quarterly data are medium or long-run elasticities. There is absolutely no reason to expect these elasticities would have the same value. Nonetheless, the issue is not whether the elasticity estimates differ, but which is the time frame of analysis that is more relevant in the context of the markets considered. The Commission's assumption, which appears reasonable, is that for fresh basic dairy markets customers make their purchasing decisions on a weekly basis, not a monthly or quarterly basis. This is so, inter alia, because fresh basic products are perishable and normally last much less than a month. In the case of LLDDs, however, the relevant time frame might be longer due to the long-life nature of the products.

5.3.1.3.Inference of wholesale behaviour using retail elasticities

153. RBB also raises a common criticism that concerns the use of retail level elasticities to inform wholesale or producer behaviour. RBB argues that "it is at best doubtful to what extent this can inform the wholesale elasticity, the necessary quantity to inform the merger appraisal". RBB cites a well known paper by Hosken et al. (2002), where the authors decompose the elasticity faced by the manufacturer, E(M), into the elasticity of demand facing the retailer, E(R), and the elasticity of the retail price with respect to the wholesale price, E(p,w), via the following relationship E(M) = E(R) * (w/p) * (pass-through rate) = E(R) * E(p,w). If this second elasticity term is one, for example if retailers keep a constant percentage margin, then the wholesale elasticity will equal the retail elasticity. However, in general we would expect the elasticity of the retail price with respect to the wholesale price to be less than one, in which case the wholesale elasticity will be less than the retail elasticity".

154. RBB further argues that "there is no a priori reason to suppose that all products considered by the Commission are sold at equal mark-ups (i.e. fresh v. long-life products, etc.). Equally, there is no reason to suppose that the average supermarket margins throughout the Netherlands or Belgium country are equal. To what extent the retail elasticity differs from the relevant manufacturer elasticity is an empirical matter. However, to the extent that there is a significant difference this would imply that markets delineated using the Commission's methodology will tend to be too narrow and should therefore be expanded".

155. It should again be stressed that the purpose of estimating the top-level elasticity matrices (for fresh basic dairy and for Flavoured drinks) was only to test the claims made by the parties that in the former case all products belong to the same market and in the latter that LL chocolate flavoured drinks and LL fruit flavoured drinks belong to separate markets.

156. Concerning Fresh Basic dairy the evidence provided by the parties, as well as the descriptive analysis presented in the relevant section, suggests that retail markets for fresh milk, fresh yoghurt, buttermilk and vla are highly competitive, with very thin margins for retailers as well as producers (with a strong presence of private label particularly in fresh milk). It would thus appear reasonable to make the assumption that the pass-through rate is high and it margins across supermarkets are unlikely to differ significantly, due to the competitive pressure that supermarkets exert on each other particularly with respect to fresh basic dairy products.

157. Concerning Long-Life Dairy Drinks note that precisely the argument made by the parties is that LL chocolate and LL fruit flavoured drinks should be considered as part of a separate narrow market. Thus, if anything, the use of elasticity estimates in this case to inform this issue would be a conservative test of the parties claim. In any event it is not necessary to explore the question of whether retail margins and wholesale margins differ significantly in the case of LLDD since there are other reasons to dismiss the elasticity estimates presented by the Commission in the SO for this market, as discussed in the next section.

5.3.2. Criticisms of the econometric analysis in the LLDD presented in the SO

5.3.2.1.Lack of robustness to changes in segment definition

158. RBB points to a technical difficulty with the use of the AIDS model. They estimate the Commission's model for long-life flavoured drinks and add a product that is not part of the candidate relevant market and which is also not similar in terms of characteristics. The model contains 5 product groups as per the Commission's analysis, namely Bonomel, Yogho Yogho, Chocomel, Fristi and Private Label. In addition to these long-life flavoured drinks, Boerenvla (a custard product) is added into the segment. The results show that all elasticities with respect to Boerenvla (i.e. the shaded cells in the table below) are significant at 1% using the Commission's methodology and generally (with the exception of private label) have the sign that would suggest that the product is a substitute. Moreover, the results suggest that Boerenvla is most branded products' closest competitor. The results suggest that Boerenvla's own-price elasticity is -3.18 and the cross-price elasticity of Bonomel with respect to Boerenvla is 0.516. Hence the Commission's model predicts, for instance, that Bonomel's closest substitute is Boerenvla which is clearly counterintuitive.

159. RBB concludes that "due to the severe problem of finding substitution between products when there is none - a direct consequence of the Commission's modelling assumption - the elasticities implied by the model are not reliable and we do not believe that they can inform market definition or the closeness of substitution between products". Further RBB argues that "the restriction that market shares sum to 100% implies that if a product is falsely included in the group the model will find significant substitution even when there is none in reality".

160. As explained above, the selection of brands within a given bottom level segment is not arbitrary. In fact, the segmentation into milk, buttermilk, vla, flavored drinks and so on is dictated by qualitative evidence. The main goal of the Commission's bottom level modelling exercise is to measure the cross elasticities between the brands in a given segment.

5.3.2.2.Failure to estimate long-term substitution patterns

161. RBB remarks that the Commission's analysis is conducted using weekly sales data which fluctuate significantly due to promotions. As Hosken et al. (2002), points out "[i]f inventory effects are important (this is likely to be the case if the predominate source of price variation are the sales which generate inventory effects), the estimated elasticities will likely be too large".

162. RBB argues that the Commission's claim that there is no stocking behaviour due to the limited shelf life of dairy products does not apply "at very least when one considers long-life dairy products". RBB also points out that "there are several other characteristics about the nature of promotions that could lead to an overstatement of the own-price elasticity. Specifically, sales surge in response to a promotion not only due to stocking behaviour but also as a result of (i) impulse purchases and (ii) the design of the promotion. Impulse purchases are defined as sales by consumers who would otherwise not purchase any product in the entire segment. The design of promotions is such that it is geared to boost sales of a given product (for example in the case of bundles).

163. RBB shows that the sales evolution of long-life dairy products is such that one often observes a significant, sometimes more than three-fold increase in sales of one product while the sales of the other products are unaffected. This suggests that the sales do not result from business stealing from other products but are market expansions. Moreover, there is often no dip in sales equal to the surge in the following periods which is what one would expect if the surges were entirely due to stocking behaviour. As a result of stocking behaviour, impulse purchases and the design of promotions, the own-price effect for some products is likely to be overstated in the Commission's model.

164. We believe RBB's line of reasoning is sound as regards Long-Life Dairy Drinks. Indeed the evidence highlighted by RBB that sales increase significantly in periods of heavy promotion, with limited apparent substitution suggests that customers engage in inventory behaviour or impulse purchases with respect to long-life dairy drinks. This appears reasonable since, the key characteristic of Long-Life products is that they can be stocked and last for months, unlike fresh dairy products. It follows that the assumption maintained in the SO that the customer make purchasing decisions in each week which are independent of their purchases on the previous week is unlikely to be met in the case of Long-Life Dairy Drinks. It is for this reason that the results presented in the Statement of Objections concerning LLDD cannot be given any significant weight.

165. Having said that, it is important to note that inventory or impulse behaviour is not likely in the case of fresh basic dairy or fresh flavoured drinks. This is for two reasons: First, as argued above fresh dairy products are perishable and have expiry dates often shorter than one or two weeks. This makes it difficult to purchase large quantities to keep in stock. Second brands of fresh basic dairy products, in particular, fresh milk, do not experience any significant promotions, price or otherwise, comparable to those observed for Long-life dairy drinks. This is the case across all supermarkets and throughout the period of analysis. As a result this criticism, whilst relevant in the case of LLDD it is not applicable for Fresh basic dairy, or Fresh Flavoured Drinks.

5.3.3. Specification problems concerning AIDS model

5.3.3.1.Spurious findings due to non-stationarity

166. RBB remarks that the Commission relies mostly on its findings based on the fixed effects specifications and that "the FD models of milk, buttermilk and natural yoghurt [...] provide a surprisingly poor fit." RBB further argues that this is likely due to non-stationarity. RBB shows that a non-stationarity test of the errors of the fixed effects model confirms this. They present the table below which shows that for every fixed effects model reported by the Commission at least one equation does not pass Fisher's unit root test.

<table>

167. RBB argues that based on this finding any elasticity based on the fixed effects specifications must be treated with great caution as it may be a purely spurious finding.

168. Stationarity is the stability concept of time series econometrics. If a series is stationary then it has a constant mean over time around which it fluctuates: it does not have a trend (it does not explode or implode). Stationarity of a given series can be tested. The Fisher test is a special test for panel data where a given variable consists of several time series. In its Null hypothesis, the test assumes that all the stacked time series in the panel variable are nonstationary. Rejection of this Null implies that at least one of the series is stationary.

169. Non-stationarity of the estimated error terms of an econometric regression does reveal specification problems. In fact, econometric models assume that the error term is stationary. The tests reported by RBB for the fixed effects models in most cases do not reject the Null of non-stationarity, so revealing a specification problem for these estimations. This problem requires considering extensions or alternatives to the model specification.

170. One way to deal with non-stationarity is to estimate the model using the first difference method. The estimated error terms from these specifications are not detected to be nonstationary by the Fisher test. As mentioned in the SO, the problem with these models is that with some exceptions they tend to provide a poor fit of the data and so can not be regarded as a reliable alternative despite the lack of non-stationarity problems.

171. It is also interesting to note that RBB makes a connection between the poor fit of the first difference models and non-stationarity. This connection is not valid. In fact, all of the FD specifications presented in the SO pass the stationarity test. The lower R-squares in the cases of the FD specifications are a completely different matter. It is explained by the fact that the dependent variable in any FD model is the change in the level of market shares. This makes natural that the fit in these models is less strong than that of the fixed effects models which fit the level of the market shares.

172. Another way to fix the specification problem is to set up dynamic models in which the lagged dependent variables are added to the set of explanatory variables. This approach might resolve the non-stationary problem even in a fixed-effect framework. In addition and as explained below, this dynamic model can be given an economic interpretation as representing habit formation effects.

5.3.3.2.Failure to pass basic diagnostic checks

173. RBB also remarks that none of the Commission's specifications pass Ramsey's (1969) RESET test, a common specification test, at the 5% level. This test suggests therefore that the Commission used the inadequate functional form of its model. Misspecification of a model has serious consequences as necessary conditions for inference such as consistency and unbiasedness are violated. Any calculations based on parameters of such a model are likely to give unreliable results.

174. RBB's use of the RESET test is misguided. RESET is purely a functional form test and it is not to be used detecting omitted variables bias. (29) The test is able to detect some un- modelled non-linearity. As explained above, the linear AIDS model is a first order flexible form approximation of the true data generating demand system. This fact can explain why some omitted non-linearity can be shown by the test. In the same time, it does not lead to biased estimates as long as the first order approximation holds in the narrow vicinity of the parameters of the true, not approximate data generating model.

5.3.3.3.Undue use of linear trends

175. Finally RBB criticises that in all specifications the Commission includes two linear time trends without any motivation of why a linear trend may be necessary to explain share variation over time. The use of a linear trend when the depended variable is bounded - such as a market share which lies between zero and 1 by construction - is problematic as negative market share predictions and market shares greater than 100% are possible. This property is clearly problematic when the parameters of the model are used to make predictions about market shares.

176. The Commission does not intend to predict out-of-sample market shares. Hence, the problem of negative predicted market shares does not apply. The goal of the econometric modeling exercise is to estimate the magnitude of own and cross price elasticities which hold on average in the sampled period. Including a time trend is a way of improving the in-sample fit by including 'time effects'. This is the approach followed by Hausman et al. (1994).

5.4. Dynamic extension of the model: Introducing habit persistence

177. Though not mentioned by RBB, one should contemplate the possibility that in the case of Fresh dairy products there might be some degree of habit persistence. To this effect we have considered an extension to the basic AIDS specification presented in the SO, to include dynamics that account for the possibility of habit persistence.

178. Habit persistence refers to the case when the consumer is accustomed to the purchase and consumption of a given brand and, eventually, his/her consumption pattern of the brand becomes less sensitive to changes in its price. The existence of habit persistence in itself does not render econometric modelling theoretically infeasible. However, if it is not controlled for it may cause specification problems. This problem might be the cause of the non-stationarity detected in some of the model specifications that is described above. As a consequence, the dynamic extension of the model can not only be economically meaningful but it may also be able to resolve the specification problems.

179. Following Alley et al. (1992) (30) the model is the following: ln ln , , 1 , 1 m,t 1 imt j mt mt im t ij jm t j j imt ij jmt j w = ? p + ß g +d x +? w + f p +? g +e - - - S S (21)

180. Here the vector x represents the exogenous shifters (trends, fixed effects). Habit persistence is introduced in the model by adding the lagged value of the product's market share as well as the lagged values of all products prices and the real segment expenditure.

181. The implied elasticity matrices of the estimation results as well as diagnostic tests are presented in the Appendix 8-9. The results show that the dynamic model does indeed remove the non-stationarity problem. The Fisher test finds significant non-stationarity only in two equations (one in the vla and an other in the fresh flavored segment). (31) The implied elasticity matrices are economically meaningful, the own price elasticities are negative and the crossprice elasticities, with a very few exceptions, are positive. The elasticities are generally significant.

182. However, there is a main qualitative difference in the results with respect to those of the SO. (32) The elasticity matrices, with the exception of the vla and fresh flavored segments, do not show that the merging parties' brands exercise significantly stronger competitive constraint on each other than the private labels. In this sense, the pure brand effect of the parties on each other seems to be on par with the competitive constraint exercised by the private label brands.

6. CONCLUSIONS

183. The Commission has attached to its Statement of Objections a set of econometric results on estimation of retail demand systems. These results were additional to another, larger set of qualitative and basic quantitative evidence. The goal of the econometric models was twofold. First, to shed more light on specific product market definition questions; and second, to support the claim that the merging parties' brands are providing on each other a significantly stronger competitive constraint than the private label brands.

184. The parties provided criticism of the econometric models on both of these accounts. Concerning market definition, after careful review of the arguments the Commission decided that they have merit and, on balance, it does not attach weight to the econometric results. The market definitions presented in the Decision can be supported by other evidence and can be maintained without the econometric results.

185. Regarding the relative competitive constraints, the parties' arguments are partly acceptable, partly can be replied to by suitable modifications of the model specification. In particular, the argument that the econometric framework proposed in the SO for the LLDD segment is not capable of satisfactorily depicting the structure of the demand has merit. As a consequence, the Commission does not regard the econometric results in its Decision on these issues as evidence.

186. Regarding the fresh basic dairy econometric models, though the parties' criticism is partly well founded it can be addressed by suitably modifying the specification. In particular, introducing habit persistence in the form of a dynamic model can satisfactorily improve the diagnostic properties of the estimation results. The Commission argued in the SO that the qualitative results of the econometric models showed into the same direction as the other, qualitative and quantitative, evidence. This is no longer true for the modified results as they are rather inconclusive. It is important to emphasize that the econometric evidence from the modified specifications does not contradict the theories of harm put forward in the Decision. In this sense, the econometric modeling exercise provided a one sided test of the whole set of theory of harms: It either contributed to this set as an add-on or remained indifferent. On balance, the Commission's assessment is that the second case applies. As a consequence, the Commission attaches no weight to the econometric results in its Decision.

7. REFERENCES

Alley, A. G., D. G. Ferguson and K. G. Stewart (1992), An Almost Ideal Demand System for Alcoholic Beverages in British Columbia, Empirical Economics, 17., 401-418.

Alston, J. M., Foster, K.A. and Green, R. D. (1994). Estimating Elasticities with the Linear Approximate Almost Ideal Demand System: Some Monte Carlo Results. Review of Economics and Statistics, 76(2), 351-56.

Banks, J., R. Blundell, A. Lewbel. 1997. Quadratic Engel Curves and Consumer Demand. The Review of Economics and Statistics,Vol. LXXIX, N°4.

Baker, Jonathan B. and Rubinfeld, Daniel L.,Empirical methods in antitrust litigation: review and critique. American Law and Economics Review, Vol. 1, No. 1, pp. 386-435, 1999 Available at SSRN: http://ssrn.com/abstract=874160

Bouamra-Mechemache, Réquillart, Soregaroli and Trévisiol (November 2007) Demand for dairy products in the EU,

Burrell, A. , R. Jongeneel. 1999. Demand for Dairy Products: A Case Study for the Netherlands- A report on work in progress for FAIR5-PL97-3481, Novembre 3, 1999- University of Wageningen.

Capps, Oral, Jr., Jeffrey Church, and H. Alan Love (2003). "Specification Issues and Confidence Intervals in Unilateral Price Effects Analysis." Journal of Econometrics, 113, 3-31.

Christen, M., S. Gupta, J.C. Porter, R. Staelin, and D. R. Wittink. (1997). Using Market-Level Data to Understand Non-Linear Promotion Effects. Journal of Marketing Research 34(3):322- 334.

Cotterill, R.W. Y Haller, L.E. (1996). "Evaluating traditional share -price and residual demand measures of market power in the catsup industry", Review of Industrial Organization, 11 (3), pp. 293-306.

Cotterill, R.W. Y Putsis Jr., W.P. (1999). "Share, price and category expenditure - geographic market effects and private labels", Managerial and Decision Economics, 20, pp. 175-187.

Cotterill, R.W.; Putsis, Jr., W.P. Y Dhar, R. (2000). "Assessing the competitive interaction between private labels and national brands", Journal of Business, 73 (1), pp. 109-137.

Crooke, Philip, Luke Froeb, Steven Tschantz, and Gregory J. Werden (1999). "The Effects of Assumed Demand Form on Simulated Post-Merger Equilibria." Review of Industrial Organization, 15, 205-17.

Deaton, A., J. Muellbauer. 1980. Economics and Consumer Behavior. Cambridge University Press.

Dhar, S.K. Y Hoch, S.J. (1997). "Why store brand penetration varies by retailer", Marketing Science Institute, Working Paper, Julio, pp. 97-114.

Edgerton, David L. 'Weak Separability and the Estimation of Elasticities in Multistage Demand Systems'. American Journal of Agricultural Economics, 1997, 79(1) 62-79.

Green, R. D. and Alston, J. M. 1990. Elasticities in Aids Models. American Journal of Agricultural Economics, 72(2), 442-45.

Greene, W. H. 2003. Econometric Analysis, 5th es. Upper Saddle River, NJ: Prentice Hall.

Hausman, et al. (1994) 'Competitive Analysis with Differentiated Products', 34 ANNALES D'ECONOMIE ET DE STATISTIQUE (1994),

Hausman, (1997)'Valuation of New Goods Under Perfect and Imperfect Competition', in Timothy F. Bresnahan and Robert J. Gordon (eds), The Economics of New Goods

Hausman and Leonard (1997), 'Economic Analysis of Differentiated Products Mergers Using Real World Data', 7 GEO. MASON L. REV.

Hausman and Leonard (2005) Competitive analysis using a flexible demand specification Journal of Competition Law and Economics 1: 279-301.

Igal Hendel (1999), 'Estimating Multiple-Discrete Choice Models: An Application to Computerization Returns.' The Review of Economic Studies, Vol. 66, No. 2 (Apr.), pp. 423-446

Hoch, S. J., et al. "Determinants of Store-Level Price Elasticity." Journal of Marketing Research, XXXII, no. February(1995): 17-29.

Hosken, Daniel, Daniel O'Brien, David Scheffman, and Michael Vita (2002). "Demand System Estimation and its Application To Horizontal Merger Analysis." Federal Trade Commission, Bureau of Economics, Working Paper (246), available at http://www.ftc.gov/be/workpapers/wp246.pdf.

INRA-Wageningen. Consortium 2002. Study on the impact of future options for the milk quota system and the common market organisation for milk and milk products. European Commission, DG Agri. Summary report. http://europa.eu.int/comm/agriculture/publi/reports/milkquota/index_en.htm

Kaserman, D.L., and H. Zeisel, 1996. "Market Definition: Implementing the Department of Justice Merger Guidelines", Antitrust Bulletin, Vol. 41, No. 3 (Fall), pp. 665-690.

Moschini, G.. 1995. Units of Measurement and the 'Stone Index' in Demand System Estimation. American Journal of Agricultural Economics, 77: 63-68.

Nevo, A. (2000) "Mergers with Differentiated Products: The Case of the Ready-to-Eat Cereal Industry,"

Food and Agricultural Policy Research Institute (FAPRI). 2006. FAPRI Agricultural Outlook. Center for Agricultural and Rural Development, Iowa State University. Available at: http://www.fapri.org/outlook2006/text/outlookPub2006.pdf

Rubinfeld, D. (2000) "Market Definition with Differentiated Products: The Post-Nabisco Cereal Merger" Antitrust Law Journal, Volume 68, Number 1, 2000, pp. 163-185

Slade, M. E. 1995. Product rivalry with multiple strategic weapons: An analysis of price and advertising competition. Journal of Economics and Management Strategy 4, no. 3 (Fall): 445-76.

Sheffman, D. (1991). Statistical Measures of Market Power: Uses and Abuses. In The Cutting Edge of Antitrust: arket Power, American Bar Association, An Econometric Analysis for RTE Cereal Section of Antitrust, Willard Intercontinental Hotel, Washington, D.C., October 17-18

Feleke S. and H. Liu (1995), Aggregate Deand for Imported Whole Milk in Spain: Implications for the European Union (EU), Journal of Food Distribution Research, 36(2), 20-28.

Werden, et al. (1996) 'The Use of the Logit Model in Applied Industrial Organization', 3 INT. J. ECON. BUS.

Werden, G.J., (1992). "Four Suggestions on Market Delineation", Antitrust Bulletin, Vol. 37, No.1, (Spring), pp. 107-121.

Wooldridge, J. M. 2002. Econometric Analysis of Cross Section and Panel Data, MIT

Wooldridge, J. M. 2003. Introductory Econometrics. A modern approach, Thomson, South- Western.

Case COMP/M.5046 - Friesland / Campina

Appendices of Annex 1

1. APPENDIX 1: FIXED EFFECTS SUR ESTIMATES

Fresh basic dairy

1.1.1. Milk

Table 2: AIDS model with aggregated private labels iterated SUR estimates; robust p-values in parentheses; dropped equation: private label

Table 3: AIDS model with disaggregated private labels iterated SUR estimates; robust p-values in parentheses; dropped equation: mid-range private label

1.1.2. Buttermilk

Table 4: AIDS model with aggregated private labels iterated SUR estimates; robust p-values in parentheses; dropped equation: private label

1.1.3. Natural yoghurt

Table 5: AIDS model with aggregated private labels iterated SUR estimates; robust p-values in parentheses; dropped equation: private label

Table 6: AIDS model with disaggregated private labels iterated SUR estimates; robust p-values in parentheses; dropped equation: mid-range private label

1.1.4. Vla

Table 7: AIDS model with aggregated private labels iterated SUR estimates; robust p-values in parentheses; dropped equation private label:

Table 8: AIDS model with disaggregated private labels iterated SUR estimates; robust p-values in parentheses; dropped equation: mid-range private label

1.2. Fresh flavored

Table 9: AIDS model with aggregated private labels iterated SUR estimates; robust p-values in parentheses; dropped equation: private label

1.3. Long-life flavored

Table 10: AIDS model with aggregated private labels iterated SUR estimates; robust p-values in parentheses; dropped equation: private labels

Table 11: AIDS model with disaggregated private labels iterated SUR estimates; robust p-values in parentheses; dropped equation: fruit flavored private label

1.4. Top level estimates

1.4.1. Basic dairy: milk, yoghurt, buttermilk and vla segments

Table 12: AIDS model on segment aggregates iterated SUR estimates; robust p-values in parentheses; dropped equation: vla

1.4.2. Flavored drinks: fresh health, fresh other, long-life choco and long-life fruit

Table 13: AIDS model on segment aggregates iterated SUR estimates; robust p-values in parentheses; dropped equation: long-life fruit

2. APPENDIX 2: FIRST DIFFERENCE SUR ESTIMATES

2.1. Fresh basic dairy

2.1.1. Milk

Table 14: AIDS model with aggregated private labels iterated SUR estimates; robust p-values in parentheses; dropped equation: private label

Table 15: AIDS model with disaggregated private labels iterated SUR estimates; robust p-values in parentheses; dropped equation: mid-range private label

2.1.2. Buttermilk

Table 16: AIDS model with aggregated private labels iterated SUR estimates; robust p-values in parentheses; dropped equation: private label

2.1.3. Natural yoghurt

Table 17: AIDS model with aggregated private labels iterated SUR estimates; robust p-values in parentheses; dropped equation: private label

Table 18: AIDS model with disaggregated private labels iterated SUR estimates; robust p-values in parentheses; dropped equation: mid-range private label

2.1.4. Vla

Table 19: AIDS model with aggregated private labels iterated SUR estimates; robust p-values in parentheses; dropped equation private label:

Table 20: AIDS model with disaggregated private labels iterated SUR estimates; robust p-values in parentheses; dropped equation: mid-range private label

2.2. Fresh flavored

Table 21: AIDS model with aggregated private labels iterated SUR estimates; robust p-values in parentheses; dropped equation: private label

2.3. Long-life flavored

Table 22: AIDS model with aggregated private labels iterated SUR estimates; robust p-values in parentheses; dropped equation: private label

Table 23: AIDS model with disaggregated private labels iterated SUR estimates; robust p-values in parentheses; dropped equation: fruit flavored private label

2.4. Top level estimates

2.4.1. Basic dairy: milk, yoghurt, buttermilk and vla segments Table 24: AIDS model on segment aggregates iterated SUR estimates; robust p-values in parentheses; dropped equation: vla

2.4.2. Flavored drinks: fresh health, fresh other, long-life choco and long-life fruit

Table 25: AIDS model on segment aggregates iterated SUR estimates; robust p-values in parentheses; dropped equation: long-life fruit

3. APPENDIX 3: FIRST DIFFERENCE 3SLS ESTIMATES

3.1. Fresh basic dairy

3.1.1. Milk

Table 26: AIDS model with aggregated private labels iterated 3SLS estimates; robust p-values in parentheses; dropped equation: private label

Table 27: AIDS model with disaggregated private labels iterated 3SLS estimates; robust p-values in parentheses; dropped equation: mid-range private label

3.1.2. Buttermilk

Table 28: AIDS model with aggregated private labels iterated 3SLS estimates; robust p-values in parentheses; dropped equation: private label

3.1.3. Natural yoghurt

Table 29: AIDS model with aggregated private labels iterated 3SLS estimates; robust p-values in parentheses; dropped equation: private label

Table 30: AIDS model with disaggregated private labels iterated 3SLS estimates; robust p-values in parentheses; dropped equation: mid-range private label

3.1.4. Vla

Table 31: AIDS model with aggregated private labels iterated 3SLS estimates; robust p-values in parentheses; dropped equation private label:

Table 32: AIDS model with disaggregated private labels iterated 3SLS estimates; robust p-values in parentheses; dropped equation: mid-range private label

3.2. Fresh flavored

Table 33: AIDS model with aggregated private labels iterated 3SLS estimates; robust p-values in parentheses; dropped equation: private label

3.3. Long-life flavored

Table 34: AIDS model with aggregated private labels iterated 3SLS estimates; robust p-values in parentheses; dropped equation: private label

Table 35: AIDS model with disaggregated private labels iterated 3SLS estimates; robust p-values in parentheses; dropped equation: fruit flavored private label

3.4. Top level estimates

3.4.1. Basic dairy: milk, yoghurt, buttermilk and vla segments

Table 36: AIDS model on segment aggregates iterated 3SLS estimates; robust p-values in parentheses; dropped equation: vla

3.4.2. Flavored drinks: fresh health, fresh other, long-life choco and long-life fruit

Table 37: AIDS model on segment aggregates iterated 3SLS estimates; robust p-values in parentheses; dropped equation: long-life fruit

4. APPENDIX 4: ELASTICITIES (FIXED EFFECTS SUR ESTIMATES)

4.1. Fresh basic dairy

4.1.1. Milk

Table 38: AIDS model with aggregated private labels

Table 39: AIDS model with disaggregated private labels

4.1.2. Buttermilk

Table 40: AIDS model with aggregated private labels

4.1.3. Natural yoghurt

Table 41: AIDS model with aggregated private labels

Table 42: AIDS model with disaggregated private labels

4.1.4. Vla

Table 43: AIDS model with aggregated private labels

Table 44: AIDS model with disaggregated private labels

4.2. Fresh flavored

Table 45: AIDS model with aggregated private labels

4.3. Long-life flavored

Table 46: AIDS model with aggregated private labels

Table 47: AIDS model with disaggregated private labels

4.4. Top level estimates

4.4.1. Basic dairy: milk, yoghurt, buttermilk and vla segments

Table 48: AIDS model on segment aggregates

4.4.2. Flavored drinks: fresh health, fresh other, long-life choco and long-life fruit Table 49: AIDS model on segment aggregates

5. APPENDIX 5: ELASTICITIES (FIRST DIFFERENCE SUR ESTIMATES)

5.1. Fresh basic dairy

5.1.1. Milk

Table 50: AIDS model with aggregated private labels

Table 51: AIDS model with disaggregated private labels

5.1.2. Buttermilk

Table 52: AIDS model with aggregated private labels

5.1.3. Natural yoghurt

Table 53: AIDS model with aggregated private labels

Table 54: AIDS model with disaggregated private labels

5.1.4. Vla

Table 55: AIDS model with aggregated private labels

Table 56: AIDS model with disaggregated private labels

5.2. Fresh flavored

Table 57: AIDS model with aggregated private labels

5.3. Long-life flavored

Table 58: AIDS model with aggregated private labels

Table 59: AIDS model with disaggregated private labels

5.4. Top level estimates

5.4.1. Basic dairy: milk, yoghurt, buttermilk and vla segments

Table 60: AIDS model on segment aggregates

5.4.2. Flavored drinks: fresh health, fresh other, long-life choco and long-life fruit

Table 61: AIDS model on segment aggregates

6. APPENDIX 6: ELASTICITIES (FIRST DIFFERENCE 3SLS ESTIMATES)

6.1. Fresh basic dairy

6.1.1. Milk

Table 62: AIDS model with aggregated private labels

Table 63: AIDS model with disaggregated private labels

6.1.2. Buttermilk

Table 64: AIDS model with aggregated private labels

6.1.3. Natural yoghurt

Table 65: AIDS model with aggregated private labels

Table 66: AIDS model with disaggregated private labels

6.1.4. Vla

Table 67: AIDS model with aggregated private labels

Table 68: AIDS model with disaggregated private labels

6.2. Fresh flavored

Table 69: AIDS model with aggregated private labels

6.3. Long-life flavored

Table 70: AIDS model with aggregated private labels

Table 71: AIDS model with disaggregated private labels

6.4. Top level estimates

6.4.1. Basic dairy: milk, yoghurt, buttermilk and vla segments

Table 72: AIDS model on segment aggregates

6.4.2. Flavored drinks: fresh health, fresh other, long-life choco and long-life fruit

Table 73: AIDS model on segment aggregates

7. APPENDIX 7: SARGAN/HANSEN AND HAUSMAN TESTS

Table 74: p-values from Sargan/Hansen and Hausman tests for panel models

8. APPENDIX 8: DYNAMIC MODEL, ELASTICITIES (FIXED EFFECTS SUR ESTIMATES)

8.1. Fresh basic dairy

8.1.1. Milk

Table 75: AIDS model with disaggregated private labels

8.1.2. Buttermilk

Table 76: AIDS model with aggregated private labels

8.1.3. Natural yoghurt

Table 77: AIDS model with disaggregated private labels

8.1.4. Vla

Table 78: AIDS model with disaggregated private labels

8.2. Fresh flavored

Table 79: AIDS model with aggregated private labels

9. APPENDIX 9: DYNAMIC MODEL, DIAGNOSTICS (FIXED EFFECTS SUR ESTIMATES)

Table 80: R-squares, p-values from the RESET and Fisher tests

based on fixed effects SUR models with disaggregated private label specification

MILK

R-squares: .99560448 .99102544 .99995467 .99719254

Ramsey's RESET test, p-value: 2.008e-28 Fisher's test of residual of equation CAMPINA (H0: unit root in all series in the panel), p-value: .03394832

Fisher's test of residual of equation FRIESCHE_VLAG (H0: unit root in all series in the panel), p-value: .05417647

Fisher's test of residual of equation PL_HIGH (H0: unit root in all series in the panel), p-value: .00169369

Fisher's test of residual of equation PL_LOW (H0: unit root in all series in the panel), p-value: .03647598

BUTTERMILK

R-squares: .98922184 .98093726

Ramsey's RESET test, p-value: .00120894

Fisher's test of residual of equation CAMPINA (H0: unit root in all series in the panel), p-value: .08685699

Fisher's test of residual of equation FRIESCHE_VLAG (H0: unit root in all series in the panel), p-value: .00363272

NATURAL YOGHURT

R-squares: .96728881 .98222495 .83828157 .9974097

Ramsey's RESET test, p-value: 2.190e-32

Fisher's test of residual of equation CAMPINA (H0: unit root in all series in the panel), p-value: .0255448

Fisher's test of residual of equation FRIESCHE_VLAG (H0: unit root in all series in the panel), p-value: .20216484

Fisher's test of residual of equation PL_HIGH (H0: unit root in all series in the panel), p-value: .00713966

Fisher's test of residual of equation PL_LOW (H0: unit root in all series in the panel), p-value: .05353031

VLA

R-squares: .87844283 .90818852 .99859795 .99109797

Ramsey's RESET test, p-value: 3.427e-15

Fisher's test of residual of equation CAMPINA (H0: unit root in all series in the panel), p-value: .05924405

Fisher's test of residual of equation FRIESCHE_VLAG (H0: unit root in all series in the panel), p-value: .06961404

Fisher's test of residual of equation PL_HIGH (H0: unit root in all series in the panel), p-value: .09246824

Fisher's test of residual of equation PL_LOW (H0: unit root in all series in the panel), p-value: .43806745

FRESH FLAVORED

R-squares: .8484987 .84539128 .88527873 .83905493

Ramsey's RESET test, p-value: 6.910e-64

Fisher's test of residual of equation CAMPINA (H0: unit root in all series in the panel), p-value: .00181192

Fisher's test of residual of equation MONA (H0: unit root in all series in the panel), p-value: .08288464

Fisher's test of residual of equation OPTIMEL (H0: unit root in all series in the panel), p-value: .00243543

Fisher's test of residual of equation FRIESCHE_VLAG (H0: unit root in all series in the panel), p-value: .43870887

27 November 2008

By e-mail

European Commission

DG Competition

Rue Joseph II 70 / Jozef II-straat 70

B-1000 Brussels

CASE M.5046 - Friesland Foods / Campina

COMMITMENTS TO THE EUROPEAN COMMISSION

Pursuant to Article 8(2)of Council Regulation (EC) No. 139-2004 (the "Merger Regulation"), Zuivelcoöperatie Friesland Foods U.A. and Koninklijke Friesland Foods N.V. ("Friesland Foods") and Zuivelcoöperatie Campina U.A. and Campina BV ("Campina") (hereafter collectively referred to as the "Parties"), hereby provide the following Commitments (the "Commitments") in order to enable the European Commission (the "Commission") to declare the merger between the Parties compatible with the common market and the EEA Agreement by its decision pursuant to Article 8(2) of the Merger Regulation (the "Decision"). The Commitments shall take effect upon the date of adoption of the Decision. This text shall be interpreted in the light of the Decision to the extent that the Commitments are attached as conditions and obligations, in the general framework of Community law, in particular in the light of the Merger Regulation, and by reference to the Commission Notice on remedies acceptable under Council Regulation (EC) No. 139-2004 and under Commission Regulation (EC) No. 802-2004.

SECTION A. DEFINITIONS

For the purpose of the Commitments, the following terms shall have the following meaning:

Affiliated Undertakings: undertakings controlled by the Parties, whereby the notion of control shall be interpreted pursuant to Article 3 of the Merger Regulation and in the light of the Commission Notice on the concept of concentration under Council Regulation (EC) No. 139-2004.

Campina: Zuivelcoöperatie Campina U.A. ("Coöperatie Campina "), a Dutch dairy cooperative counting 5,443 member farmers in the Netherlands, as per 31 December 2007 and Affiliated Undertakings. Coöperatie Campina is incorporated under the laws of the Netherlands and has its registered office at Hogeweg 9, Zaltbommel, and registered with the Commercial/Company Register at Midden-Nederland under number 11024221.

Closing: the transfer of the legal title of the Divestment Business to the Purchaser. Divestment Business(es): the business or businesses that the Parties commit to divest and as defined in the Schedules (which are submitted for each remedy separately as Annex A, B, C or D to the Commitments) submitted by the Parties.

Divestiture Trustee: one or more natural or legal person(s), independent from the Parties, who is approved by the Commission and appointed by the Parties and who has received from the Parties the exclusive Trustee Mandate to sell the Divestment Business to a Purchaser at no minimum price. Effective Date: the date of adoption of the Decision.

First Divestiture Period: The period of [CONFIDENTIAL]* from the Effective Date.

FrieslandCampina: the merged entity consisting of Campina and Friesland Foods.

Friesland Foods: Zuivelcoöperatie Friesland Foods U.A. ("Coöperatie Friesland Foods"), a Dutch dairy cooperative counting 9,417 member farmers as per 31 December 2007, and Affiliated Undertakings. Coöperatie Friesland Foods is incorporated under the laws of the Netherlands and has its registered office at Blankenstein 142, Meppel, and registered with the Commercial/Company Register at Noord-Nederland under number 04055219.

Hold Separate Manager: the persons appointed by the Parties for the Divestment Business to manage the day-to-day business under the supervision of the Monitoring Trustee.

Key Personnel: all personnel necessary to maintain the viability and competitiveness of the Divestment Business, listed in the Schedules (which are submitted for each remedy separately as Annex A, B, C or D to the Commitments).

Monitoring Trustee: one or more natural or legal person(s), independent from the Parties, who is approved by the Commission and appointed by the Parties, and who has the duty to monitor the Parties' compliance with the conditions and obligations attached to the Decision.

Personnel: all personnel currently employed by the Divestment Business, including Key Personnel, staff seconded to the Divestment Business, shared personnel and the additional personnel listed in the Schedules (which are submitted for each remedy separately as Annex A, B, C or D to the Commitments).

Purchaser: the entity approved by the Commission as acquirer of the Divestment Business in accordance with the criteria set out in Section D.

Trustee(s): the Monitoring Trustee and the Divestiture Trustee.

Trustee Divestiture Period: the period of [CONFIDENTIAL]* from the end of the First Divestiture Period.

SECTION B. THE DIVESTMENT BUSINESS

Commitment to divest

1. In order to restore effective competition, the Parties commit to divest the Divestment Businesses by the end of the Trustee Divestiture Period as a going concern to a purchaser and on terms of sale approved by the Commission in accordance with the procedure described in paragraph 15. To carry out the divestiture, the Parties commit to find a purchaser and to enter into a final binding sale and purchase agreement for the sale of the Divestment Businesses within the First Divestiture Period. If the Parties have not entered into such an agreement at the end of the First Divestiture Period, the Parties shall grant the Divestiture Trustee an exclusive mandate to sell the Divestment Businesses in accordance with the procedure described in paragraph 24 in the Trustee Divestiture Period.

2. The Parties shall be deemed to have complied with this commitment if, by the end of the Trustee Divestiture Period, the Parties have entered into final binding sale and purchase agreements, if the Commission approves the Purchasers and the terms in accordance with the procedure in paragraph 15 and if the closing of the sale of the Divestment Businesses takes place within a period not exceeding 3 months after the approval of the purchasers and the terms of sale by the Commission.

3. In order to maintain the structural effect of the Commitments, the Parties shall, for a period of 10 years after the Effective Date, not acquire direct or indirect influence over the whole or part of the Divestment Businesses, unless the Commission has previously found that the structure of the market has changed to such an extent that the absence of influence over the Divestment Businesses is no longer necessary to render the proposed concentration compatible with the common market.

Structure and definition of the Divestment Businesses

4. The Divestment Businesses concern 1) the entire fresh dairy business of Friesland Foods in the Netherlands ("Fresh Divestment Business"), 2) a significant part of the cheese business of Campina ("Cheese Divestment Business"), 3) two brands of Campina used in the market for long-life dairy drinks and 4) the availability of raw milk under the conditions as determined in the Schedule submitted as Annex D to the Commitments. The present legal and functional structure of the different Divestment Businesses as operated to date is described in the Schedules (which are submitted for each remedy separately as Annex A, B, C or D to the Commitments). The Divestment Businesses include

(a) all tangible and intangible assets (including intellectual property rights), which contribute to the current operation or are necessary to ensure the viability and competitiveness of the Divestment Businesses;

(b) all licences, permits and authorisations issued by any governmental organisation for the benefit of the Divestment Businesses;

(c) all contracts, leases, commitments and customer orders of the Divestment Businesses; all customer, credit and other records of the Divestment Businesses (items referred to under (a)-(c) hereinafter collectively referred to as "Assets");

(d) the Personnel;

(e) the benefit, for a transitional period after Closing and on terms and conditions equivalent to those at present afforded to the Divestment Businesses, of all current arrangements under which the Parties or Affiliated Undertakings supply products or services to the Divestment Businesses, as detailed in the Schedules (which are submitted for each remedy separately as Annex A, B, C or D to the Commitments). The purchaser of the Fresh Divestment Business will have right of first refusal to offer for the Cheese Divestment Business, at a price matching the best alternative offer if any by a third party and matching the purchaser requirements set out in paragraph 13.

SECTION C. RELATED COMMITMENTS

5. From the Effective Date until Closing, the Parties shall preserve the economic viability, marketability and competitiveness of the Divestment Businesses, in accordance with good business practice, and shall minimise as far as possible any risk of loss of competitive potential of the Divestment Businesses. In particular the Parties undertake:

(a) not to carry out any act upon its own authority that might have a significant adverse impact on the value, management or competitiveness of the Divestment Businesses or that might alter the nature and scope of activity, or the industrial or commercial strategy or the investment policy of the Divestment Businesses;

(b) to make available sufficient resources for the development of the Divestment Businesses, on the basis and continuation of the existing business plans;

(c) to take all reasonable steps, including appropriate incentive schemes (based on industry practice), to encourage all Key Personnel to remain with the Divestment Businesses.

Hold-separate obligations of Parties

6. The Parties commit, from the Effective Date until Closing, to keep the Divestment Businesses separate from the businesses it is retaining and to ensure that Key Personnel of the Divestment Businesses - including the Hold Separate Managers - have no involvement in any business retained and vice versa. The Parties shall also ensure that the Personnel does not report to any individual outside the Divestment Businesses.

7. Until Closing, the Parties shall assist the Monitoring Trustee in ensuring that the Divestment Businesses are managed as a distinct and saleable entity separate from the businesses retained by the Parties. The Parties shall appoint Hold Separate Managers who shall be responsible for the management of the Divestment Businesses, under the supervision of the Monitoring Trustee. The Hold Separate Managers shall manage the Divestment Businesses independently and in the best interest of the businesses with a view to ensuring its continued economic viability, marketability and competitiveness and its independence from the businesses retained by the Parties.

Ring-fencing

8. The Parties shall implement all necessary measures to ensure that it does not after the Effective Date obtain any business secrets, know-how, commercial information, or any other information of a confidential or proprietary nature relating to the Divestment Businesses. In particular, the participation of the Divestment Businesses in a central information technology network shall be severed to the extent possible, without comprising the viability of the Divestment Businesses. The Parties may obtain information relating to the Divestment Businesses which is reasonably necessary for the divestiture of the Divestment Businesses or whose disclosure to the Parties is required by law.

Non-solicitation clause

9. The Parties undertake, subject to customary limitations, not to solicit, and to procure that Affiliated Undertakings do not solicit, the Key Personnel transferred with the Divestment Businesses for a period of 2 years after Closing.

Due Diligence

10. In order to enable potential purchasers to carry out a reasonable due diligence of the Divestment Businesses, the Parties shall, subject to customary confidentiality assurances and dependent on the stage of the divestiture process:

(a) provide to potential purchasers sufficient information as regards the Divestment Businesses;

(b) provide potential purchasers sufficient information relating to the Personnel and allow them reasonable access to the Personnel.

Reporting

11. The Parties shall submit written reports in English on potential purchasers of the Divestment Businesses and developments in the negotiations with such potential purchasers to the Commission and the Monitoring Trustee no later than 10 days after the end of every month following the Effective Date (or otherwise at the Commission's request).

12. The Parties shall inform the Commission and the Monitoring Trustee on the preparation of the data room documentation and the due diligence procedure and shall submit a copy of an information memorandum to the Commission and the Monitoring Trustee before sending the memorandum out to potential purchasers.

SECTION D. THE PURCHASER

13. In order to ensure the immediate restoration of effective competition, the Purchaser, in order to be approved by the Commission, must:

(a) be independent of and unconnected to the Parties;

(b) have the financial resources, proven expertise and incentive to maintain and develop the Divestment Businesses as a viable and active competitive force in competition with the Parties and other competitors;

(c) neither be likely to create, in the light of the information available to the Commission, prima facie competition concerns nor give rise to a risk that the implementation of the Commitments will be delayed, and must, in particular, reasonably be expected to obtain all necessary approvals from the relevant regulatory authorities for the acquisition of the Divestment Businesses (the before-mentioned criteria for the purchaser hereafter the "Purchaser Requirements").

14. The final binding sale and purchase agreement(s) shall be conditional on the Commission's approval. When the Parties have reached an agreement with a purchaser, it shall submit a fully documented and reasoned proposal, including a copy of the final agreement(s), to the Commission and the Monitoring Trustee. The Parties must be able to demonstrate to the Commission that the purchaser meets the Purchaser Requirements and that the Divestment Businesses are being sold in a manner consistent with the Commitments. For the approval, the Commission shall verify that the purchaser fulfils the Purchaser Requirements and that the Divestment Businesses are being sold in a manner consistent with the Commitments. The Commission may approve the sale of the Divestment Businesses without one or more Assets or parts of the Personnel, if this does not affect the viability and competitiveness of the Divestment Businesses after the sale, taking account of the proposed purchaser.

SECTION E. TRUSTEE

I. Appointment Procedure

15. The Parties shall appoint a Monitoring Trustee to carry out the functions specified in the Commitments for a Monitoring Trustee. If the Parties have not entered into a binding sales and purchase agreement one month before the end of the First Divestiture Period or if the Commission has rejected a purchaser proposed by the Parties at that time or thereafter, the Parties shall appoint a Divestiture Trustee to carry out the functions specified in the Commitments for a Divestiture Trustee. The appointment of the Divestiture Trustee shall take effect upon the commencement of the Extended Divestment Period.

16. The Trustee shall be independent of the Parties, possess the necessary qualifications to carry out its mandate, for example as an investment bank or consultant or auditor, and shall neither have nor become exposed to a conflict of interest. The Trustee shall be remunerated by the Parties in a way that does not impede the independent and effective fulfilment of its mandate. In particular, where the remuneration package of a Divestiture Trustee includes a success premium linked to the final sale value of the Divestment Businesses, the fee shall also be linked to a divestiture within the Trustee Divestiture Period.

Proposal by the Parties

17. No later than one week after the Effective Date, the Parties shall submit a list of one or more persons whom the Parties propose to appoint as the Monitoring Trustee to the Commission for approval. No later than one month before the end of the First Divestiture Period, the Parties shall submit a list of one or more persons whom the Parties propose to appoint as Divestiture Trustee to the Commission for approval. The proposal shall contain sufficient information for the Commission to verify that the proposed Trustee fulfils the requirements set out in paragraph 17 and shall include:

(a) the full terms of the proposed mandate, which shall include all provisions necessary to enable the Trustee to fulfil its duties under these Commitments;

(b) the outline of a work plan which describes how the Trustee intends to carry out its assigned tasks;

(c) an indication whether the proposed Trustee is to act as both Monitoring Trustee an Divestiture Trustee or whether different trustees are proposed for the two functions.

Approval or rejection by the Commission

18. The Commission shall have the discretion to approve or reject the proposed Trustee(s) and to approve the proposed mandate subject to any modifications it deems necessary for the Trustee to fulfil its obligations. If only one name is approved, the Parties shall appoint or cause to be appointed, the individual or institution concerned as Trustee, in accordance with the mandate approved by the Commission. If more than one name is approved, the Parties shall be free to choose the Trustee to be appointed within one week of the Commission's approval, in accordance with the mandate approved by the Commission.

New proposal by the Parties

19. If all the proposed Trustees are rejected, the Parties shall submit the names of at least two more individuals or institutions within one week of being informed of the rejection, in accordance with the requirements and the procedure set out in paragraphs 16 and 19.

Trustee nominated by the Commission

20. If all further proposed Trustees are rejected by the Commission, the Commission shall nominate a Trustee, whom the Parties shall appoint, or cause to be appointed, in accordance with a trustee mandate approved by the Commission.

II. Functions of the Trustee

21. The Trustee shall assume its specified duties in order to ensure compliance with the Commitments. The Commission may, on its own initiative or at the request of the Trustee or the Parties, give any orders or instructions to the Trustee in order to ensure compliance with the conditions and obligations attached to the Decision.

Duties and obligations of the Monitoring Trustee

22. The Monitoring Trustee shall:

(i) propose in its first report to the Commission a detailed work plan describing how it intends to monitor compliance with the obligations and conditions attached to the Decision.

(ii) Oversee the on-going management of the Divestment Business with a view to ensuring its continued economic viability, marketability and competitiveness and monitor compliance by the Parties with the conditions and obligations attached to the Decision. To that end the Monitoring Trustee shall:

(a) monitor the preservation of the economic viability, marketability and competitiveness of the Divestment Businesses, and the keeping separate of the Divestment Businesses from the businesses retained by the Parties, in accordance with paragraphs 6 and 7 of the Commitments; (b) supervise the management of the Divestment Businesses as a distinct and saleable entity, in accordance with paragraph 8 of the Commitments;

(c) (i) in consultation with the Parties, determine all necessary measures to ensure that the Parties do not after the Effective Date obtain any business secrets, know-how, commercial information, or any other information of a confidential or proprietary nature relating to the Divestment Businesses, in particular strive for the severing of the Divestment Businesses' participation in a central information technology network to the extent possible, without compromising the viability of the Divestment Businesses, and (ii) decide whether such information may be disclosed to the Parties as the disclosure is reasonably necessary to allow the Parties to carry out the divestitures or as the disclosure is required by law;

(d) monitor the splitting of assets and the allocation of Personnel between the Divestment Businesses and the Parties or Affiliated Undertakings;

(iii) assume the other functions assigned to the Monitoring Trustee under the conditions and obligations attached to the Decision;

(iv) propose to the Parties such measures as the Monitoring Trustee considers necessary to ensure the Parties' compliance with the conditions and obligations attached to the Decision, in particular the maintenance of the full economic viability, marketability or competitiveness of the Divestment Businesses, the holding separate of the Divestment Businesses and the non-disclosure of competitively sensitive information;

(v) review and assess potential purchasers as well as the progress of the divestiture process and verify that, dependent on the stage of the divestiture process, (a) potential purchasers receive sufficient information relating to the Divestment Business and the Personnel in particular by reviewing, if available, the data room documentation, the information memorandum and the due diligence process, and (b) potential purchasers are granted reasonable access to the Personnel;

(vi) provide to the Commission, sending the Parties a non-confidential copy at the same time, a written report within 15 days after the end of every month. The report shall cover the operation and management of the Divestment Businesses so that the Commission can assess whether the business is held in a manner consistent with the Commitments and the progress of the divestiture process as well as potential purchasers. In addition to these reports, the Monitoring Trustee shall promptly report in writing to the Commission, sending the Parties a non-confidential copy at the same time, if it concludes on reasonable grounds that the Parties are failing to comply with these Commitments;

(vii) within one week after receipt of the documented proposal referred to in paragraph 15, submit to the Commission a reasoned opinion as to the suitability and independence of the proposed purchaser and the viability of the Divestment Businesses after the Sale and as to whether the Divestment Businesses are sold in a manner consistent with the conditions and obligations attached to the Decision, in particular, if relevant, whether the Sale of the Divestment Businesses without one or more Assets or not all of the Personnel affects the viability of the Divestment Businesses after the sale, taking account of the proposed purchaser.

Duties and obligations of the Divestiture Trustee

23. Within the Trustee Divestiture Period, the Divestiture Trustee shall sell at no minimum price the Divestment Business to a purchaser, provided that the Commission has approved both the purchaser and the final binding sale and purchase agreement in accordance with the procedure laid down in paragraph 15. The Divestiture Trustee shall include in the sale and purchase agreement such terms and conditions as it considers appropriate for an expedient sale in the Trustee Divestiture Period. In particular, the Divestiture Trustee may include in the sale and purchase agreement such customary representations and warranties and indemnities as are reasonably required to effect the sale. The Divestiture Trustee shall protect the legitimate financial interests of the Parties, subject to the Parties' unconditional obligation to divest at no minimum price in the Trustee Divestiture Period.

24. In the Trustee Divestiture Period (or otherwise at the Commission's request), the Divestiture Trustee shall provide the Commission with a comprehensive monthly report written in English on the progress of the divestiture process. Such reports shall be submitted within 15 days after the end of every month with a simultaneous copy to the Monitoring Trustee and a non-confidential copy to the Parties.

III. Duties and obligations of the Parties

25. The Parties shall provide and shall cause its advisors to provide the Trustee with all such cooperation, assistance and information as the Trustee may reasonably require to perform its tasks. The Trustee shall have full and complete access to any of the Parties' or the Divestment Businesses' books, records, documents, management or other personnel, facilities, sites and technical information necessary for fulfilling its duties under the Commitments and the Parties and the Divestment Businesses shall provide the Trustee upon request with copies of any document. The Parties and the Divestment Businesses shall make available to the Trustee one or more offices on their premises and shall be available for meetings in order to provide the Trustee with all information necessary for the performance of its tasks.

26. The Parties shall provide the Monitoring Trustee with all managerial and administrative support that it may reasonably request on behalf of the management of the Divestment Businesses. This shall include all administrative support functions relating to the Divestment Businesses which are currently carried out at headquarters level. The Parties shall provide and shall cause its advisors to provide the Monitoring Trustee, on request, with the information submitted to potential purchasers, in particular give the Monitoring Trustee access to the data room documentation and all other information granted to potential purchasers in the due diligence procedure. The Parties shall inform the Monitoring Trustee on possible purchasers, submit a list of potential purchasers, and keep the Monitoring Trustee informed of all developments in the divestiture process.

27. The Parties shall grant or procure the Affiliated Undertakings to grant comprehensive powers of attorney, duly executed, to the Divestiture Trustee to effect the sale, the Closing and all actions and declarations which the Divestiture Trustee considers necessary or appropriate to achieve the sale and the Closing, including the appointment of advisors to assist with the sale process. Upon request of the Divestiture Trustee, the Parties shall cause the documents required for effecting the sale and the Closing to be duly executed.

28. The Parties shall indemnify the Trustee and its employees and agents (each an "Indemnified Party") and hold each Indemnified Party harmless against, and hereby agrees that an Indemnified Party shall have no liability to the Parties for any liabilities arising out of the performance of the Trustee's duties under the Commitments, except to the extent that such liabilities result from the wilful default, recklessness, gross negligence or bad faith of the Trustee, its employees, agents or advisors.

29. At the expense of the Parties, the Trustee may appoint advisors (in particular for corporate finance or legal advice), subject to the Parties' approval (this approval not to be unreasonably withheld or delayed) if the Trustee considers the appointment of such advisors necessary or appropriate for the performance of its duties and obligations under the Mandate, provided that any fees and other expenses incurred by the Trustee are reasonable. Should the Parties refuse to approve the advisors proposed by the Trustee the Commission may approve the appointment of such advisors instead, after having heard the Parties. Only the Trustee shall be entitled to issue instruction to the advisors. Paragraph 29 shall apply mutatis mutandis. In the Trustee Divestiture Period, the Divestiture Trustee may use advisors who served the Parties during the Divestiture Period if the Divestiture Trustee considers this in the best interest of an expedient sale.

IV. Replacement, discharge and reappointment of the Trustee

30. If the Trustee ceases to perform its functions under the Commitments or for any other good cause, including the exposure of the Trustee to a conflict of interest:

(a) the Commission may, after hearing the Trustee, require the Parties to replace the Trustee; or (b) the Parties, with the prior approval of the Commission, may replace the Trustee.

31. If the Trustee is removed according to paragraph 31, the Trustee may be required to continue in its function until a new Trustee is in place to whom the Trustee has effected a full hand over of all relevant information. The new Trustee shall be appointed in accordance with the procedure referred to in paragraphs 16-21.

32. Besides the removal according to paragraph 31, the Trustee shall cease to act as Trustee only after the Commission has discharged it from its duties after all the Commitments with which the Trustee has been entrusted have been implemented. However, the Commission may at any time require the reappointment of the Monitoring Trustee if it subsequently appears that the relevant remedies might not have been fully and properly implemented.

SECTION F. THE REVIEW CLAUSE

33. The Commission may, where appropriate, in response to a request from the Parties showing good cause and accompanied by a report from the Monitoring Trustee:

(i) Grant an extension of the time periods foreseen in the Commitments, or

(ii) Waive, modify or substitute, in exceptional circumstances, one or more of the undertakings in these Commitments.

Where the Parties seek an extension of a time period, it shall submit a request to the Commission no later than one month before the expiry of that period, showing good cause. Only in exceptional circumstances shall the Parties be entitled to request an extension within the last month of any period.

duly authorised for and on behalf of Friesland Foods

duly authorised for and on behalf of Campina

Annex A: Schedule Fresh

Scope of the Divestment Business

(1) The Divestment Business consists of all fresh dairy activities of Friesland Foods ("FF Fresh") in the Netherlands. FF Fresh is a stand alone operating unit of Friesland Foods and is responsible for the production, sales and marketing in the Netherlands of non-organic (i) fresh milk, (ii) fresh buttermilk, (iii)fresh custard (vla), (iv) fresh plain yoghurt, (v) fresh dairy desserts, (vi) fresh flavoured dairy drinks, (vii) value added yoghurts and quark and (viii) fresh liquid cream which are sold under the Friesche Vlag brandand under private label. In addition, FF Fresh sells and markets all (i) organic fresh milk, (ii) organic fresh basic yoghurt, (iii) organic fresh custard (vla) and (iv) organic fresh buttermilk which is sold under the Friesche Vlag brand (together referred to as the "FF Fresh product portfolio"). The Friesche Vlag organic products are produced by Friese Ekologische Zuivel BV ("FEZ") in Drachten which is not part of the Divestment Business.

(2) FF Fresh currently employs approximately 513 FTEs. The majority is permanent staff (363 FTEs), most of whom are in operations (318 FTEs), with the remainder in commerce and functional support areas. All commercial and functional support staff is situated next to the production facility and distribution centre in Nijkerk.

(3) Please find below an organizational chart showing all departments which report to the managing director.

<table>

(4) The temporary staff of approximately 150 FTEs is predominantly employed as production personnel.

(5) The management team of FF Fresh consists of six individuals: the managing director, director operations, manager F&A, director marketing, director sales, and manager HR. The Customer Service Management department ("CSM") and Purchasing department are not represented in the management team, however these departments report directly to the managing director. Operations

(6) Out of the 318 operational FTEs, roughly one half is equally distributed across the mainstream and specialty facilities, about 25 percent is deployed in the distribution center, and the remainder comprise operational support activities (i.e. maintenance, quality control, etc.).

(7) The operational support groups comprise a number of different activities:

(a) Maintenance, the largest group consisting of 28.4 FTEs, is responsible for all maintenance that is not covered by maintenance contracts with suppliers

(b) Quality Assurance / Quality Control (14.8 FTEs)

(c) In Bound Logistics (15.7 FTEs)

(d) Supply Chain Planning (10.6 FTEs)

(e) Products and Process Development (9.3 FTEs). This group is responsible for investigating small improvements or renewals within the current FF Fresh portfolio, but does not include any "research". Research is organized centrally in the Friesland Foods group. Finance & Accounting

(8) The Finance & Accounting ("F&A") group consists of one manager and staff of 17.7 FTEs. The FF Fresh F&A group can be further divided in a commerce and operations group, both equal in size. The operations group is responsible for controlling the operational activities of FF Fresh, whereas the commerce group is responsible for controlling the commercial activities, including debtors' management. The sub-group "gegevensbeheer" is supporting the invoice process. Marketing

(9) FF Fresh Marketing totals 8.8 FTEs. The responsibilities of the marketing group primarily consist of Brand Management and Trade Marketing. Activities of Brand Management focus on analysing and anticipating consumer trends, development and introduction of new product concepts, brand activation based on Above The Line and Below The Line promotions and portfoliomanagement. Activities of Trade Marketing focus onshopper analysis, category analysis and advice, space management and in-store support.

Sales

(10) The sales group of FF Fresh totals 17.7 FTEs. The group is split into a retail sales team and OOH sales team (about 7 FTEs). Activities focus on account management. On the retail side this includes key account management for several retailers such as Albert Heijn, C1000, Jumbo, Plus and Hoogvliet, as well as account management at store level (winkel account management, 5 FTE). On the OOH side, account management is organized according to different types of OOH customers (e.g. catering, gas stations, etc.).

Human Resources

(11) The HR department employs 3.9 FTEs (1 manager human resource, 1.9 HR manager and 1 manager education). The FF Fresh HR department is responsible for HR activities such as support of local management, recruiting and educating local production personnel and for the registration of hours (3.0 FTE). A part of the administrative support has been categorised in the Shared Service Centre of Friesland Foods.

ICT

(12) The FF Fresh ICT department cooperates with the shared Corporate ICT centre. The FF Fresh ICT infrastructure (hard- and software) is an integrated part of the total Corporate ICT infrastructure of Friesland Foods. FF Fresh currently deploys an ERP system from CSB-System (headquartered in Germany). Theentire supply chain of FF Fresh is integrated into this platform.

(13) As explained above, CSM and Purchasing are not represented in the management team but do report to the managing director. Therefore, a description of their activities is included below.

Customer Service Management

(14) The CSM department of FF Fresh employs 19.7 FTEs. Main tasks of the CSM department are receiving and processing orders, managing imports, in-bound (returning) packaging logistics, invoicing, transport management / optimization. Logistic Account Management as well as the replenishment for a key retail account.

Purchasing

(15) Within the Friesland Foods organisation, purchasing of product and non-product related materials and services is considered as an important strategic lever. Therefore purchasing is centrally coordinated at group level. Strategic purchasing teams exist for all key product and services categories. These teams consist of purchasing personnel throughout the various operating companies who combine their purchasing responsibility for the operating company with the role of team member in the group-wide Purchasing category teams. The purchasing team in FF Fresh is therefore small (2 FTEs) and reports directly to the MD. The key responsibility is to ensure proper supply of ingredients, packaging material, services and the like. Raw milk purchasing is done by the planning department of Friesland Foods.

Legal structure

(16) FF Fresh is a separate operating entity of Friesland Foods, but not a separate legal entity. FF Fresh is an operating entity residing under Friesland Foods BV All shares of Friesland Foods BV are held by Koninklijke Friesland Foods N.V. On its turn Koninklijke Friesland Foods N.V. is wholly owned by Zuivelcoöperatie Friesland Foods U.A. It should be noted that Friese Ekologische Zuivel BV ("FEZ") from

Drachten is part of the operating company FF Fresh. However, FEZ is a 100% subsidiary of Friesland Foods BV and has a separate production facility which is not part of the Divestment Business. A legal structure chart is submitted below.

<table>

(17) More in particular, the Divestment Business consists of the following tangible and intangible assets:

Tangible assets

(18) The ownership of the modern production facility and the highly automated distribution centre of FF Fresh at Nijkerk, as well as the dedicated (commercial and administrative) organisation of FF Fresh; The Nijkerk site is strategically located in an area in which significant volumes of raw milk are produced and which is in the proximity of distribution centres of all major Dutch retailers and Out-of-Home wholesalers. The Divestment Business will include the whole site and the distribution centre of FF Fresh, including the ownership of the real estate on which the buildings are located as well as the buildings themselves. The Nijkerk plant is a state-of-the-art plant, one of the most modern and efficient plants in Western Europe, and consists of a mainstream and a specialty products facility. Overall the plant produces [...]* litres of fresh dairy products per year and has a capacity of [...]* litres of fresh dairy products per year.

The Nijkerk production facility has a total of 18 packaging lines (1), which produce 190 store keeping units ("SKUs"), divided into 146 branded and 44 PL products. Of the 18 packaging lines, FF Fresh owns 11 and leases 7 lines (2). 8 packaging lines are dedicated to mainstream products and 10 lines are dedicated to specialty products.

Distribution centre

(19) The Divestment Business will also include a highly automated distribution center that processes the entire volume sold (i.e. own production and sourced products). The distribution centre is located in Nijkerk, adjacent to the production facility. The Divestment Business will include the ownership of the whole distribution centre.

Offices

(20) Next to the production site in Nijkerk (Gildenstraat), FF Fresh has an office where overhead activities are located. Approximately 300 meters beyond the production site in Nijkerk, FF Fresh rents an office where Marketing, Sales and F&A employees employ their activities (Ambachtsstraat). Furthermore, FF Cheese Benelux rents an office of FF Fresh, located at Nijverheidsstraat 45 in Nijkerk. This lease agreement will be terminated on 1 September 2009 at the latest.

Intangible assets

(21) Friesland Foods commits to transfer the following intangible assets: Brands

(22) The Divestment Business will include an exclusive, five year, royalty-free license to use the Friesche Vlag brand name in the Netherlands for (i) fresh milk, (ii) fresh buttermilk, (iii) fresh plain yoghurt, (iv) fresh custard, (v) porridge, (vi) fresh flavoured dairy drinks, (vii) value added yoghurts & quark, (viii) liquid fresh cream, (ix) organic fresh milk, (x) organic fresh plain yoghurt, (xi) organic fresh custard (vla) and (xii)organic fresh buttermilk. The five-year royalty-free license may, at the option of the buyer of the Divestment Business, be prolonged for a period of maximum five years. The license period will be followed by an indefinite black out period for the use of the Friesche Vlag brand for these products in the Netherlands. This means that after the license period FrieslandCampina will never be allowed to use the Friesche Vlag brand for these fresh dairy products in the Netherlands.

(23) Furthermore, in addition to the proposed licence for the Friesche Vlag brand the remedy proposal also includes the transfer of Campina's Melkunie brand at the option of the Purchaser. The Melkunie brand is a well-known brand in the Netherlands which was phased-out from the market by Campina in 2002. The Melkunie brand was at that time one of the strongest brands in the Netherlands. In 2005 - 2 years after it was phased out - the brand was still better known than for example the Danone brand. Despite its strength the Melkunie brand was phased out by Campina due to its corporate strategy. According to this strategy Campina aims at promoting all of its fresh products, in the Netherlands and abroad, under the same brand. The Campina brand was considered to be potentially a more international brand than Melkunie. Nevertheless, due to the strong position of the Melkunie brand in the Netherlands, the Melkunie logo was placed on the packaging for a long period after the Melkunie brand had been phased out.

(24) The Parties submit that the Melkunie brand is still well-known today. This is among others evidenced by a recent legal dispute between Campina and Cono in which an injunction was granted against Cono in order for Campina to protect its Melkunie brand (see District Court 's-Hertogenbosch, 11 May 2007, Case 158946, KG ZA 07-313). The cause for the initiation of the injunction by Campina was an attempt by Cono to link its Beemster cheese with the Melkunie brand. The Parties submit that the attempt of Cono to link its own successful range of Beemster cheese with Melkunie and the fact that Campina still protects the Melkunie brand indicates that the Melkunie brand is still considered relevant today. In addition, the Parties submit that a recent market survey showed that the so called "top-of-mind" awareness with respect to the Melkunie brand was on average 4% until week 41 of 2008 and the so called "spontaneous" brand awareness of theMelkunie brand was on average 15% over the same period. The "top-of-mind" awareness measures thepercentage of respondents in the Netherlands that identified Melkunie as the first brand that comes to their mind when asked which brands they are familiar with in fresh dairy. The "spontaneous" awareness measures the percentage of respondents in the Netherlands that named the brand Melkunie (but not necessarily as the first brand) when asked which brands they are familiar with in fresh dairy. The survey showed that six years after the withdrawal of the Melkunie brand from the Dutch market it is still identified by 4% of respondents in the Netherlands as the first brand that comes to their mind in fresh dairy and 15% of respondents identifies the brand as one of the known brands in fresh dairy. In the view of the Parties these percentages show that the Melkunie brand should still be considered a strong brand with a high potential for further growth.

(25) In view of the above the Parties submit that the inclusion of the Melkunie brand also ensures that the Purchaser of the Divestment Business will be in the position to re-brand the Friesche Vlag products within the five-year licence period (and the indefinite black-out period) to a brand which is already well known in the Netherlands.

(26) Furthermore, the Divestment Business will include a perpetual royalty-free license for the use of the Weidemelk logo, under the condition that the logo is used for products that fulfil the criteria of the Weidemelk foundation. The Weidemelk concept, as defined by the Weidemelk foundation, implies that the cows producing the raw milk have spent a minimum of 120 days (for 6 hours a day) of grazing time in the open air in the period between the spring and autumn;

(27) Finally, the Divestment Business will include the full ownership of all Friesche Vlag sub-brand names and all brands that are specific to the products of FF Fresh (with the exception of the Friesche Vlag brand itself).The purchaser of the Divestment Business will be able to continue using all Friesche Vlag sub-brands even after having re-branded the FF Fresh product portfolio from Friesche Vlag to, for instance, Melkunie. The use of all sub-brands provides the purchaser additional tools to ensure an undisrupted re-branding process, by focussing its promotion activities on the sub-brands instead of the Friesche Vlag brand;

Recipes

(28) The Divestment Business will include the transfer of ownership of recipes, production and process know how and related patents for all products currently consisting the FF Fresh product portfolio.

Licenses, permits and authorizations

(29) Friesland Foods commits to transfer to the purchaser of the Divestment Business all licenses, permits and authorisations which have been issued for the operation of the Divestment Business, to the extend permitted by the competent authority.

Supply agreements

Raw milk supply

(30) In order to ensure the undisrupted continuation of the Divestment Business, raw milk has to be supplied on a continuous basis on the basis of a Supply Agreement (see Annex 1). The Parties submit that the raw milk supplies are necessary to maintain the full economic viability and competitiveness of the Nijkerk production facility during a transitional period of two years on the basis of a supply agreement. In order to facilitate the migration of the raw milk supply to other sources, like the Raw Milk Remedy, the merged entity commits to supply the raw milk required by the Divestment Business to be able to utilize the current maximum capacity of the Nijkerk facility for a period of two years. The current maximum capacity of raw milk processed at the Nijkerk production facility is [...]* kg. The raw milk supplied to the Nijkerk production facility will meet all quality requirements which will apply to FrieslandCampina. Currently, Friesland Foods supplies FF Fresh with raw milk that meets the criteria set by the Weidemelk foundation. At the request of the purchaser of the Divestment Business FrieslandCampina will continue to supply raw milk which meets the Weidemelk criteria.

(31) Under the Raw Milk Remedy, the purchaser of the Divestment Business may continue to procure raw milk from the Dutch Milk Fund acting as agent for FrieslandCampina under the same conditions as under the supply agreement with FrieslandCampina.

(32) The supply agreement stipulates that, should the Dutch Milk Fund not become operational within two years, at the option of the buyer of the Divestment Business the duration of two year supply agreement may be extended until 1 January 2017 at the latest.

Price raw milk

(33) The price for the raw milk under the Supply Agreement will be the monthly "guaranteed milk price" of FC being applicable in the month of delivery and including any costs as specified in the Supply Agreement. The Guaranteed Price will be subject to a rebate of 1% for the period until six months after the Dutch Milk Foundation has become operational within the meaning of clause 13.2. The guaranteed milk price is based on standard fat (4,41%) & protein (3,47%) contents and will be objectively adjusted for the actual fat and protein contents of the raw milk supplied (on the same basis as is being done for the members of FrieslandCampina). The guaranteed milk price is the cash paid out price exclusive of dividend, interest for debt, loans and other returns on member capital of FrieslandCampina, which is paid by FrieslandCampina to its members for the supply of raw milk.

(34) The guaranteed milk price is a representative price for raw milk in North West Europe and reflects the weighted average milk price over the year paid out by dairy companies in Denmark, Germany, Belgium and the Netherlands ("the Dairy Companies"), altogether constituting the average price paid for about 45 billion kg of raw milk. The milk prices of the Dairy Companies are based on verifiable sources like ZMP in Germany and LTO Nederland (Land- en Tuinbouw Organisatie Nederland) in the Netherlands. The development is determined by the monthly price developments in Germany, Belgium, Denmark and the Netherlands.

(35) The table below demonstrates how the guaranteed milk price using the weighted average milk price over the year paid out by the Dairy Companies is determined. In the last column of the table the weight attributed to each of the Dairy Companies involved is mentioned (i.e. CONO, Bel Leerdammer, DOC each account for7.9%, Arla accounts for 10%, Milcobel accounts for 6.7% and the dairy companies reflected in the ZMP data account for 59.6%).

<table>

(36) According to the statutes of FrieslandCampina, the guaranteed price will be reviewed once every three years. However, such review is only meant to ensure that the guaranteed price is still a representative market price.

(37) Under the Supply Agreement, the Divestment Business will pay the guaranteed milk price to FrieslandCampina, minus a 1% rebate as set out in paragraph,33, plus a fee for logistical and other costs. Logistics and transport (collection at farms and delivery at processing plants) as well as quality control,quality assurance and administration are costs borne by FrieslandCampina. These costs are not charged separately to the farmers. Consequently, the farmers receive the guaranteed price for the raw milk delivered without a discount for these costs. FrieslandCampina carries these costs as overhead. Therefore, thesecosts are included in the profit and loss account of FrieslandCampina and hence affect the profits of FrieslandCampina.

(38) In this situation, it goes without saying that the logistical and other costs relating to the collection of raw milk made available to the Divestment Business should be borne by Divestment Business. Otherwise, FrieslandCampina would have to carry these costs twice: on behalf of itself (in relation to its members) and in relation to the raw milk supplied to Divestment Business.

(39) All costs that arise from the logistics and transport (collection at farms and delivery at Nijkerk) of raw milk sold to the Divestment Business as well as quality control, quality assurance and administration will be calculated in advance by FrieslandCampina and will be charged on an objective, transparent and verifiable basis. The costs charged by FrieslandCampina to the Divestment Business per 100 kg Raw Milk are the same as for FrieslandCampina internally. The costs are based on optimal logistic solutions, maximumeconomies of scale and the average actual costs of FrieslandCampina per 100 kg Raw Milk (which is currently EUR [...]*). In other words, the Divestment Business will benefit from the economies of scale that are achieved by FrieslandCampina. The (breakdown of the) costs concerned will be made transparent for the Divestment Business.

(40) The mentioned EUR [...]* is based on the actual costs FrieslandCampina has to make ([CONFIDENTIAL]*).

(41) At the request of the purchaser of the Divestment Business FrieslandCampina will supply raw milk which meets the Weidemelk criteria. For supplies of raw milk that meets the Weidemelk criteria a surcharge will be paid which is equal to the surcharge which is currently charged internally by Friesland Foods. Currently, this surcharge amounts to EUR 0.01 per kg of raw milk that meets the Weidemelk criteria.

(42) In the unforeseen case of a dispute the Supply Agreement prescribes for an efficient and speedy arbitration.

Planning of the supplies of raw milk

(43) Under the two-year Supply Agreement, FrieslandCampina will supply all raw milk ordered by the Divestment Business with a maximum of the volume needed if the Nijkerk production facility operates at full capacity. The volume needed from year to year will be communicated by the purchaser to FrieslandCampina six months before the beginning of the calendar year concerned. The annual forecast may be modified afterwards but shall be made definitive at the latest by 1 April of the calendar year concerned, as set forth in further detail in the Supply Agreement.

(44) The Nijkerk production facility shall provide FrieslandCampina before the beginning of every month with a 13 month rolling forecast for the volume of Product required per month based on the Annual Forecast. Moreover, the Nijkerk production facility shall provide the Supplier before the beginning of every week with a 13 week rolling forecast for the volume of Product required per week.

(45) The weekly planning provides for deliveries 7 days a week, unless explicitly agreed otherwise in writing. The Supplier shall accommodate normal fluctuation in ordered volume of Product during the week.

(46) The Annual Forecast of Product as provided by the Divestment Business in accordance with the Supply Agreement, will be supplied by FrieslandCampina to the Divestment Business in equal weekly volumes (the Average Weekly Volume), provided that orders may be adjusted by a maximum of +20% or -20% of the Average Weekly Volume in aggregate in any week. FrieslandCampina and the Divestment Business will agree that the volume supplied over a period of 13 weeks will correspond to the average volumes for a 13 week period (i.e. 13/52 x the Annual Forecast (the Average 13 Week Volume), but may be adjusted by a maximum of +5% or -5% of the Average 13 Week Volume. Finally, Fresh may deviate from the Annual Forecast over the entire calendar year concerned by a maximum of +5% or -5%.

(47) Experience shows that the flexibility percentages (yearly, monthly and weekly) are largely sufficient to accommodate the Divestment Business also in the light of commercial cycles.

(48) All other terms and conditions are described in detail in the Raw Milk supply agreement (Annex 1).

Supply of other milk-based ingredients

(49) The Divestment will also include an eight-year supply agreement for all skimmed milk, sweet buttermilk, goat milk and whey-paste required to produce the maximum volumes of end-products comprising the current product assortment of FF Fresh. The price of skimmed milk, sweet buttermilk, goat milk and whey paste will be determined in consultation with the purchaser of the Divestment Business; Supply of residual cream by the Divestment Business to FrieslandCampina

(50) The Divestment Business will at the option of the purchaser also include separate arrangements for the sale of the residual cream of the Divestment Business production facility to FrieslandCampina for a period of eight years in proportion to the same quantities as the merged entity supplies raw milk to the Nijkerkproduction facility, either under the two-year supply agreement or the Raw Milk Remedy. It should be stressed that the supply of rest streams to FrieslandCampina is at the option of the Purchaser of the Divestment Business. Hence, the Purchaser of the Divestment Business is free to sell the cream and wheyrest streams on the market or use it for its own further processing. If the Purchaser of the Divestment Business wishes to sell the rest streams to FrieslandCampina, the following price calculation applies:

Price paid for rest streams

(51) The price of the residual cream (in proportion to the same quantities as the merged entity supplies raw milk to the Divestment Business production facility under the Supply Agreement or the Preferential Drawing Rights) will be determined on the basis of the fat-price as a component of the guaranteed price for raw milk (the garantieprijs), minus any costs for additional transport, handling and taxes which Friesland Campina will incur.

(52) The residual cream produced by the Divestment Business can be transported over a distance of more than 300 km. Hence, the Divestment Business will have ample opportunities to supply its residual cream to alternative outlets.

(53) All terms and conditions are described in detail in the supply agreement for Cream (Annex 2). Supply of organic basic dairy products

(54) The Divestment Business will include a one year agreement for the supply by FrieslandCampina of organic (i) milk, (ii) plain yoghurt, (iii) custard (vla) and (iv) buttermilk. The Parties are convinced that a one year supply agreement provides sufficient time to the purchaser of the Divestment Business to find alternative sources for the replacement of the organic product portfolio of FF Fresh. In this context it is noted that the purchaser of the Divestment Business will be entitled to market the sourced organic products under the Friesche Vlag brand regardless the origin of these products;

(55) All terms and conditions are described in detail in the supply agreement for organic products (Annex 3);

Supply of other product and non-product related goods and services

(56) The Divestment Business will include, at the option of the purchaser, Temporary Service Level Agreements for ICT, Training & Development, Payroll Services, Employee Health Insurances and Pension, Job evaluation, Food Safety, Consumer Services, Marketing and Research & Development;

(57) Moreover, the Divestment Business will include a one-year supply agreement by FrieslandCampina for all other product and non-product related goods and services which Friesland Foods currently sources centrally, to the extend permitted under those agreements;

(58) Finally, the Divestment Business will also include the transfer of the current sourcing agreement for quark, to the extend transferable;

Financial data

(59) The Divestment Business achieved a turnover of EUR [...]* in 2007 and EBIT of EUR [...]*. Based on the latest estimates, the Divestment Business is expected to achieve a turnover of EUR [...]* in 2008 and an EBIT of EUR [...]*.

Annex B: Schedule Cheese

Scope of the Divestment Business

(60) Parties offer to divest a very attractive cheese business, consisting of 1) one of Campina's most cost efficient cheese production facilities, namely in Bleskensgraaf (the Netherlands), specialised in the production of Dutch type nature cheese and 2) an experienced sales team carved out from the sales organisation of the contemplated merged entity (FrieslandCampina) and other employees for R&D, planning and logistics and general support (the "Divestment Business"). The Bleskensgraaf cheese production facility is a stand alone plant, with modern and flexible production lines, storage facilities and office space. The site is located in the Netherlands in an area in which significant volumes of raw milk are produced. It is responsible for theproduction of Dutch type nature cheese (currently dedicated to Gouda cheese, but also suitable for the production of Edam and Maasdam cheese).

(61) The Divestment Business includes the whole of the Bleskensgraaf production facility including the land and premises which are wholly owned by Campina. The Bleskensgraaf production facility produces Dutch type nature cheese. Currently, the Bleskensgraaf production facility produces both wheel shaped (12 kg) and rectangular shaped (Euroblock shaped cheese has a standardised shape of 30x50x10 cm) Dutch type nature cheese. The Bleskensgraaf facility does not produce rindless Dutch type cheese.

(62) In 2007, the Bleskensgraaf production facility realised a production of approximately [...]* Dutch type cheese (i.e. producing both wheel and rectangular shaped cheese). At the moment the Bleskensgraaf production facility produces only Gouda cheese, but the plant is also suitable for the production of other Dutch type cheeses. In addition, with limited investments the Divestment Business production facility can also be made suitable for the production of rindless Dutch type cheese which will give additional flexibility on top of the current round wheel/Euroblock nature capacity. Taking into account certain conditions in the current configuration the Bleskensgraaf production facility has the capability of achieving a higher output (i.e. approximately [...]*) and can with limited investments further extend its capacity with [...]* additional KT. With additional investments the Bleskensgraaf plant should within a period of three years be able to further expand to an output of [...]*. The wastewater treatment facility has recently been upgraded to facilitate increased output; with some additional measures to reduce water usage in the factory the treatment facility and the corresponding wastewater disposal license can cope with the output of approx. [...]*. It is noted that the relevant permits and licences will allow for each of the above mentioned expansion options. In addition, is can be noted that in the industrial area where the Bleskensgraaf production facility is located there is space to further expand the cheese production (i.e. creating additional capacity by building an additional production plant).

(63) The Bleskensgraaf production facility has no ripening and packaging facilities on site, but is using outsourced (internal and external) services for these activities. The basic assumption is that the Divestment Business will sell the current production volume predominantly as ex-factory cheese (i.e. 15-day old cheese). However, the Divestment Business will also be capable to sell ripened and/or pre packed cheese through the outsourcing of ripening and co packing. A number of third parties offering ripening facilities and/or co packing are available to provide the necessary services to the Divestment Business at its request. Using outsourced services adds to the efficiency and flexibility of the Divestment Business. As noted above in the industrial area where the Bleskensgraaf production facility is located there is space to expand. Therefore, the Divestment Business could also choose to create its own ripening facility in the vicinity of the Bleskensgraaf production facility.

(64) However, in order to further improve the Divestment Business' viability, the Parties offer the Divestment Business, at the option of the purchaser and for a transitional period of one year, to ripen cheese produced at the Divestment Business production facility at third parties under its current agreements and/or pre pack the cheese in their own facilities on a cost plus basis.

(65) The Parties will include in the Divestment Business the recipes for the production of the cheese currently produced by the Divestment Business and -if needed- also recipes for the production of other Dutch type cheese (Maasdam and Edam).

Commercial attractiveness of Divestment Business

(66) The Bleskensgraaf production facility is one of the most profitable cheese production facilities of the contemplated merged entity (FrieslandCampina). The Bleskensgraaf cheese production facility was established in 1990. In the course of the years a number of expansions and upgrades took place, creating one of Campina's most cost efficient cheese production facilities. The most recent investments are: (i) an upgrade to meet the growing market demand for Euroblock shaped cheese in 2006 (i.e. an investment in a new production line of approx. EUR [...]*), (ii) the automation of the factory cheese ripening facility for 15 day old cheese in 2008 (i.e. an investment in automation systems of approx. EUR [...]*), and (iii) an upgrade of the wastewater treatment plant in 2008 (an investment of approx. [...]*).

(67) The equipment of the Bleskensgraaf production facility is up to standards and maintenance is at an up to date level. There are no major replacement investments needed for the next five years. The competitiveness of the Bleskensgraaf production facility can be demonstrated by the table below. This confidential table shows the conversion costs for the production of 15 days old cheese of the different plants in EUR/100kg. The conversion costs exist of: labour, energy, maintenance, factory overhead, environmental costs and depreciation. From this table it is evident that the cost structure of Bleskensgraaf plant is indeed very competitive in relation to other cheese factories.

Table 1: The Bleskensgraaf production facility

Personnel [...]*

Energy [...]*

Maintenance [...]*

Depreciation [...]*

Other* [...]*

Total [...]*

* costs for general laboratory tests, third party services, and taxes etc.

(68) The Bleskensgraaf production facility has a well developed certified quality system (based on ISO 9000, ISO 14000, process certification Qlip, IFS, GMP Feed). The Bleskensgraaf production facility is with regard to quality up to standard compared to other cheese production facilities (e.g. the Bedum, Lutjewinkel, Steenderen and Workum production facilities). This is illustrated in the confidential table 2 below with the key performance indicators (kpi) loss (%) and costs of complaints (EUR/100 kg). The definitions of these kpi's are: loss (%) is the percentage of the produced volume that deviates from the standards and which is sold with a discount. Cost of complaints (EUR/ 100 kg) are the cost in EUR that are incurred as a result of complaints from customers regarding cheese products that do not meet the standards/demands of the customers and for which the producer has to pay a compensation to the customer.

Table 2: The Bleskensgraaf production facility

Loss [%]* [...]*

Costs of complaints [EUR/100kg]* [...]*

(69) In 2007, the Bleskensgraaf production facility generated a pro forma turnover of approximately EUR [...]*. For 2008, the Bleskensgraaf production facility is expected to generate a turnover of approximately EUR [...]*. The aforementioned turnovers for 2007 and 2008 are calculated on a pro forma basis due to the fact that they are realised by the Bleskensgraaf production facility fulfilling its current role as one of the production facilities of the Campina cheese production organisation. It is noted that producing another product mix/supplying other customers could lead to different financial results for the Bleskensgraafproduction facility.

(70) The Divestment Business will include the whole site, including the ownership of the real estate on which the buildings are located as well as the buildings, personnel and all necessary licences, permits and authorisations to operate the plant. As noted above, the Bleskensgraaf production facility produces Dutch type nature cheese for the Campina cheese production organisation which uses the Dutch type naturecheese originating from Bleskensgraaf to sell this to various customers of the Campina cheese productionorganisation. The Bleskensgraaf production facility therefore does not have its own customers. Consequently, the Divestment Business will not include customer contracts and customer records and thePurchaser will be completely free to decide to who the cheese produced by the Divestment Business will be sold.

Personnel Divestment Business

(71) The Bleskensgraaf production facility has 56 FTE (Fixed: 46 FTEs, Temps: 10 FTEs). Currently, the Bleskensgraaf facility has a production administration department on site. They will remain with Divestment Business and fall within the 56 FTE.

(72) In addition to these 56 FTE's, the Divestment Business will include an experienced sales team carved out from sales organisation of FrieslandCampina (i.e. a senior sales manager and supporting staff (3 FTE)). The sales team that will be allocated to the Divestment Business has experience in both retail and wholesale sales. (i.e. the senior sales manager has sufficient experience to make sales to value adding wholesalers and retailers such as e.g. Albert Heijn).

(73) Currently, the planning of the cheese production in the Bleskensgraaf production facility is organised by the central office of Campina Holland Cheese in Tilburg (the Netherlands). The allocation of the supplies of raw milk that is required for the production of the Dutch type cheese and the planning of the by products cream and whey are done by central office of Campina in Zaltbommel (the Netherlands). The sales of the cheese production in the Bleskensgraaf production facility is organised by the central sales office of Campina Holland Cheese in Tilburg (the Netherlands). The factory also receives QA and technological support and some general support from Campina Holland Cheese in Tilburg (The Netherlands).

(74) The Divestment Business will be allocated appropriate planning, administration, quality assurance and sales employees to ensure that it will be a viable stand alone operation. In addition, the Divestment Business will also include a Sales & Operations planner and supporting staff (2 FTE), support related to Development and Quality Assurance (2 FTE), support on administration (a senior controller and supporting staff (3 FTE)), as well as support the stand alone operation in terms of general support (e.g. HR (1.5 FTE)).

Figure 1: Future organisation diagram of Bleskensgraaf cheese production facility

(75) If the Purchaser of the Divestment Business believes that the sales force offered in the Divestment Business is too large, FrieslandCampina will offer to reduce the number of FTEs for the sales team.

Legal structure

(76) The Bleskensgraaf production facility is owned by Campina. The Bleskensgraaf production facility is not a separate legal entity, but is part of Campina Holland Cheese BV which is a separate operating and legal entity of Campina. The Divestment Business will be carved out from the current Campina organisation into a separate legal entity

Raw milk supply

(77) In order to ensure the undisrupted continuation of the Divestment Business, raw milk has to be supplied on a continuous basis on the basis of a Supply Agreement (see Annex 4). The Parties submit that the raw milk supplies are necessary to maintain the full economic viability and competitiveness of the Bleskensgraaf production facility during a transitional period of two years under a supply agreement. In order to facilitate the migration of the raw milk supply to other sources, like the Raw Milk Remedy, the merged entity commits to supply the raw milk required by the Divestment Business to sustain the current production volumes for a period of two years. The maximum annual volume of raw milk supplied to the Bleskensgraaf production facility will be 432 million kg, which is the volume of raw milk needed if the Bleskensgraaf production facility would produce at maximum capacity (48 KT). The raw milk supplied to the Bleskensgraaf production facility will meet all quality requirements which will apply to FrieslandCampina.

(78) Under the Raw Milk Remedy, the purchaser of the Divestment Business may continue to procure raw milk from the Dutch Milk Fund acting as agent for FrieslandCampina under the same conditions as under the supply agreement with FrieslandCampina.

(79) The supply agreement stipulates that, should the Dutch Milk Fund not become operational within two years, at the option of the buyer of the Divestment Business the duration of two year supply agreement may be extended until 1 January 2017 at the latest.

Price Raw milk

(80) The price for the Raw Milk under the Supply Agreement will be the monthly "guaranteed milk price" of FrieslandCampina being applicable in the month of delivery, including any costs as specified in the Supply Agreement. The Guaranteed Price will be subject to a rebate of 1% for the period until six months after the Dutch Milk Foundation has become operational within the meaning of clause 13.2. The guaranteed milk price is based on standard fat (4,41%) & protein (3,47%) contents and will be objectively adjusted for the actual fat and protein contents of the Raw Milk supplied (on the same basis as is being done for the members of FrieslandCampina). The guaranteed milk price is the cash paid out price exclusive of dividend, interest for debt, loans and other returns on member capital of FrieslandCampina, which is paid by FrieslandCampina to its members for the supply of Raw Milk.

(81) The guaranteed milk price is a representative price for raw milk in North West Europe and reflects the weighted average milk price over the year paid out by dairy companies in Denmark, Germany, Belgium and the Netherlands ("the Dairy Companies"), altogether constituting the average price paid for about 45 billion kg of raw milk. The milk prices of the Dairy Companies are based on verifiable sources like ZMP in Germany and LTO Nederland (Land- en Tuinbouw Organisatie Nederland) in the Netherlands. The development is determined by the monthly price developments in Germany, Belgium, Denmark and the Netherlands.

(82) The table below demonstrates how the guaranteed milk price using the weighted average milk price over the year paid out by the Dairy Companies is determined. In the last column of the table the weight attributed to each of the Dairy Companies involved is mentioned (i.e. CONO, Bel Leerdammer, DOC each account for 7.9%, Arla accounts for 10%, Milcobel accounts for 6.7% and the dairy companies reflected in the ZMP data account for 59.6%).

<table>

(83) According to the statutes of FrieslandCampina, the guaranteed price will be reviewed once every three years. However, such review is only meant to ensure that the guaranteed price is still a representative market price.

(84) Under the Supply Agreement, the Divestment Business will pay the guaranteed milk price to FrieslandCampina, minus a rebate as set out in paragraph 80, and including any costs as specified in the Supply Agreement and plus a fee for logistical and other costs. Logistics and transport (collection at farms and delivery at processing plants) as well as quality control, quality assurance and administration are costs borne by FrieslandCampina. These costs are not charged separately to the farmers. Consequently, the farmers receive the guaranteed price for the Raw Milk delivered without a discount for these costs. FrieslandCampina carries these costs as overhead. Therefore, these costs are included in the profit and loss account of FrieslandCampina and hence affect the profits of FrieslandCampina.

(85) In this situation, it goes without saying that the logistical and other costs relating to the collection of Raw Milk made available to the Divestment Business should be borne by Divestment Business. Otherwise, FrieslandCampina would have to carry these costs twice: on behalf of itself (in relation to its members) and in relation to the Raw Milk supplied to Divestment Business.

(86) All costs that arise from the logistics and transport (collection at farms and delivery at Bleskensgraaf) of Raw Milk sold to the Divestment Business as well as quality control, quality assurance and administration will be calculated in advance by FrieslandCampina and will be charged on an objective, transparent and verifiablebasis. The costs charged by FrieslandCampina to the Divestment Business per 100 kg Raw Milk are the same as for FrieslandCampina internally. The costs are based on optimal logistic solutions, maximum economies of scale and the average actual costs of FrieslandCampina per 100 kg Raw Milk (which is currently EUR [...]*). In other words, the Divestment Business will benefit from the economies of scale thatare achieved by FrieslandCampina. The (breakdown of the) costs concerned will be made transparent for the Divestment Business.

(87) The mentioned EUR [...]* is based on the actual costs FrieslandCampina has to make ([CONFIDENTIAL]*).

(88) In case of a dispute the Supply Agreement prescribes an efficient and speedy arbitration. Planning of the supplies of raw milk

(89) Under the two-year Supply Agreement, FrieslandCampina will supply all raw milk ordered by the Divestment Business with a maximum of the volume needed if the Bleskensgraaf production facility operates at full capacity. The volume needed from year to year will be communicated by the purchaser to FrieslandCampina six months before the beginning of the calendar year concerned. The annual forecast may be modified afterwards but shall be made definitive at the latest by 1 April of the calendar year concerned.

(90) The Bleskensgraaf production facility shall provide FrieslandCampina before the beginning of every month with a 13 month rolling forecast for the volume of Product required per month based on the Annual Forecast. Moreover, the Bleskensgraaf production facility shall provide the Supplier before the beginning of every week with a 13 week rolling forecast for the volume of Product required per week.

(91) The weekly planning provides for deliveries 7 days a week, unless explicitly agreed otherwise in writing. The Supplier shall accommodate normal fluctuation in ordered volume of Product during the week.

(92) The Annual Forecast of Product as provided by the Divestment Business in accordance with the Supply Agreement, will be supplied by FrieslandCampina to the Divestment Business in equal weekly volumes (the Average Weekly Volume), provided that orders may be adjusted by a maximum of +10% or -10% of the Average Weekly Volume in aggregate in any week. FrieslandCampina and the Divestment Business will agree that the volume supplied over a period of 13 weeks will correspond to the average volumes for a 13 weeks period (i.e. 13/52 x the Annual Forecast (the Average 13 Week Volume), but may be adjusted by a maximum of +5% or -5% of the Average 13 Week Volume. Finally, Cheese may deviate from the AnnualForecast over the entire calendar year concerned by a maximum of +5% or -5%.

(93) Experience shows that the flexibility percentages (yearly, monthly and weekly) are largely sufficient to accommodate the Divestment Business also in the light of commercial cycles.

(94) All other terms and conditions are described in detail in the Raw Milk supply agreement (Annex 4).

Supply of rest streams Cream and Whey

(95) The Divestment Business will at the option of the purchaser also include separate arrangements for the sale of the residual cream and whey of the Divestment Business production facility to FrieslandCampina for a period of eight years in proportion to the same quantities as the merged entity supplies raw milk to the Bleskensgraaf production facility, either under the two-year supply agreement or the Raw Milk Remedy. It should be stressed that the supply of rest streams to FrieslandCampina is at the option of the Purchaser of the Divestment Business. Hence, the Purchaser of the Divestment Business is free to sell the cream and whey rest streams on the market or use it for its own further processing. If the Purchaser of the Divestment Business wishes to sell the rest streams to FrieslandCampina, the following price calculation applies:

Price paid for rest streams

(96) The price of the residual cream (in proportion to the same quantities as the merged entity supplies raw milk to the Divestment Business production facility under the Supply Agreement or the Preferential Drawing Rights) will be determined on the basis of the fat-price as a component of the guaranteed price for raw milk (the garantieprijs), minus any costs for additional transport, handling and taxes which FrieslandCampina will incur.

(97) The residual cream currently produced by the Divestment Business can be transported over a distance of more than 300 km. Already within a range of 300 km from the Divestment Business there are various alternative outlets other than the facilities of the FrieslandCampina for the residual cream of the Divestment Business. There are for instance within this range producers of butter (e.g. VIV Vreeland BV, in Vreeland, the Netherlands and Corman, in Goé Belgium,), liquid cream (e.g. in the Netherlands: Koningszuivel in Bergambacht, Holland Dairy Star in Leusden, Katshaar in Coevorden, Twente Foods in Losser, Vreugdenhil in Voorthuizen, Van Roemburg in Putten and Lyempf in Kampen), spraycream (e.g. Incopack in Dilsen- Stokkem, Belgium and Hochwald in Erftstadt Germany), ice cream (IJspaleis in Sprundel, the Netherlands, IJsboerke in Tielen, Belgium and Unilever in Rotterdam, the Netherlands), whipped cream (e.g. Maitre Paul in Tilburg the Netherlands). In addition to direct sales the DB could also use traders such as Apollo, De Brandt and Hoogwegt that sell the cream to parties such as Milcobel, Inex and Danone.

(98) The price of the residual whey will be determined by a cost calculation on the basis of the whey price quotation published by the Dutch Dairy Organisation (Productschap Zuivel), minus any costs for additional transport and handling which FrieslandCampina will incur.

(99) The liquid whey currently produced by the Divestment Business is a concentrated product (i.e. the whey represents approx. 30% in dry matter). Therefore, the whey produced by the Divestment Business can be transported over distances of more than 300 KM. Already within a range of 300 km from the Divestment Business there are various alternative outlets other than the facilities of the FrieslandCampina for theresidual whey cream of the Divestment Business. The whey can be transported to the various potential alternative buyers in the Netherlands, Belgium, Germany and France (e.g. Vreugdenhil in Gorinchem, the Netherlands, Sloten in Deventer, the Netherlands, Eurolat in Cologne, Germany, or Milcobel in Antwerp, Belgium).

(100) It should be noted that with additional investments (approx. EUR [...]*) the Divestment Business can also further concentrate the whey up to 60 % in dry matter. This process will further reduce transportation costs enabling transports over distances of 800-1000 km. This concentrated product can be used for various purposes and the Divestment Business will have ample opportunities to supply this whey product to alternative outlets and within a range of 800-1000 km.

(101) All terms and conditions are described in detail in the supply agreements for Cream and Whey (Annexes 5 and 6).

Supply of other product and non-product related goods and services

(102) A potential purchaser may not be able to have purchasing agreements with all suppliers in place immediately at the closing date. The procurement of most product and non-product related goods and services is handled centrally by the Campina procurement function. Purchasers of all Campina operating companies are a member of such team. These teams source product categories or services and compile framework agreements for Campina as a whole. It will take some time for the Bleskensgraaf production facility to negotiate new contracts. Therefore, the Bleskensgraaf production facility needs sufficient time in order to agree on new contracts with its (new) suppliers.

(103) For this reason - and at the option of the purchaser - the Divestment Business will include an agreement for the supply by FrieslandCampina of product and non-product related goods required for the production of the current volumes of the Bleskensgraaf production facility product portfolio with a maximum duration of 12 months, to the extend permitted by those agreements. The product and non-product goods or services will be supplied by FrieslandCampina at the same conditions applicable to FrieslandCampina. It is noted that the ingredients that are currently sourced centrally by Campina and are used in the Bleskensgraaf production facility for the production of Dutch type cheese form approximately 1% of the total costs of the production of Dutch type cheese in the Bleskensgraaf production facility.

(104) With regard to IT and Payroll services which are handled centrally by Campina - at the option of the purchaser - a Service Level Agreement (SLA) on a cost plus basis with a maximum duration of 12 months needs to be in place.

Annex C: Schedule Long - Life Flavoured Dairy Drinks Scope of the Divestment Business

(105) The commitments offered comprise two brands which are qualified by the Commission as A-brands in the LLDD market. The proposal consists of a divestiture of 1) "Choco Choco", Campina's main brand in the chocolate flavoured LLDD segment in both Belgium and the Netherlands and 2) "Yogho Yogho", Campina'smain brand in the fruit flavoured LLDD segment in the Netherlands and in Belgium.

(106) The Purchaser of the Yogho Yogho brand will have to grant an exclusive, perpetual, royalty free, irrevocable license to FrieslandCampina for the use of the Yogho Yogho brand outside the Netherlands and Belgium. The presence of the Yogho Yogho brand outside the Netherlands and Belgium is limited.

(107) The divestiture of Choco Choco and Yogho Yogho also includes full access to recipes and formulations, copyrights to packaging and advertising and customer records in the form of data on customers, all customer lists with contact details, overdue lists with details per customer and credit terms. All these items which are included in the divestiture of the brands will also be included in the exclusive, perpetual, royalty free, irrevocable license that will be granted to FrieslandCampina by the Purchaser.

(108) At the option of the purchaser, the divestiture of Choco Choco and Yogho Yogho brand will include temporary two-year cost-plus sourcing agreements and manufacturing technical assistance. The proposed Divestment Business thus comprises a viable business which will be able to operate on a stand-alone-basis.

Choco Choco

(109) Choco Choco is Campina's main brand in the chocolate flavoured LLDD market segment in Belgium and in the Netherlands. Currently the brand Choco Choco is used for two types of chocolate flavoured LLDD, a classic version and a low-fat version. The classic version is sold as "Choco Classic" and the low-fat version as "Choco Choco". The Parties note that Choco Choco has generated in 2007 a turnover of EUR [...]* in Belgium (market share of [0-5]* in total LLDD market and [5-10]*% in chocolate flavoured segment) andEUR [...]* in the Netherlands (market share of [0-5]*% in the total LLDD market and [0-5]*% in chocolate flavoured segment).

Yogho Yogho

(110) Yogho Yogho is Campina's main brand in the fruit flavoured LLDD segment in the Netherlands. Under the brand name Yogho Yogho fruit flavoured LLDD are supplied with different flavours and different packaging. Currently Yogho Yogho is supplied in six flavours: strawberry, raspberry, pear, peach, yellow fruits and red fruits. In addition to the 1 litre gable tops, Yogho Yogho is also available in 500 ml plastic bottles and 200 ml cartons. In 2007 Campina achieved with Yogho Yogho in the Netherlands sales of EUR [...]* and a market share of [0-5]*% in the total LLDD market in the Netherlands and [10-20]*% in the fruit flavoured LLDD segment. In Belgium Campina achieved sales of EUR [...]* and a market share of [0-5]*% in the total LLDD market in Belgium and [0-5]*% in the fruit flavoured LLDD segment.

Strong brands and sourcing agreement

(111) The brands Choco Choco and Yogho Yogho in the Netherlands and in Belgium will provide the purchaser with strong brands in the LLDD market without needing the investment and time to build a new brand. Entry in the market for the purchaser of Choco Choco and Yogho Yogho will be facilitated by the proposed sourcing agreement with the Parties. This will allow the potential purchaser to be immediately present in the Belgian and Dutch markets without any production disruption. There is adequate capacity in the Campina factory in Aalter (Belgium), where all these products are/can be manufactured, to facilitate future growth. Price will be based on a cost-plus arrangement, including transportation costs. The purchaser(s) will therefore acquire viable brands and will be able to continue operating on a stand alone basis, subject to the proposed sourcing agreement during a transitory period of up to one year.

(112) Besides, the purchaser will have ample possibilities to source LLDD from other suppliers that already play a major role in the sourcing of LLDD to be sold under PL. Some examples are Inex, Inza, Müller, Immergut, Hochwald, Nöm or Milchunion Hocheifel.

Annex D: Schedule Raw Milk Scope of the Divestment Business

(113) The Parties submit two cumulating commitments with regard to the market for the procurement of raw milk.

The first commitment provides existing members of FC in the Netherlands a financial incentive to leave FC and to enter into a supply arrangement of any kind (either on a contractual basis or as member of a cooperative) with a certain minimum duration with a buyer of raw milk in the Netherlands.

(114) The second commitment aims to make raw milk available for plants of dairy processing companies ("DPC plants") that are registered with the Dutch Dairy Board ("Productschap Zuivel") as processor of raw milk in the Netherlands and that are or will become downstream competitors of the merged entity. The raw milk willbe made available by FC through an independent body (the Dutch Milk Foundation ("DMF")) by granting "drawing rights" to owners of DPC plants;

(115) The two commitments will be referred to as "Raw Milk Remedy 1" and "Raw Milk Remedy 2" and collectively as the "Raw Milk Remedies". The Raw Milk Remedies should be seen in conjunction with the sourcing arrangements that are offered in the context of the remedies for fresh dairy products and Dutch type cheese. These sourcing arrangements are meant for a transitional period not exceeding two years, after which the buyers of the divestment businesses in the markets for fresh dairy products and Dutch type cheese will be in a position to attract former members as suppliers (Raw Milk Remedy 1) or enter into supply agreements under Raw Milk Remedy 2.

(116) Raw milk as referred to in the Raw Milk Remedies ("Raw Milk") is untreated conventional raw milk produced in the Netherlands. For the purpose of Raw Milk Remedy 2, the following categories of Raw Milk are excluded:

* Organic raw milk;

* Specialty raw milk ("CLA") used for Campina's branded fresh dairy products;

* Milk used for the production of PDO (Noord-Holland) cheese;

* "Weidemelk" (to the extent used for other than fresh dairy applications).

(117) The Raw Milk covered by the Raw Milk Remedies meets all the EU hygiene and quality requirements controlled by an independent accredited quality control organisation, also known as Qlip.

(118) Raw Milk Remedy 1 provides for the grant of a lump sum to dairy farmers that leave FC as a member (the "Start Up Payment"). The amount of the Start Up Payment is 5 EUR per 100 kg raw milk delivered in the year immediately preceding the year in which the application for the Start Up Payment is made. Thus, the aggregate Start Up Payment to a dairy holding of average size in the Netherlands (500,000 kg/year) will be in the range of the average net yearly income of a dairy farmer (EUR 25,000).

(119) Any member of FC may apply for the Start Up Payment by giving three months written notice to FC, provided it becomes a supplier to any buyer of Raw Milk in the Netherlands ("Buyer"). After the three months notice period the applicant farmer may start supplying a Buyer forthwith. The Membership of the applicant farmer of FC is terminated automatically at the end of the first calendar year following the expiration of the three months notice period. Upon termination, applicant farmers may sell their Member Certificates and/or Member Bonds to FC or to remaining members at normal (=nominal) value. For an average farmer, the value of these Certificates and/or Bonds will be in the range of EUR [...]* (EUR [...]* per [...]* kg of raw milk).

(120) The Start Up Payment will be financed by FC and granted by DMF directly to the dairy farmer concerned on a case by case basis. DMF will monitor and supervise the application of Raw Milk Remedy 1, including verification that the applicant farmer concerned has indeed started supplying Raw Milk to a Buyer following termination of his supply obligations towards FC. DMF will pay out the Start Up Payment as soon as it has made the afore-mentioned verification, but in any event not later than two months from the date the applicant farmer has started supplying the Buyer.

(121) A dairy farmer that has been granted a Start Up Payment under Raw Milk Remedy 1 may rejoin FC as a member. However, if a member re-accedes to FC or definitively stops supplying Raw Milk altogether within three years from the date he ceases supplying Raw Milk to FC, he will have to repay the exit fee on a pro rata basis. The applicable entrance fee will be increased with the difference between the Start UP Paymentand pro rata paid back amount on a per KG basis. After three years from exit FC may apply the then applicable entrance fee to such former member who may wish to re-accede to FC. FC shall not offer terms toformer members more attractive than those offered to new members applicable at the relevant time.

(122) The total raw milk volume to which the Raw Milk Remedies apply is 1.2 billion kg. Before the end of each calendar quarter, the DMF will calculate the aggregate volume of raw milk for which applications have been made under Raw Milk Remedy 1. This aggregate volume will be deducted from the absolute total volume of raw milk that will have to be made available by FC under Raw Milk Remedy 2 as from the next calendar quarter until the Raw Milk Remedies will expire. However, if a member re-accedes to FC or definitively stops supplying Raw Milk altogether within three years from the date he ceases supplying raw milk to FC, the raw milk volume for which the Start Up Payment was paid is added to the volume that should be made available by FC through DMF from the date the re-accession becomes effective.

(123) Raw Milk Remedy 1 will remain effective until members representing a volume of 1.2 billion have left FC (taking into account paragraph 122 above), or until the Raw Milk Remedies have been withdrawn as a result of a review carried out by the Commission (cf. paragraph 124).

(124) DMF will remain in operation until the volume of raw milk to be made available by FC through DMF has been reduced to 0 following the application of Raw Milk Remedy 1. Moreover, the Commission will conduct a review at the request of FC following the abolition of the milk quota regime. The Commission will decide to withdraw the Raw Milk Remedy if it is satisfied that sufficient Dutch raw milk is available to Qualified Buyers as defined below.

(125) The volume of Raw Milk that will be made available under the Raw Milk Remedies should be more than sufficient (i) to allow access to raw milk for purchasers of the two divestments businesses in the markets for Fresh Basic Dairy Products and Dutch type nature cheese, and (ii) to facilitate access for existing competitors and new entrants that wish to build or increase production capacity in the downstream marketsconcerned within reasonable limits.

(126) Raw Milk Remedy 2 will be available to any buyer of Raw Milk that owns a DPC plant ("Qualified Buyer") that produces Fresh Basic Dairy Products, Dutch type nature cheese or one of these products in combination with other dairy products. Qualified Buyers can only buy Raw Milk for the benefit of:

(i) The DPC plants of the Divested Businesses of FC;

(ii) Expansion of production in existing DPC plants;

(iii) New DPC plants;

(iv) Additional DPC plants meant to increase production of Fresh Dairy Products including but not limited to Fresh Basic Dairy Products.

(127) The Parties note for the avoidance of doubt that the plants of the Divestment Businesses are within the scope of this category and that traders in raw milk and DPC plants of FC are outside the scope of Raw Milk Remedy 2.

(128) Fresh Basic Dairy Products as defined in the SO comprise plain milk, plain yoghurt in gabletop packaging and buttermilk and are produced in the Netherlands.

Raw milk remedy 1 - Incentive to leave

(129) Raw Milk Remedy 1 aims to decrease the market share of FC on the market for the procurement of Raw Milk in the Netherlands (expressed by the aggregate milk quotas of its members), by providing members of FC a financial incentive to leave FC and to continue their activities as a supplier of Raw Milk to users in the Netherlands.

(130) Since a cooperative does not have the statutory means to unilaterally terminate the membership of any members, unless for urgent reasons relating to the conduct of the member concerned, FC is not in a position to reduce its market share by terminating membership relations. FC, however, is not constrained by itsstatutes to provide financial incentives to its members to terminate their membership. The financial incentive of Raw Milk Remedy 1 should be sufficient to stimulate a significant number of members, representing a significant volume of Raw Milk, to leave FC as a member.

Raw Milk Remedy 2 - Dutch Milk Fund

(131) The Dutch Milk Fund ensures the supply of Raw Milk through an independent body (DMF) by granting "drawing rights" to DPC plants. FC will enter into contracts with Qualified Buyers of Raw Milk through the intermediary services of DMF. DMF will remain in operation until the volume of raw milk to be made available by FC through DMF has been reduced to 0 following the application of the Incentive to leave. Moreover, the Commission will conduct a review at the request of FC following the abolition of the milk quota regime. The Commission will decide to withdraw the Raw Milk Remedy if it is satisfied that sufficient Dutch raw milk is available to Qualified Buyers.

(132) In order to fulfil the role of an agent in the supply of Raw Milk, DMF will be designed as follows:

(a) DMF will be an independent non-profit organisation (in the form of a foundation -- stichting) acting as a contracting agent between FC and the DPC plants in the Netherlands;

(b) DMF will be incorporated before 1 July 2009;

(c) The costs of incorporation of DMF will be paid by FC;

(d) The operational costs of DMF will not be borne by the Qualified Buyers;

(e) The board of DMF consists of 3 qualified members, currently not active in the dairy sector and independent from and not related to the Parties, to be appointed by the Minister of Agriculture (Ministerie van Landbouw, Natuurbeheer en Voedselkwaliteit) or the Minister of Economic Affairs in the Netherlands;

(f) The operational execution of DMF's activities may be outsourced (entirely or partly) by the board (at its discretion) to (i) Productschap Zuivel, a semi-governmental organisation already executing EU market support regulations in the dairy sector (including milk quota system) or (ii) any other independent organisation;

(g) DMF will report annually to the Monitoring Trustee on the application of the present remedy.

(133) The maximum volume of Raw Milk that will be made available by FC in the form of "drawing rights" is 1.2 billion kg per year. Contracts must be for a fixed volume, such volume to be evenly spread over the contract term on a weekly basis plus or minus 5%, also taking into account seasonal fluctuations in raw milk production at normal commercial conditions. This means (in principle): A is 95 to 105% of B / 52 (+ or - the seasonal fluctuations of the total Raw Milk of FC) Whereby A = the volume to be delivered to the DPC Plants per week; B = the contract volume. A buyer may request a deviation from the above. FC will accommodate such requests applying a mark up on the guaranteed price. In case of disagreement about the amount of the mark up, DMF may give binding instructions to FC.

(134) Purchasers of the Divestment Businesses have preferential "drawing rights" up to the volume representing the total production capacity of those businesses. However, FC is not obliged to supply raw milk to the divestment businesses in excess of the volume that should be made available under this Remedy. In case preferential drawing rights exercised by the two divestment business exceed the aforementioned volume, the available volume will be allocated on a pro rata basis.

(135) These Purchasers may obtain Raw Milk via DMF on the same conditions as those of the supply agreementsoffered under the remedies for the Fresh Dairy and the Cheese businesses, subject to paragraph 137 below.

(136) Otherwise, Raw Milk is allocated on a first come, first serve basis. In case of force majeure (including animaldiseases), all Raw Milk supplied by FC will be allocated between FC and DPC plants on a pro rata basis(unless prohibited by the competent authorities). In case of disputes, DMF may give binding instructions to FC.

(137) Contracts will come into existence through DMF acting as agent for FC. DMF will prepare the contracts according to a standard format, but the contracts will be entered into by FC and the Qualified Buyers concerned. A contract may enter into force at the beginning of each calendar month with a fixed contract term of 12 months. The minimum volume per DPC plant per month is 2 million kg based on full truckloads. The minimum order term depends on the volume involved: between 2 and 4 million kg/month, the minimum order term is three months. For volumes of 4 up to 6 million kg/month it is 4 months. For volumes of 6 million kg/month or more, the order term is 6 months.

(138) Contracts for different periods and different volumes may be accumulated. However, contracts concluded by a single Qualified Buyer starting within a time period of two months will be accumulated for the application of the above-mentioned order terms as well.

(139) The price for the Raw Milk for all Qualified Buyers under the Raw Milk Remedy 2 will be the monthly "guaranteed milk price" of FC being applicable in the month of delivery minus 1% for the first five years from the date this Remedy becomes effective (i.e. DMF will be operational) and the aforesaid guaranteed milk price without reduction in the years thereafter. The guaranteed milk price is based on standard fat (4,41%) & protein (3,47%) contents and will be objectively adjusted for the actual fat and protein contents of the Raw Milk supplied (on the same basis as is being done for the members of FC). The guaranteed milk price is the cash paid out price exclusive of dividend, interest for debt, loans and other returns on member capital of FC, which is paid by FC to its members for the supply of Raw Milk.

(140) The guaranteed milk price is a representative price for raw milk in North West Europe and reflects the weighted average milk price over the year paid out by dairy companies in Denmark, Germany, Belgium and the Netherlands ("the Dairy Companies"), altogether constituting the average price paid for about 45 billion kg of raw milk. The milk prices of the Dairy Companies are based on verifiable sources like ZMP in Germany and LTO Nederland (Land- en Tuinbouw Organisatie Nederland) in the Netherlands. The development is determined by the monthly price developments in Germany, Belgium, Denmark and the Netherlands.

(141) The table below demonstrates how the guaranteed milk price using the weighted average milk price over the year paid out by the Dairy Companies is determined. In the last column of the table the weight attributed to each of the Dairy Companies involved is mentioned (i.e. CONO, Bel Leerdammer, DOC each account for 7.9%, Arla accounts for 10%, Milcobel accounts for 6.7% and the dairy companies reflected in the ZMP data account for 59.6%).

Details of guaranteed price Development of milk prices weighted by country (at 4.41 fat and 3.47 protein)

DOC 31.01 31.13 30.46 29.94 37.52 32.01 3.6 7.9 31.29 30.21 31.73 30.90 32.38 31.92 33.20 Av e. 5 y ear 45.0 3.0 4.5 26.8 3.6 3.6 Supplied kg milk Ave. Netherlands 31.80 31.16 30.93 30.45 37.55 35.55 35.57 32.70 35.23 37.41 37.73 2007 100 6.7 10.0 59.6 7.9 7.9 % Guaranteed price 31.09 30.45 29.87 29.47

Mi lcobel = Belgium 29.37 29.94 28.42 27.72 Arla = Denmark 34.00 31.79 30.54 29.61 ZMP = Germany 30.51 30.00 29.49 29.26 Bel Leerdammer 31.77 30.62 30.38 29.44 Cono 32.62 31.74 31.97 31.96 2003 2004 2005 2006 In NL total 10.6 b illion kg milk In DK 4.49 b illion In B 3.01 billion

(142) According to the statutes of FC, the guaranteed price will be reviewed once every three years. However, such review is meant to ensure that the guaranteed price is still a representative market price.

(143) Qualified Buyers buying Raw Milk via the DMF will pay the purchase price to FC plus a fee for logistical costs (see below).

(144) Logistics and transport (collection at farms and delivery at FC plants) as well as quality control, quality assurance and administration are and will be costs borne by FC. These costs are not charged separately to the farmers. Consequently, the farmers receive the guaranteed price for the Raw Milk delivered without a discount for these costs. FC carries these costs as overhead. Therefore these costs are included in the profit and loss account of FC and hence affect the profits of FC. This, however, is completely different from a hypothetical scenario where the logistical costs would be charged to the farmers concerned. Where the aim is to create a level playing field, it goes without saying that the logistical costs relating to the collection of Raw Milk made available to the Qualified Buyers procuring Raw Milk via DMF are borne by these Buyers. Otherwise, FC would have to carry these costs twice: on behalf of itself (in relation to its members) and in relation to the Qualified Buyers of the Raw Milk concerned.

(145) Logistics and transport (collection at farms and delivery at DPC plants) of Raw Milk sold via the DMF will be organised by FC as well as quality control, quality assurance and administration. Additional costs that arise from these activities will be calculated in advance by FC and will be charged on an objective and transparent basis. The costs charged by FC to DPC plants per 100 kg Raw Milk are the same as for FC internally. The costs are based on optimal logistic solutions, maximum economies of scale and the average actual costs of FC per 100 kg Raw Milk (which is currently EUR [...]*). In other words, the buyers of Raw Milk will benefit from the economies of scale that are achieved by FC. The (breakdown of the) costs concerned will be madetransparent for DMF and the Qualified Buyers. The allocation of costs by FC will be supervised by DMF and in case of a dispute, DMF may give binding instructions to FC.

(146) Upon the request of a Qualified Buyer buying Raw Milk, logistics, quality testing and administration may (partly) be organized by that Buyer, to be agreed upon between FC and the Buyer concerned. Collection on farms of members of FC may be effectuated by Qualified Buyers, only (i) upon consultation of FC and (ii) with the prior approval of the individual farmers concerned. If a Qualified Buyer decides to arrange collection and transportation of raw milk itself, it should engage to do so for at least three years without interruption. Such Qualified Buyer shall not be charged by FC any mark up for collection or transportation.

(147) DMF will monitor compliance by FC with the conditions of this remedy on an ongoing basis and may give binding instructions to FC in case of non-compliance. FC and DMF will agree upon an effective arbitrationclause, subject to approval of the Monitoring Trustee, in connection with disputes and complaints (on interpretation etc.) ensuring quick dispute resolution (and providing for continuity of supply of Raw Milk pending the arbitration procedure). This includes disputes between FC and DMF in connection with binding instructions issued by DMF.

(148) The Forms RM relating to fresh dairy products and cheese provide for supply agreements. As contemplated in Forms RM relating to fresh dairy products and cheese, the supply agreements are meant to be replaced by sourcing arrangements under this Raw Milk Remedy 2.

Notes

1 OJ L 24, 29.01.2004, p. 1.

2 OJ C ...,...200. , p....

3 OJ C ...,...200. , p....

4 Form CO, section 1, paragraph 10.

5 Form CO, section 1, paragraph 3.

6 Council Regulation (EC) n°1255/1999 of 17 May 1999 of the Common organisation of the market in milk and milk products, OJ L 160, 26.6.1999, p. 48-72.

7 See "A historical turnaround in global dairy - Implications for Europe" Report by Rabobank International, November 2007, available in Form CO, Annex 6.A.4

8 See Report by Rabobank International, referred to in footnote 7, p.30.

9 See Report by Rabobank International, referred to in footnote 7, p.18.

10 Apart for the distinction between conventional and organic raw milk. (See recitals 47-52).

11 See PowerPoint presentation dated 13 August 2008 "Cooperative Structure FrieslandCampina".

* Parts of this text have been edited to ensure that confidential information is not disclosed; those parts are enclosed in square brackets and marked with an asterisk

12 Form CO, Section 6.B.14, paragraphs 38 and 39. Campina has member farmers in the Netherlands, Belgium and Germany (see recital 53).

13 See Memorandum submitted by the notifying parties dated 12 August 2008 "Of members, milk prices, and the remuneration from financing" and the PowerPoint presentation dated 13 August 2008 "Milk pricing system FrieslandCampina".

14 Clause 10-5-5 of Merger Agreement.

15 See Powerpoint presentation dated 13 August 2008 "Financing instruments FrieslandCampina".

16 Article 6-3 of the Articles of Association of Campina (Form CO, Annex 6.B.7); Article 9-1 of the articles of association of Friesland Foods (Form CO, Annex 6.B.5).

17 Form CO, Section 6.B.13, paragraph 35.

18 Form CO, Section 6.B, paragraph 4.

19 Commission Decision of 10 June 2003 in Case No. COMP/M.3130 - Arla/Express Dairies, OJ C 297, 9.12.2003 p.25 and Commission Decision of 19 September 2006 in Case No. COMP/M.4344 - Lactalis/Nestlé/JV, OJ C 199, 24.8.2006, p. 2.

20 Form CO, Section 6B, paragraph 7.

21 OJ L 198 of 22.07.1991.

22 Form CO, Section 6.B.7, table 1.

23 Questionnaire to competitors raw milk procurement sent on 17 June 2008, Question 9 and 10.

24 Reply to the Statement of Objections, paragraphs 24 and 25.

25 Form CO, Section 6.B, paragraph 17.

26 In the Netherlands, quotas are managed by The Dutch Dairy Board (Productschap Zuivel).

27 Form CO, Section 6.B, paragraph 18.

28 Article 13 of the Articles of Association of Friesland Foods (Form CO, Annex 6.B.5).

29 Article 20 of the Articles of Association of Campina (Form CO, Annex 6.B.7).

30 Without relocating his farm in the other party's working area.

31 Case 1132 Friesland/De Kievit, decision of 7 July 1999.

32 Questionnaire to competitors raw milk procurement sent on 17 June 2008, Question 13.

33 Form CO, Section 7.B.17, para 64.

34 Answer from CO-RM-I-5 to questionnaire on raw milk procurement sent to competitors on 17 June 2008, question 15.

35 For Campina and Friesland Foods, these figures have been calculated by dividing the volume of raw milk procured in a given year by the number of farms.

36 Answers from CO-RM-I-14, CO-RM-I-3 and CO-RM-I-8 to questionnaire to competitors on raw milk procurement sent on 17 June 2008, question 11.

37 Form CO, Section 6.B, paragraph 23.

38 Answer from CO-RM-I-3 to 1st phase questionnaire to competitors on raw milk procurement sent on 17 June 2008, Question 28 " Pay-out price is the result of business results and based on the protein percentage and the fat percentage"; answer from CO-RM-I-14 to 1st phase questionnaire to competitors on raw milk procurement sent on 17 June 2008"(We) aim to pay out a price in conformity with the market and their own results".

39 Answer from CO-RM-I-5 to 1st phase questionnaire sent to competitors on raw milk procurement on 17 June 2008, question 28: "Friesland price is taken as a benchmark together with other regional and national buyers. Objective is to pay a competitive milk price, not on a monthly or yearly basis, but over a number of years. Since competition is made up of multiple factors, of which the Friesland milk price is only one, it can be considered as a loose benchmark. Finally the milk price paid has to be earned in the business results, which are made inside the EU and the world markets".

40 Articles from the website of Dutch newspaper Het Agrarisch Dagblad, dated 24 September 2007 and 16 January 2008, form CO, annex 6.B.9.

41 "Memorandum case COMP M.5046 Friesland Foods/Campina comments on the 6(1)c decision", dated 12 August 2008, page 7.

42 Decision COMP/M.3130 Arla Foods / Express Dairies of 10 June 2003.

43 Decision COMP M. 4344 Lactalis/Nestlé (JV) of 19 September 2006.

44 Form CO, Section 6.B, paragraph 63.

45 Friesland Foods procures organic raw milk in the same working area as where it procures conventional raw milk.

46 Source: Notifying parties' estimates, confirmed by the market investigation.

47 If the market for procurement of conventional raw milk were considered as sub-national in the manner proposed by the notifying parties, the overlaps between the notifying parties' activities would be limited. In Friesland Foods' working area, Friesland Foods is the dominant player with a share of [70-80]*% in volume in 2007 and Campina is almost not active. Competitors are DOC ([5-10]*%), Bel-Leerdamer ([0-5]*%), Vreugdenhill and Hochwald Nederland ([0-5]*% each). In Campina's working area, the combined market share of the new entity would be [60-70]*% (Campina: [60- 70]*%, Friesland Foods: [0-5]*%) with competition stemming from Bel-Leerdamer ([5-10]*%), DOC ([5-10]*%), Vreugdenhill ([5-10]*%) and Cono ([5-10]*%).

48 Guidelines on the assessment of horizontal mergers under the Council regulation on the control of concentrations between undertakings, OJ C31-5, 05.02.2004, paragraph 17. Case T-210-01 General Electric v Commission [2005] ECR II-5575, paragraph 115 and Case T-177/04 easy Jet v Commission [2006] ECR II-1913, paragraph 174.

49 Reply to the Statement of Objections, paragraph 29.

50 Reply to the Statement of Objections, paragraph 35.

51 Minutes of a conference call with DOC dated 25 July 2008.

52 See minutes of a conference call with CO-BD-I-27 dated 13 August 2008.

53 Reply to the Statement of Objections, paragraph 33.

54 Guidelines on the assessment of horizontal mergers under the Council regulation on the control of concentrations between undertakings, OJ C31-5, 05.02.2004, paragraph 61 ("Horizontal Merger Guidelines").

55 Neither of the notifying parties currently supplies raw milk to third parties. Therefore the impact on competitors with regard to access to raw milk is not direct as it would be the case in a "classical" vertical input foreclosure.

56 Answer from CO-RM-II-9 to 2nd phase questionnaire to competitors on raw milk procurement sent on 5 August 2008, question 24.

57 Answer from CO-RM-II-3 to 2nd phase questionnaire to competitors on raw milk procurement sent on 5 August 2008, question 24.

58 Answer from CO-RM-II-14 to 2nd phase questionnaire to competitors on raw milk procurement sent on 5 August 2008, question 25.

59 Answer from PE-RM-4 to 2nd phase questionnaire to potential entrants on raw milk procurement sent on 5 August 2008, question 16: "Eine deutliche Milchpreisesteigeruung von 5 bis 10% würde neue Milcherzeugerbetriebe anziehen, umliegende Wettbewerber warden weiter aus dem Markt gedrängt. Monopolstellung droht für Holland und auch angrenzende Lânder". Answer from PE-RM-2 to 2nd phase questionnaire to potential entrants on raw milk procurement sent on 5 August 2008, question 16 "Später (nach Vollzug der Fusion) ist es denkbar, dass das neue Gebilde in NL höhere Auszahlungspreise auch marktaktisch zum Squeeze-out von Wettbewerbn mit Milchbaisis in NL nutzt".

60 Minutes of conference call with CO-RM-I-4 of 25 July 2008.

61 Answer from CO-RM-II-6 to 2nd phase questionnaire to competitors on raw milk procurement sent on 5 August 2008, question 18. In its reply to Q.24 of the same questionnaire, CO-RM-II-6 underlined that "With their market share, FrieslandCampina would be able to set a political/tactical pay-out price for a period of time to squeeze out competitors in certain products areas in the Netherlands or in periods of milk shortage".

62 See table 6-1 and paragraph 79: [...]* kt for Lyempf and [...]* kt for Noorderlandmelk.

63 "FrieslandCampina: Comments on the Commission's 6(1)c decision in relation to the procurement of raw milk", RBB Economics, 25 July 2008.

64 The Commission maintains (see reply to the Statement of Objections, paragraph 37) that the notifying parties present these articles as elements which would show that "competitors see opportunities arising from the merger". See page 3 of the conclusion of the memorandum on raw milk sent by the notifying parties by e-mail on 12 August 2008, 18:56.

65 Memorandum dated 12 August 2008.

66 Form CO, Annex 7.I.7

67 Reply to the Statement of Objections, paragraphs 19 and 20.

68 Reply to the Statement of Objections, annex 2.1 "Note on dairy markets and their dynamics" by Roel Jongeneel, AERI, Wageningen and Vincent Requillart, INRA, Toulouse.

69 This volume would not be entirely available since the increase would also concern cooperatives such as Cono or DOC where farmers are also "locked-in" since they have to deliver the whole of their production to the cooperative.

70 Reply to the Statement of Objections, paragraphs 11 and 40.

71 First phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to customers, questions 11-13.

72 First phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to customers, questions 5 and 6 and first phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to competitors, questions 5 and 6.

73 For details see Section 7.3.

74 Commission Decision of 15.1.2007 in Case No COMP/M.4323 - Arla/Ingman Foods OJ C24, 2.2.2007, p.1, the Commission concluded that fresh and long-life milk form separate relevant product markets.

75 See Form CO Section 6.C.6-8.

76 See Annex 6.C.6 Form CO for Germany (23.5% microfiltration technology in 2007) and Annex 6.C.7 Form CO for the Netherlands (89.4% bactufugation technology in 2007).

77 European Commission, "Commission Notice on the definition of the relevant market for the purposes of Community competition law," OJ C 372, 9.12.1997, p.5 ("Relevant Market Notice", par. 20).

78 See Form CO Section 6.C.10-11 as well as Annexes 6.C.7-9, Annex 6.C.22 of Form CO. Claim reiterated in the notifying parties' reply to the Article 6.(1)(c) Decision "Memorandum Case M.5046 - Friesland Foods/Campina, Comments on the Article 6.1c decision of 17 July 2008", 12 August 2008.

79 See Form CO Section 6.C.14.

80 The notifying parties in their reply to the Statement of Objections did not respond in detail to the sections of the Statement of Objections dealing with the fresh dairy products. Their main comments related to the importance of Dutch origin of the fresh products, the analysis of internal pricing data by the Commission in its competitive assessment and the increase in supply by competitors. This decision will discuss these arguments in the sections on the relevant geographic market and the competitive assessment.

81 European Commission, "Commission Notice on the definition of the relevant market for the purposes of Community competition law," OJ C 372, 9.12.1997, p.5 ("Relevant Market Notice"), para 7.

82 First phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to customers, question 14.

83 First phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to customers, question 10 and first phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to competitors, question 13.

84 First phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to customers, question 10 and first phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to competitors, question 13.

85 In the Statement of Objections, an econometric result on estimation of retail demand systems was attached. The result complemented the other, larger set of qualitative evidence. The goal of the econometric model was to shed more light on specific product market definition questions. The notifying parties in their response to the Statement of Objections have raised a number of criticisms regarding the interpretation and the robustness of the econometric results (see RBB Economics, Case COMP/M.5046 Campina/Friesland Foods - Response to the Statement of Objections - Long Life Flavoured Dairy Drinks, 17 October 2008). Most of the criticisms put forward by the notifying parties are invalid or based on a misunderstanding of the methodology. Moreover, a number of criticisms regarding the robustness of the results can be addressed by extending the econometric model. The Commission has analysed these possibilities of additional arguments and extending its modeling framework and has concluded that, on balance, it does not attach weight to the econometric evidence concerning the definition of the relevant market (All modeling assumptions, the main results and conclusions of the SO, summary of the notifying parties' Replies to SO as well as details regarding the additional arguments and modeling are contained in Annex 1 and its attached Appendices). It is important to emphasise that this conclusion does not in any way contradict or affect the merits of the other qualitative and quantitative evidence put forward in this Decision on the definition of the relevant markets.

86 Relevant Market Notice, para 20.

87 According to the Relevant Market Notice potential competition is not taken into account when defining markets, since the conditions under which potential competition will actually represent an effective competitive constraint depend on the analysis of specific factors and circumstances related to the conditions of entry.

88 To illustrate this point, suppose that products A and B are not interchangeable from a demand viewpoint. Suppose further that some, but not a majority, of manufacturers of product B can readily switch production to manufacture product A. If, as a result, the markets for products A and B were aggregated, the market shares of the manufacturers of product A would be clearly underestimated. This is because by aggregating these two markets, the output of all manufacturers of product B would be taken into consideration for the calculation of market shares, thus ignoring that only some producers of B could switch to produce A. For similar reasons market aggregation is not meaningful when suppliers have the ability and incentive to switch only part of their production capacity from B to A. For example, brand repositioning may require fundamental changes in the nature of the product or it may mean the establishment of a new brand or the modification of an existing brand through changes in advertising and marketing strategy. Also, a "de novo" entrant, even if can enter quickly and at little sunk cost will not be included in the list of active market suppliers and assigned market shares. Hit-and-run entry, or where supply-side substitution is partial or not nearly universal are all taken into account only at the competitive assessment stage.

89 First phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to competitors, question 7.

90 First phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to competitors, question 14.

91 CO-BD-I-4 replied that such a switch would technically not possible because of "completely different ph-value, clearing of filling line necessary after each production, different packaging types and processing."

92 See reply of CO-BD-I-27 to question 14.

93 See reply of CO-BD-2-5 to question 16.

94 See reply of CO-BD-2-5 to question 16.

95 See reply of CO-BD-I-7 to additional questions and minutes CO-BD-I-2.

96 See Relevant Market Notice para. 23, where branded beverages are given as an example, but also applies to fresh basic dairy.

97 See Form CO Section 6.C.18.

98 See Form CO Section 6.C.18.

99 See notifying parties' reply to the Article 6.(1)(c) Decision "Memorandum Case M.5046 - Friesland Foods/Campina, Comments on the Article 6.1c decision of 17 July 2008", 12 August 2008, page 14.

100 Some customers in the OOH segment are directly targeted by dairy manufacturers. For details see section 7.1.1.3.

101 See for example Commission Decision in Case No. COMP/M.4533 - SCA/P&G, OJ 275, 16.11.2007, p.1.

102 Since private label products for fresh basic dairy are not present in the out of home segment, the question whether private labels must be distinguished from branded products only applies to the retail segment.

103 First phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to competitors, question 37 and first phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to customers, questions 44 and 45.

104 For the exact market share data see Section 7.2.3 on the competitive assessment.

105 First phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to customers, question 22 and first phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to competitors, question 16.

106 See Form CO Section 7.C.19-27. The figures refer to the overall market including sales by retailers and OOH wholesalers. Private label is only present in the retail segment, its share there would be 70.4 % (fresh milk), 50.5 % (fresh buttermilk) and 57.1 % (plain yoghurt).

107 See Form CO Annex 6.C.36.

108 Results are similar in terms of sales value.

109 This is in contrast to the market for fresh flavoured dairy drinks where products with an explicit health claim gain more and more importance. For details see Section 11.2.1.1.

110 The present case is different from Commission Decsion of 5.9.2007 in Case No COMP/M.4533 - SCA/P&G where the private label supply was supplied by a completely different set of producers compared to the branded market, thereby leading to "asymmetric" conditions of competition - while the branded producers could constrain the private label producers, the opposite was not possible as a private label producer would have to invest into a brand and marketing before being able to compete. The equally strong presence of Campina as well as Friesland Foods in private labels and branded products ensures symmetry in the conditions of competition.

111 Starting in mid 2007 the prices of all brands and private labels experienced an important price hike, driven by the increase in the price of raw milk. It appears, however, that the prices of low-level private labels increased relatively more than the price of high-level private labels. This has narrowed down the price difference between both types of private label to around 10-15% in the first quarter of 2008. The implications of this development are discussed below in the competitive assessment.

112 See for example reply CU-BD-2-12 minutes with CU-BD-2-8, who explained "that due to the fact that PL and A-branded products divide the market equally among them, a price which is too high set by Friesland would currently lead to a switch to PL".

113 See reply CU-BD-2-12.

114 See Form CO Section 6.C.28 Footnote 17.

115 See Commission Decision of 8.8.2001 in Case No. COMP/M.2399-Friesland Coberco/Nutricia, OJ C18, 22.1.2002, p.14.

116 See for example minutes CU-BD-2-7 and minutes CU-BD-2-1.

117 See reply to question 10 in M5223173-1-20385846 of 4 September 2008.

118 Second phase questionnaire customers fresh basic dairy products, question 7 and 8.

119 See for example minutes CO-BD-2-16, minutes CO-BD-2-20 or reply CU-BD-I-3.

120 Second phase questionnaire competitors value added yoghurt and quark, question 4, which covered general features of the OOH segment.

121 See Form CO Section 6.C.35.

122 Case M.3130-Arla Foods/Express Dairies and Case M.4117-Dairy Crest/Arla Foods both cover the UK.

123 See Market Share Data CD-ROM submitted 27-05-2008, Section Basic Dairy.

124 See "Memorandum Case M.5046 - Friesland Foods/Campina, Comments on the Article 6.1c decision of 17 July 2008", 12 August 2008, page 14.

125 See Form CO Annex 6.C.37 - at the end of 2007 the price difference between Belgium and the Netherlands for 1 l semi-skimmed fresh milk was EUR 0.3 (Belgium EUR 0.8, Netherlands EUR 0.5).

126 See replies CO-BD-I-11, CO-BD-I-15, CO-BD-I-19.

127 See replies CO-BD-2-5, CO-BD-2-4, CO-BD-2-22.

128 See CU-BD-2-11 "transport from further distances could have a negative effect on the expiry date and freshness." Similar CU-BD-2-12 who states that "importing foreign milk raises the very important issue of remaining shelf life. [...] Inherently, the remaining shelf life of fresh products is more limited if the product is sourced from abroad than in the case the product are sourced nationally."

129 See reply of CU-BD-2-8 "fresh basic dairy can have a burned taste"

130 See reply to Article 6.(1)(c) Decision page 14: "Only milk that has undergone en ESL treatment that prolongs the shelf life to 30 days will have a different taste than other fresh products."

131 See minutes CU-BD-2-8.

132 See minutes CU-BD-2-11

133 See reply CU-BD-I-1.

134 See Form CO Section 7.C.12-13.

135 See reply of CU-BD-I-10.

136 In their reply to the Statement of Objections, 17 October 2008, the notifying parties argue that "the reference to the promotion campaign is [...] entirely misplaced" (page 137) as the NZO had the aim "to promote milk in general and not in particular Dutch milk" (page 137). While indeed most of the campaigns available on the NZO website have no particular reference to Dutch origin, that seems to be based on the fact that they all at least 20 years old. Thus, fresh basic dairy products were produced and supplied locally and the fresh milk was identical to Dutch milk. Moreover, the notifying parties seem to ignore completely the fact that the NZO (which has 11 members, among them Campina and Friesland Foods) "aims to ensure that the dairy industry can preserve the Netherlands' high-quality and profitable and sustainable milk production in the long term. Dairy production should be based on respect for nature and environment, and dairy products should be able to meet the expectations of buyers and consumers to maintain the fine reputation of Dutch milk production [emphasis added]*." See http://www.nzo.nl/?PageID=82.

137 For some examples of past campaigns see http://www.zuivelonline.nl/?PageID=572.

138 See for example CU-BD-I-10 and CU-BD-I-1.

139 See http://www.campina.nl/default.aspx?selected=campina-nl.bijdeboer.boerderij-dag-2008&l=nl.

140 The Netherlands also has a schoolmilk programme, where children in primary schools receive milk (fresh Dutch milk). This programme has been in place for more than half a century. In their reply to the Statement of Objections (17 October 2008, page 136) the notifying parties clarified that [5-10]*% of children in primary school are currently supplied with Campina's fresh milk, thus the impact of the programme creating a "Dutch preference" seems to be limited.

141 "Melk is melk, zuivel is zuivel, maar Campina is anders. Daarom maakte Campina medio 2006 bekend dat in de lente van 2007 een nieuwe en unieke Campina melk op de Nederlandse markt wordt gebracht. Als grootste merk in de Nederlandse supermarkt (in 2006 voor het vijfde jaar achtereen) is Campina ook het grootste A-merk in melk. Met haar nieuwe standaard in zuivel onderscheidt Campina zich nadrukkelijk van andere melk. Want Campina melk heeft vanaf begin april een evenwichtiger vetzuursamenstelling (10 procent minder verzadigde vetzuren, 20 procent meer onverzadigde vetzuren en een verdubbeling van het gehalte aan Omega 3) omdat Nederlandse Campina-boeren hun koeien natuurlijk evenwichtiger bijvoeding geven in combinatie met weidegang. Campina stimuleert bovendien de teelt van verantwoord geproduceerde soja voor het veevoer van de koeien. De nieuwe melk is het schoolvoorbeeld van waar Campina voor staat: gezonde producten, duurzame productiemethoden, zorg voor koeien en landschap, en topkwaliteit boerderijmelk van boeren dichtbij consumenten. Het gaat om de grootste, aparte melkstroom ooit in Nederland."See http://www.campina.nl/Default.aspx?selected=campina-nl.nieuws&item=campinanl. nieuws.campina2006autonomegroeiineuropa

142 ."dat de melk die bij de Nederlandse consument komt, ook gemaakt is door de Nederlandse koelen" See "Campina communiceert ook op emotie" in Zuivel in Retail Dossier, ZuivelZicht, 8 November 2006, page 17-19), where the Campina's Group director consumer products is asked about the marketing strategy of the company in http://www.zuivelzicht.nl/Afbeeldingen/PDF/2006/dossier_retail.pdf. The article further mentions 8 metre high milk glasses at the border to create a typical Dutch feeling "Welkom thuis, we hebben jullie gemist. En de charme van de boerderij werd opnieuw ontdekt."

143 For more details on the Weidemelk concept see http://www.frieslandfoods.com/nl/frieslandfoods/themaseninitiatieven/Pages/Weidegang1.aspx or http://www.frieschevlag.nl/weidegang/index.html where it is stated that "Waarom stmuleert Friesche Vlag weidemelk? De Nederlandse koe hoort bij het typisch Nederlandse landschap. Friesche Vlag vindt het van belang dat de koe zichtbaar blijft in de wei".

144 A fact the notifying parties re-iterated in their reply to the Statement of Objections, 17 October 2008, page 137.

145 See Startverklaring weidegang melkkoeien submitted as Annex 2.3 by the notifying parties in M5223173/1/20385846.

146 For plain yoghurt 7 out of 8 products refer to Dutch origin, while for buttermilk 7 of 13 do. See Annex A.4a submitted by the notifying parties in M5084549-1-20385846.

147 See reply to question A4 submitted by the notifying parties in M5084549-1-20385846.

148 See reply to Article 6.(1)(c) Decision page 15.

149 Similar CU-BD-2-8 "the origin of the milk is of importance for especially the Dutch customer".

150 Similar competitors like CO-BD-2-16 "for Albert Heijn it will be difficult to explain that origin of fresh milk would be German" or CO-BD-2-20 (minutes) "waere aufgrund dieser Wahrnehmung der Konsumenten ein Absatz auslaendischer Ware fuer den Lebensmitteleinzelhandel (LEH) mit Schwierigkeiten verbunden."

151 First phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to competitors, question 26 and first phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to customers, questions 31.

152 Campina indirectly admits in an internal document that Dutch preferences could be viewed as specific. In fresh dairy Campina [...]*. See Annex 3.1 in reply M5223173-1-20385846 of 04-09-2008.

153 See Form CO Section 7.C.7-13.

154 See Reply to Statement of Objections, page 137, 17 October 2008.

155 See for example CU-BD-I-1 who stated that "Pre merger, CU-BD-I-1 has always been able to negotiate with either the merging parties knowing that it could go to the other if unfavourable terms were offered. Post merger, CU-BD-I-1 will no longer be in a position to play off one cooperative society against the other and will likely be confronted with significant price increases."

156 A view shared in general by the majority of respondents to the market investigation, which considered Friesland Foods and Campina as the closest competitors with respect to product range, price, ability to supply large quantities and brands - First phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to customers, question 56 and 57 and first phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to competitors, question 53 and 54.

157 In their reply to the Statement of Objections the notifying parties critice this analysis as an oversimplification as it would not take into account that factors other than competition between dairy companies would determine prices. They explain that the price decrease in private labels for [...]* would be due to a re-organisation by [...]* using distribution centres, thereby lowering the price for one litre private label milk by EUR [...]*. Since at the same time the price for one litre of private label fresh milk increased by EUR [...]*, the overall effect was a reduction of EUR [...]*. The Commission however did address the price differences between [...]* and [...]*, this difference narrowed to EUR [...]* as a result of a price increase for [...]*, which is the opposite of what the notifying parties have put forward. According to the notifying parties the gap would have increased by EUR [...]*. Moreover, the alleged price decrease at the beginning of 2006 was not unique to [...]*, but a general feature for all retailers delivered by Campina. Thus, the criticsm seems to be unfounded.

158 This is also evidenced by an internal document of Campina, where in the description of the current situation in 2007 it is noted that "[...]* "[...]*" - See Annex 3.1 in reply M5223173-1-20385846 of 04-09-2008.

159 RBB Economics, Case COMP/M.5046 Campina/Friesland Foods - Response to the Statement of Objections - Long Life Flavoured Dairy Drinks, 17 October 2008.

160 All modeling assumptions, the main results and conclusions of the Statement of Objections, summary of the notifying parties' replies to the Statement of Objections as well as details regarding the additional arguments and modeling are contained in Annex 1 and its attached Appendices.

161 The notifying parties have submitted examples of past switching behaviour in the Form CO Section 6.C.25. All the examples cover Friesland Foods losing eight contracts since 2005 - seven to Campina and one to Katshaar in 2008.

162 See minutes CU-BD-2-8.

163 See reply CU-BD-I-1.

164 See reply CO-BD-2-5.

165 See reply to second phase questionnaire competitors fresh basic dairy products.

166 See minutes CO-BD-1-27.

167 See for example CU-BD-2-12 "It must be noted that (...) is also dependent on the merging parties for the raw milk input for its private label products".

168 For a more detailed analysis see also section 6.4.1.

169 See minutes CO-BD-I-27.

170 See minutes CO-BD-2-20.

171 See Annex 3.1 in reply M5223173-1-20385846 of 04-09-2008.

172 Guidelines on the assessment of horizontal mergers under the Council regulation on the control of concentrations between undertakings, OJ C31-5, 05.02.2004, paragraph 65.

173 Guidelines on the assessment of horizontal mergers under the Council regulation on the control of concentrations between undertakings, OJ C31-5, 05.02.2004, paragraph 65.

174 Commission Decision of 31 January 2001 in Case COMP/M.2097 SCA/Metsa Tissue, OJ L57, 27.2.2002, pp.1-33.

175 Economic theory has also pointed to another, more subtle channel by which larger buyers can obtain a discount. Somewhat loosely speaking, if suppliers have strictly increasing incremental costs of production, then each of a number of small buyers essentially negotiates "at the margin", where incremental costs are higher. In other words, a supplier incurs high incremental costs when selling to small buyers because sales to small buyers come "on top" of the already large quantity that the supplier sells to other buyers, thereby justifying a higher price per unit for these buyers. In contrast, if these smaller buyers merged then they would account for a larger fraction of the supplier's total sales and would thus negotiate less "at the margin", thereby paying a lower price per unit. Note that this argument does not apply if the supplier has constant marginal costs. The available evidence suggests this is indeed the case in the supply of fresh milk and in particular private label where little or no marketing efforts or related fixed costs affect pricing.

176 A "gatekeeper's" bargaining position allows is the retailer to credibly argue that the supplier should give it a lower price because it will not have to worry about losing more profitable business with other buyers that compete in the same market and that would purchase at higher prices (because there are no other buyers in the same market).

177 See for example CU-BD-I-1 who stated that "Pre merger, CU-BD-I-1 has always been able to negotiate with either the merging parties knowing that it could go to the other if unfavourable terms were offered. Post merger, it will no longer be in a position to play off one cooperative society against the other and will likely be confronted with significant price increases."

178 A view shared in general by the majority of respondents to the market investigation, which considered Friesland Foods and Campina as the closest competitors with respect to product range, price, ability to supply large quantities and brands - First phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to customers, question 56 and 57 and first phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to competitors, question 53 and 54.

179 RBB Economics, Case COMP/M.5046 Campina/Friesland Foods - Response to the Statement of Objections - Long Life Flavoured Dairy Drinks, 17 October 2008.

180 All modeling assumptions, the main results and conclusions of the Statement of Objections, summary of the notifying parties' replies to the Statement of Objections as well as details regarding the additional arguments and modeling are contained in Annex 1 and its attached Appendices.

181 See Form CO Section 7.H.22.

182 See for example CU-BD-I-1 who stated that "Pre merger, CU-BD-I-1has always been able to negotiate with either the merging parties knowing that it could go to the other if unfavourable terms were offered. Post merger, it will no longer be in a position to play off one cooperative society against the other and will likely be confronted with significant price increases."

183 A view shared in general by the majority of respondents to the market investigation, which considered Friesland Foods and Campina as the closest competitors with respect to product range, price, ability to supply large quantities and brands - First phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to customers, question 56 and 57 and first phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to competitors, question 53 and 54.

184 RBB Economics, Case COMP/M.5046 Campina/Friesland Foods - Response to the Statement of Objections - Long Life Flavoured Dairy Drinks, 17 October 2008.

185 All modeling assumptions, the main results and conclusions of the Statement of Objections, summary of the notifying parties' replies to the Statement of Objections as well as details regarding the additional arguments and modeling are contained in Annex 1 and its attached Appendices.

186 The notifying parties have submitted examples of past switching behaviour in the Form CO Section 6.C.25. All the examples cover Friesland Foods loosing eight contracts since 2005 - seven to Campina and one to Katshaar in 2008.

187 See CU-BD-I-1.

188 See CU-BD-I-1.

189 See Form CO Section 7.H.22.

190 See Form CO Section 6.D.3-6.

191 Relevant Market Notice, para. 20.

192 See Form CO Section 6.D.8-9 as well as Annexes 6.D.4, Annex 6.D.5 of Form CO.

193 See Form CO Section 6.D.10.

194 First phase questionnaire long-life milk, long-life buttermilk, and long-life to customers, question 10 and first phase questionnaire long-life milk, long-life buttermilk, and long-life to competitors, question 11.

195 See for example reply of customer CU-LLD-1-10 who stated that "consumers do not interchange within the category long-life because of different product characteristics."

196 First phase questionnaire long-life milk, long-life buttermilk, and long-life to customers, question 10 and first phase questionnaire long-life milk, long-life buttermilk, and long-life to competitors, question 11.

197 See reply CU-LLD-1-10 to first phase questionnaire long-life milk, long-life buttermilk, and long-life custard to customers.

198 See reply CU-LLD-1-1 to first phase questionnaire long-life milk, long-life buttermilk, and long-life custard to customers.

199 According to the Relevant Market Notice potential competition is not taken into account when defining markets, since the conditions under which potential competition will actually represent an effective competitive constraint depend on the analysis of specific factors and circumstances related to the conditions of entry.

200 First phase questionnaire long-life milk, long-life buttermilk, and long-life custard to competitors, question 7.

201 First phase questionnaire long-life milk, long-life buttermilk, and long-life custard to competitors, question 13.

202 First phase questionnaire long-life milk, long-life buttermilk, and long-life custard to competitors, question 13c.

203 The notifying parties also provided market share data for long-life buttermilk and long-life custard. In long-life buttermilk Friesland Foods does not achieve a significant market share - below [0-5]*% - while in long-life custard in addition to Campina several other competitors are active. Thus, even assuming supply-side substitutability for the notifying parties, the merger would not have any impact on these two markets by eliminating a potential competitor. In addition, no customer has raised any concerns with respect to these markets. To the contrary, customers perceive these products to be in decline and soon to be phased out completely (see recital (374)).

204 See Form CO Section 6.D.10.

205 See Form CO Section 6.D.10.

206 See Form CO Section 6.D.10.

207 See for example Case No. COMP/M.4533 - SCA/P&G.

208 Since in the out of home segment private label products are not present, the discussion whether one has to distinguish private label from branded products does only apply to the retail segment.

209 First phase questionnaire long-life milk, long-life buttermilk, and long-life custard to customers, question 16 and first phase questionnaire long-life milk, long-life buttermilk, and long-life custard to competitors, question 15.

210 See Form CO Annex 7.D.2.

211 Figures based on IRI scanner data. Results are similar in terms of sales value.

212 This is in contrast to the market for fresh flavoured dairy drinks where products with an explicit health claim gain more and more importance. For details see 11.2.1.1.

213 First phase questionnaire long-life milk, long-life buttermilk, and long-life custard to competitors, question 36 and first phase questionnaire long-life milk, long-life buttermilk, and long-life custard to customers, question 36.

214 This has also been indicated during the market investigation. See for example first phase reply CULLD- 1-1 "private label products form part of the total formula and compete on price with the A brand products" and "it is very probable that customers would switch to branded long-life basic dairy products further to a permanent price increase for private label [...] products."

215 See Form CO Section 6.D.12 Footnote 5.

216 See Case No. COMP/M.2399-Friesland Coberco/Nutricia.

217 See reply to question 10 in M5223173-1-20385846 of 4 September 2008.

218 First phase questionnaire long-life milk, long-life buttermilk, long-life custard to competitors, question 14.

219 See for example reply of CU-LLD-1-2 to first phase questionnaire long-life milk, long-life buttermilk, long-life custard to customers who confirmed that a significant part of its long-life milk supply originates from Belgium and Germany.

220 First phase questionnaire long-life milk, long-life buttermilk, long-life custard to competitors, question 14.

221 See first phase reply of CO-LLD-1-4.

222 See Form CO Section 6.D.13.

223 Case M.3130-Arla Foods/Express Dairies.

224 First phase questionnaire long-life milk, long-life buttermilk, long-life custard to competitors, question 7 and first phase questionnaire long-life milk, long-life buttermilk, long-life custard to customers, question 24.

225 First phase questionnaire long-life milk, long-life buttermilk, long-life custard to competitors, question 28 and 29 and first phase questionnaire long-life milk, long-life buttermilk, long-life custard to customers, question 28 and 31.

226 First phase questionnaire long-life milk, long-life buttermilk, long-life custard to customers, question 28.

227 Reply of CU-LLD-1-1 to first phase questionnaire.

228 See Form CO Annex 7.D.1.

229 Following the notifying parties submission in the Form CO, the market shares for a potential OOH market would not be significantly different with a combined market share of [5-10]*% in 2007.

230 OJ L198 of 22.07.1991.

231 For details on organic raw milk see section 6.3.3.

232 Compared to organic products additional requirements like the use of manure (at least 60% organic), the use of feed (100% organic and at least 80% home grown) have to be fulfilled. Moreover, the farmer and the farm have to follow particular directions concerning the development of nature and the production process of end products.

233 Case 1173 Campina Melkunie - Zuiver Zuivel/De Vereenigung , decision of 01.02.1999.

234 See Form CO Section 6.P.12.

235 First phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to customers, question 10.

236 Second phase questionnaire to competitors organic fresh basic dairy, questions 5 and 12.

237 Relevant Market Notice, para. 20.

238 See Form CO Section 6.P.11.

239 First phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to competitors, question 16 and first phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to customers, question 22.

240 See Form CO Section 7.P.6.

241 First phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to competitors, question 37 and first phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to customers, questions 44 and 45.

242 See for example reply of CU-OM-2-13 or CU-OM-2-12.

243 See minutes CU-OM-2-3: "the EKO customer is not so focused on A-brands. Here, fancy or private labels [...] would be an alternative".

244 See Form CO Section 6.P.14.

245 Second phase questionnaire to customers organic fresh basic dairy, question 5.

246 Second phase questionnaire to competitors organic fresh basic dairy, question 5.

247 Since the market shares does not vary significantly across the different markets for organic fresh milk, organic buttermilk, organic plain yoghurt and organic custard, market shares are presented for the overall organic segment.

248 See reply CU-OM-2-1 second phase.

249 See reply CU-OM-2-1.

250 See reply CU-OM-2-13.

251 Second phase questionnaire to competitors organic fresh basic dairy, question 16.

252 See for example reply CO-OM-2-9.

253 See for example reply CO-OM-2-3.

254 Second phase questionnaire to competitors organic fresh basic dairy, questions 7 and 8.

255 Second phase questionnaire to competitors organic fresh basic dairy, question 17 and second phase questionnaire to customers organic fresh basic dairy, question 8.

256 See reply CU-OM-2-13.

257 See reply CU-OM-2-1.

258 See reply CU-OM-2-1.

259 Form CO Section 6.I.4.

260 Submission of the notifying parties of 18 September 2008, annex 4.1; Form CO Section 8.I.2.

261 Estimates of the notifying parties, Form CO Section 7.I.7-9.

262 Form CO Section 6.I.28.

263 Submission of the notifying parties of 11 September 2008, p. 3, 20 et seq.

264 Submissions of the notifying parties of 11 September 2008, p. 3; submission of 12 August 2008, p. 32.

265 Figure 8-1; Form CO Section 6.I.33.

266 Submission of the notifying parties of 1 September 2008, annex 4.1.

267 The notifying parties estimate that in 2007 approximately 42 000 tonnes of Dutch-type cheese were imported while the Dutch Dairy Board (Productschap Zuivel) estimates that 25 000 tonnes of imported Dutch-type cheese were imported in the Netherlands (email of the Dutch Dairy Board of 15 September 2008).

268 Submission of the notifying parties of 1 September 2008, annex 6.1.

(*) According to the market investigation, both the volumes of Dutch-type cheese sold to specialised cheese wholesalers (see section 8.4.1.1) and to downstream distribution channels (see section 8.4.1 and for the retail level section 8.4.2.1) are lower than estimated by the notifying parties.

269 Reply of CUW-C-I-18 to question 23 of the phase I questionnaire to specialised cheese wholesalers.

270 Reply of the notifying parties to the Statement of Objections of 17 October 2008, p. 17, 18; submission RBB wholesalers, p. 17.

271 Form CO Section 6.I.5.

272 Replies of CUW-C-I-21 and CUW-C-I-8 to question 23 of the phase I questionnaire to specialised cheese wholesalers.

273 Submission of the notifying parties of 5 September 2008, p. 3; reply of CUW-C-I-21 to question 19 of the phase I questionnaire to specialised cheese wholesalers.

274 Reply of CUW-C-I-21 to question 24 of the phase I questionnaire to specialised cheese wholesalers.

275 Submission of the notifying parties of 5 September 2008, p. 5; replies to questions 35 and 36 of the phase I questionnaire to specialised cheese wholesalers.

276 Supermarkets, hypermarkets and discounters are hereinafter collectively referred to as "modern types of retail".

277 Commission Decision of 18 October 2007 in Case no. COMP/M.4761 - Bongrain/Sodiaal/JV, OJ C 295, 7.12.2007, p. 7, para. 23 et seq.

278 Replies to question 16 of the phase I questionnaire to competitors.

279 Replies to question 18 of the phase I questionnaire to retailers.

280 Replies to question 19 of the phase I questionnaire to specialised cheese wholesalers.

281 For instance, the price of Emmenthal cheese in supermarkets and other modern types of retail in the Netherlands is almost twice as high as the price of Gouda (ZMP, Der Käsemarkt in den Niederlanden, 2007, p. 51).

282 18.7% of all cheese sold in Dutch modern types of retail in 2007 is non-Dutch type foreign cheese, see ZMP, Der Käsemarkt in den Niederlanden, 2007, p. 42.

283 Many Dutch cheese producers were even unable to estimate the costs and time required, see replies to question 18 of the phase I questionnaire to competitors.

284 Replies to question 21 of the phase I questionnaire to specialised cheese wholesalers.

285 Commission Decision of 18 October 2007 in Case no. COMP/M.4761 - Bongrain/Sodiaal/JV, OJ C 295, 7.12.2007, p. 7, para. 18 et seq.

286 ZMP, Der Käsemarkt in den Niederlanden, 2007, p. 44; minutes of telephone call with CUR-CNL-2-8 of 29 August 2008, confirmed on 16 September 2008.

287 Replies to questions 19 to 21 of the phase I questionnaire to retailers; replies to questions 15 of the phase I questionnaire to competitors; replies to question 18 of the phase I questionnaire to specialised cheese wholesalers.

288 Replies to question 5, 6 and 10 of the phase II questionnaire to competitors in the Netherlands.

289 Replies to questions 7, 8 and 9 of the phase I questionnaire to competitors.

290 Reply to question 17 of the phase I questionnaire to competitors; Form CO Section 6.I.32.

291 Replies to question 23 of the phase I questionnaire to specialised cheese wholesalers.

292 Replies of CUW-C-I-20, CUW-C-I-18, CUW-C-I-21 and CUW-C-I-17 to question 23 of the phase I questionnaire to specialised cheese wholesalers; note that slicing of rindless cheese is not necessary for industrial users.

293 Replies to question 23 of the phase I questionnaire to specialised cheese wholesalers.

294 Replies to question 20 of the phase I questionnaire to competitors.

295 Replies of CUR-C-I-24 and other companies active in modern types of retail to question 23 of the phase I questionnaire to retailers.

296 Replies to question 2 of the phase II questionnaire to retailers. 7% of all Dutch-type cheese sold in modern types of retail in 2007 was rindless cheese.

297 Replies to question 21 of the phase II questionnaire to specialised cheese wholesalers.

298 Submission of the notifying parties of 18 August 2008, annexes 27.1, 27.2 and 28.1; the input-corrected price correlations were submitted with the reply to the Statement of Objections.

299 Reply of the notifying parties to the Statement of Objections, annex 3.5.

300 Commission Notice on the definition of Relevant Markets for the purpose of Community Competition Law, 97-C 372-03, paras. 20-24.

301 See paras. 14, 20, 23 of the Relevant Market Notice.

302 Reply of the notifying parties to the Statement of Objections, p. 17, 18; submission RBB wholesalers of 17 October 2008, p. 17.

303 Minutes of DOC site visit of 30 October 2008.

304 DOC is a shareholder (together with the Dutch cheese producer Rouveen and the specialised cheese wholesaler Westland) in the specialised cheese wholesaler Cheese Partners Holland.

305 Reply of the notifying parties to the Statement of Objections, p. 30.

306 Replies to question 10 of the phase II questionnaire to competitors; reply of CO-CNL-2-1 to question 11 of the phase II questionnaire to competitors. Similarly, although the switching ability of competitors is more relevant for the question of supply-side substitutability than the ability of the notifying parties to switch, many factories of Friesland Foods and most factories of Campina produce either rindless or nature cheese, see submission of the notifying parties of 18 August 2008, annexes 4.1 and 4.2.

307 Reply of the notifying parties to the Statement of Objections, p. 11.

308 Minutes of telephone call with CUR-CNL-2-8 of 29 August 2008 and confirmed on 16 September 2008.

309 Submission RBB wholesalers, p. 11; reply of the notifying parties to the Statement of Objections, p. 30, 31. The notifying parties will therefore continue in-house storage of cheese during the first 15 days in the long run.

310 According to the notifying parties' submission of 16 September 2008, annex 5.1, approximately 85% of all downstream sales of cheese in the Netherlands are private label or non-branded. In the case of sales of Dutchtype cheese to specialised cheese wholesalers, the share of branded products is even lower (approximately 8%).

311 Form CO Section 7.I.27; submission of the notifying parties of 16 September 2008, annex 5.1. Replies to questions 28 and 31 of the phase I questionnaire to retailers.

312 Replies to question 44 and 48 of the to phase I questionnaire to retailers.

313 In COMP M.4761 Bongrain/Sodiaal, para. 13-17, a separate market for branded cheese was defined as opposed to private label cheese since the competitive conditions and structures upstream were held to be extremely different.

314 Replies to question 44, 45, 46, 47 and 48 of the phase I questionnaire to retailers.

315 Reply of CUW-C-I-7, CUW-C-I-17, CUW-C-I-18, CUW-C-I-20 and CUW-C-I-24 to question 12 of the phase I questionnaire to specialised cheese wholesalers.

316 Form CO Section 6.I.38.

317 Submission of the notifying parties of 16 September 2008, annex 1.1.

318 Submission of the notifying parties of 16 September 2008, annex 1.1. In contrast, the sourcing of rindless cheese by specialised cheese wholesalers is flexible in time as rindless cheese can easily be stored in containers after it has left the production line.

319 Submission of the notifying parties of 16 September 2008, annex 1.1.

320 Form CO Section 6.I.32; submission of the notifying parties of 1 September 2008, annexes 16.1 and 16.2. The notifying parties have not submitted input-corrected price correlations in their reply to the Statement of Objections.

321 Reply of CUW-C-I-27 to question 63 of the phase I questionnaire to specialised cheese wholesalers.

322 On 18 August 2008, the notifying parties were requested to "further elaborate [their]* arguments as regards the question whether or not the sale of 15-day-old [cheese] is to be considered as a separate product market (for example supply-side substitutability etc.)." Also in their reply to the Statement of Objections of 17 October 2008, the notifying parties have confined themselves to stating that "the only difference [between 15-day-old cheese and matured cheese] is that matured cheese has ripened for a longer period than 15-day-old cheese", without elaborating to which extent 15-day-old cheese and other cheese are substitutable on the supply-side.

323 Campina considers that it would take between [5-10]* and [10-20]* months to build a ripening facility, and this timing is irrespective of the size of the warehouse. The notifying parties also admit that it is the post-entry price and not the pre-entry price that will trigger such an investment decision, see RBB submission wholesalers, p. 13 and 14.

324 Commission Decsion of 23 June 2008 in Case No. COMP-M.5047 - Rewe/Adeg, OJ C 177, 12.7.2008, p.6, paras. 24, 31; Commission Decision of 23 October 2000 in Case No. IV-M.2161 - Ahold/Superdiplo, OJ C 340, 30.11.2000, p.3, para. 15.

325 Commission Decsion of 23 June 2008 in Case No. COMP-M.5047 - Rewe/Adeg, OJ C 177, 12.7.2008, p.6, para. 34.

326 Replies of CUW-C-I-12, CUW-C-I-20 and CUW-C-I-22 to question 29 of the phase I questionnaire to specialised cheese wholesalers; email of CUO-C-2-6 of 16 September 2008; reply of CUW-C-2-13 to question 2 of the questionnaire to specialised cheese wholesalers of 10 September 2008. See also replies of CUW-C-2-21 and CUW-C-2-18 to question 2 of the questionnaire to specialised cheese wholesalers of 10 September 2008; email of CUO-C-2-6 of 16 September 2008; email of CUO-C-2-6 of 15 September 2008; email of CUO-C-2-2 of 15 September 2008; email of CUO-C-2-5 of 16 September 2008.

327 Reply of CUW-C-I-18 to question 29 of the phase I questionnaire to specialised cheese wholesalers. It is to be recalled that brands do not play an important role for the sale of Dutch-type cheese in the Netherlands in terms of volume.

328 Replies of CUW-C-2-21 and CUW-C-2-13 to question 2 of the questionnaire to specialised cheese wholesalers of 10 September 2008.

329 Replies of CUW-C-2-21 and CUW-C-2-16 to question 2 of the questionnaire to specialised cheese wholesalers of 10 September 2008.

330 Reply of CUW-C-2-18 to question 2 of the questionnaire to specialised cheese wholesalers of 10 September 2008.

331 Submission of the notifying parties of 11 September 2008, p. 3.

332 Submission of the notifying parties of 11 September 2008, p. 22.

333 See e.g. COMP-M. 4590 Rewe/Delvita, para. 12; COMP/M.5112 Rewe/Plus Discount, para. 17; COMP-M. 5047 Rewe/Adeg, para. 24.

334 Submission of the notifying parties of 11 September 2008, p. 19 et seq.

335 COMP-M. 5047 Rewe/Adeg, para. 34.

336 Form CO Section 6.I.14; reply of CUW-C-2-16 to question 2 of the questionnaire to specialised cheese wholesalers of 10 September 2008.

337 Form CO Section 6.I.14.

338 According to annex 3.8 of the reply of the notifying parties to the Statement of Objections of 17 October 2008, 68% of all outlets of the most important retail chains have cutting departments and such outlets would generate 85% of all turnover at retail level. However, these figures have not been calculated on the basis of all modern types of retail (for instance, Lid is not included in annex 3.8) and will therefore in reality be lower. Furthermore, in Section 6.I.14 of the Form CO the notifying parties submit that cheese counters in Dutch modern types of retail "have been reduced to small boutiques predominantly aimed at the sale of non-Dutch-type cheeses or cheese specialties or have been removed altogether".

339 Submission of the notifying parties of 11 September 2008, p. 3, 20.

340 Differences also exist as to the demand structure of modern types of retail which is characterised by the presence of six major retail chains who together account for 94% of all modern types of retail in the Netherlands (submission of the notifying parties of 11 September 2008, Annex 27; submission of the notifying parties of 1 September 2008, p. 14) while the demand structure of specialised cheese shops has been described as comprising many buyers (reply of CUW-C-2-16 to question 2 of the questionnaire to specialised cheese wholesalers of 10 September 2008).

341 Submission of the notifying parties of 11 September 2008, p. 23.

342 Reply of CUW-C-2-16 to question 2 of the questionnaire to specialised cheese wholesalers of 10 September 2008.

343 In addition to such differences in logistics, modern types of retail often use tenders for private label cheese and contracts with a term of usually 12 months while the situation is different for specialised cheese shops, open markets and other small retail outlets where short-term agreements are used (RBB submission retailers of 17 October 2008, p. 7. Reply of CUW-C-2-16 to question 2 of the questionnaire to specialised cheese wholesalers of 10 September 2008).

344 Replies of CUW-C-I-8 and CUW-C-I-27 to question 64 and 65 of the phase I questionnaire to specialised cheese wholesalers.

345 It is to be recalled that no overlaps exist between the notifying parties for Edam. Futthermore, no overlap exists between the notifying parties for 15-day-old Maasdam cheese.

346 Commission Decision of 24 April 2006 in Case No. COMP/M.4135 - Lactalis/Galbani, OJ C 286, 23.11.2006, p.7, para. 18-19. Commission Decision of 18 October 2007 in Case No. COMP/M.4761 - Bongrain/Sodiaal/JV, OJ C 295, 7.12.2007, p.7, para. 23 et seq.

347 See section 8.2.3 (iii) - (v) above.

348 See section 8.2.3 (i) - (ii) above.

349 See section 8.2.3 (vi) - (vii) above.

350 In their reply to the Statement of Objections, the notifying parties have submitted price correlations between nature Dutch-type cheese in the Netherlands and in Germany.

351 According to ZMP, Der Käsemarkt in den Niederlanden, 2007, p. 44, prices for 1 kg of Gouda cheese at retail level range from 3.77 EUR in Poland to 5.19 EUR in Germany, 6.04 EUR in the Netherlands, 8.23 EUR in France and 9.21 EUR in the UK. Also annex 7.I.13 to the Form CO contains price variations for example between the Czech Republic and Germany for Edam (3.22 EUR vs. 2.93 EUR).

352 Replies to question 3 of the phase II questionnaire to retailers.

353 Replies to questions 35 and 36 of the phase I questionnaire to specialised cheese wholesalers; reply of CUW-CI- 24 to question 41 of phase I questionnaire to specialised cheese wholesalers.

354 Form CO Section 6.I.10.

355 Dutch Dairy Organisation (Nederlandse Zuivel Organisatie, NZO), see http://www.nzo.nl.

356 Publication OJ C 61-15 of 06.03. 2008 and OJ C 57-39 of 01.03.2008. No final decision has yet been taken and complaints have been received.

357 Application "Beschermde geografische aanduiding voor Gouda Holland" page 4.

358 Internal document 19.1.12.B and 19.1.12.C provided by the notifying parties on 20 August 2008.

359 EC Registration of Gouda and Edam, an international consumer survey commissioned by the Dutch Dairy Council, p. 28, 29.

360 Replies to questions 8, 9, 10 and 11 of the phase II questionnaire to specialised cheese wholesalers.

361 Reply of CUW-C-I-20 to question 11 of the phase I questionnaire to specialised cheese wholesalers; reply of CUW-C-I-21 to question 19 of the phase I questionnaire to specialised cheese wholesalers; reply of CUW-C-I-8 to questions 18 and 38 of the phase I questionnaire to specialised cheese wholesalers.

362 Replies to question 23 and 24 of the phase II questionnaire to specialised cheese wholesalers and replies to question 38 of the phase I questionnaire to specialised cheese wholesalers.

363 Replies to question 24 of the phase II questionnaire to specialised cheese wholesalers.

364 Replies to question 11 of the phase II questionnaire to retailers.

365 Replies to question 36 of the phase I questionnaire to retailers and to question 12 of the phase II questionnaire to retailers; Minutes of telephone call with CUR-CNL-2-8 of 29 August 2008, confirmed on 16 September 2008.

366 Minutes of telephone call with CUR-CNL-2-8 of 29 August 2008, confirmed on 16 September 2008.

367 Reply of CUR-CNL-2-8 to question 11 and 13 of the phase II questionnaire to retailers.

368 Reply of CUW-C-2-1 to question 23 of the phase II questionnaire to specialised cheese wholesalers.

369 Replies to questions 3 of the phase II questionnaire to retailers.

370 Replies to questions 17 et seq. and 30 et seq. of the phase I questionnaire to specialised cheese wholesalers.

371 Markets (i) and (ii) referred to in section 8.2.3.

372 Replies to question 14 of the phase II questionnaire to retailers.

373 Submission of the notifying parties of 1 September 2008, Annex 6.1.

374 Email of the Dutch Dairy Board (Productschap Zuivel) of 15 September 2008.

375 Submission of the notifying parties of 16 September 2008 annex 1.1. and annex 1.2 table 6; replies to question 3 of the phase II questionnaire to specialised cheese wholesalers and of the phase II questionnaire to retailers. Almost all imports of Dutch-type cheese by supermarkets are rindless cheese.

376 In 2007, the notifying parties have exported approximately [50-60]*% of their Dutch-type cheese production (approximately [...]* tonnes out of [...]* tonnes) which means that [...]* tonnes have been sold by the notifying parties in the Netherlands. Specialised cheese wholesalers have exported in 2007 a total of 98 000 tonnes. As explained below in section 8.4.1.8, it can reasonably be assumed that approximately [50-60%]* of this cheese exported by wholesalers is produced by the notifying parties and that therefore [60-70%]* of the notifying parties' cheese production are directly or indirectly exported. It follows that approximately [...]* tonnes of Dutch-type cheese produced by the notifying parties have been sold to downstream distribution channels in the Netherlands. On the basis of a total downstream volume of 232 000 tonnes (see email of the Dutch Dairy Board of 15 September 2008), this represents approximately [70-80]*%.

377 See also Form CO Section 6.I.38 according to which specialised cheese wholesalers not only purchase from manufacturers but also from each other and act at the same time as customers and suppliers of manufacturers.

378 Submission of the notifying parties of 1 September 2008, Annex 28.1. As point of departure, these estimates are based on the aggregate amount of exports by wholesalers of 98 000 tonnes (source: Dutch Dairy Board). Subsequently, the notifying parties used their best estimates for allocating these exports to the relevant wholesaler and for dividing their estimated aggregated downstream sales of wholesalers of 299 000 tonnes among the relevant wholesalers (submission of the notifying parties of 11 September 2008, p. 8-9, 18). The aggregate amount of wholesalers' exports of 98 000 tonnes is broadly in line with the aggregate amount obtained in the market investigation (118 500 tonnes after removal of obvious double-counting, see replies to question 16 of the phase II questionnaire to specialised cheese wholesalers).

379 The notifying parties submit that P. Bouter Kaascentrale is operationally very closely related to Albert Heijn, that Albert Heijn sources 15-day-old cheese and outsources it to Bouter and that Bouter matures cheese exclusively for Albert Heijn (see Form CO Section 6.I.32 and Section 7.I.15, submission of the notifying parties of 11 September 2008, p. 4).

380 Replies to question 16 of the phase II questionnaire to specialised cheese wholesalers. See also section 8.4.2.10 on a further split of these exports into nature and rindless cheese produced by the notifying parties.

381 Submission of the notifying parties of 16 September 2008, annex 1.2, table 6.

382 The market investigation has broadly confirmed the market shares of the notifying parties' competitors (DOC and Bel have slightly higher market shares while the market share of Cono is slightly lower). According to the market investigation imports into the Netherlands are only approximately 5% and the total market volume is lower than estimated by the notifying parties. The notifying parties' combined market share is therefore likely to be higher than estimated by the notifying parties.

383 Replies to question 7 of phase II questionnaire to specialised cheese wholesalers. See also RBB submission wholesalers p. 16 according to which these producers focus on branded and/or specialty cheeses.

384 In their submission of 12 August, p. 37 the notifying parties state that DOC produces at present both nature and rindless, but that [70-80]*% of its production is rindless. See also e-mail of a specialised cheese wholesaler of 31 October 2008 and reply of CUW-C-2-1 to question 7 of the phase II questionnaire to specialised cheese wholesalers.

385 The notifying parties also argue that the notifying parties' cheese production capacities are [...]*.

386 Submission of the notifying parties of 16 September 2008, annex 1.1.

387 Replies to questions 7, 8, 9, 10 and 11 of the phase II questionnaire to specialised cheese wholesalers.

388 Submission RBB wholesalers, p. 16.

389 DOC has an overall capacity of 125 000 tonnes and expects its output in 2008 to be 100 000 - 110 000 tonnes (minutes of site visit to DOC of 30 October 2008). DOC could switch even more than 74 000 tonnes if it partly outsourced maturing.

390 Form CO Section 6.I.32.

391 Reply to question 2 of phase II questionnaire to competitors.

392 Reply of the notifying parties to the SO, p. 18.

393 Email of DOC of 9 October 2008.

394 Submission of the notifying parties of 5 September 2008, p. 3.

395 Reply of the notifying parties to the Statement of Objections, p. 18.

396 Minutes of site visit to DOC of 30 October 2008.

397 Document C(2003) 3218 Europese Commissie - Steunmaatregel nr. N 499-02.

398 Form CO Section 7.I.7; RBB submission retailers, p. 15.

399 Minutes of site visit to DOC of 30 October 2008.

400 E-mail correspondence with specialised cheese wholesalers and replies to question 7 of phase II questionnaire to specialised cheese wholesalers.

401 Reply of CUR-CNL-2-10 to question 6 of the phase II questionnaire to retailers.

402 Replies to questions 7, 8, 9, 10 and 11 of the phase II questionnaire to specialised cheese wholesalers.

403 Reply of CUW-C-2-16 to question 11 of the phase I questionnaire to specialised cheese wholesalers; reply of CUW-C-1-21 to question 19 of the phase I questionnaire to specialised cheese wholesalers.

404 Replies to question 23 of the phase II questionnaire to specialised cheese wholesalers and replies to question 38 of the phase I questionnaire to specialised cheese wholesalers.

405 Replies to question 24 of the phase II questionnaire to specialised cheese wholesalers.

406 Minutes of telephone call with a specialised cheese wholesaler of 29 August 2008, confirmed on 16 September 2008.

407 Section 6.I.43 of the Form CO states that the wholesaler's switch to Germany "took a few years".

408 Reply of CUW-C-I-18 to question 67 of the phase I questionnaire to specialised cheese wholesalers; reply of CUW-C-I-22 to question 29 of the phase I questionnaire to specialised cheese wholesalers; reply of CUW-C-I- 21 to questions 39 and 42 of the phase I questionnaire to specialised cheese wholesalers.

409 Reply of CUR-CNL-2-8 to question 11 of the phase II questionnaire to retailers.

410 Replies to question 3, 5, 6, 8 and 9 of the phase II questionnaire to competitors outside the Netherlands.

411 Reply of CUW-C-I-27 to question 16 of the phase I questionnaire to specialised cheese wholesalers.

412 Reply of CUW-C-I-27 to question 63 and 64 of the phase I questionnaire to specialised cheese wholesalers.

413 Replies to question 4, 10 and 11 of the phase II questionnaire to specialised cheese wholesalers.

414 Replies of CUW-C-I-2, CUW-C-I-7, CUW-C-I-12, CUW-C-I-18, CUW-C-I-20, CUW-C-I-17, CUW-C-I-24, CUW-C-I-4 to question 62 of the phase I questionnaire to specialised cheese wholesalers.

415 Replies of CUW-C-I-2, CUW-C-I-7, CUW-C-I-8, CUW-C-I-21, CUW-C-I-22, CUW-C-I-17, CUW-C-I-24, CUW-C-I-4, CUW-C-I-27 to question 66 of phase I questionnaire to specialised cheese wholesalers.

416 Reply of CUW-C-I-17 to question 66 of the phase I questionnaire to specialised cheese wholesalers.

417 Reply of CUW-C-I-18 to question 67 of the phase I questionnaire to specialised cheese wholesalers.

418 Within the next five years CO-CNL-2-2 plans to expand its production capacity by 1 400 tonnes. This extra capacity will be used for the production of cheese specialities. CO-CNL-2-4 is planning to expand its current production with 8 000 tonnes of rindless cheese within the next five years. CO-CNL-2-1 is planning to expand its production capacity for nature Gouda cheese with 10 000-16 000 tonnes before 2015. CO-CNL-2-3 has the possibility to expand one of its current production plants and to build a third plant which is expected to be operative between 2012 and 2016 which will in turn lead to the closing of its old factory.

419 Submission of the notifying parties of 16 September 2008, annex 1.1; such 15-day-old cheese constitutes approximately 40% of all Dutch-type cheese purchased by specialised cheese wholesalers.

420 Form CO Section 6.I.31.

421 According of the submission of the notifying parties of 1 September 2008, annex 4.1, they have exported 297 000 tonnes of Dutch-type cheese in 2007. See also email of the Dutch Dairy Board (Productschap Zuivel) of 15 September 2008.

422 Around [...]* tonnes of Dutch-type cheese were exported to Germany and Belgium in 2007. Based on internal documents of the notifying parties concerning the notifying parties' best-selling products (more than [40-50]*% of the notifying parties' sales), [50-60%]* of this cheese was exported to retailers and retailers grocery, and another [20-30]*% was exported to industrial users and producers; it is unlikely that exports to these customer groups will be rechanneled to the Netherlands. According to the notifying parties' internal data, less than [10- 20]*% is exported to traders in Belgium and Germany, that is the only customer group that is likely to re-import the notifying parties' cheese, see submission of the notifying parties of 1 September 2008 - annex 4.1 and internal data of the notifying parties (SAS system).

423 Reply of CUW-C-I-18 to question 42 of the phase I questionnaire to specialised cheese wholesalers.

424 Reply to CUW-C-I-17 to question 34 of the phase I questionnaire to specialised cheese wholesalers.

425 Reply of CUW-C-2-8 to question 22 of the phase II questionnaire to specialised cheese wholesalers.

426 Submission of the notifying parties of 16 September 2008, Annex 1.1.

427 Internal information provided by the notifying parties (SAS system).

428 Submission of the notifying parties of 1 September 2008, annex 4.1.

429 In their submission of 11 September 2008, p. 12 and 13, the notifying parties explained that due to a lack of data regarding the individual wholesalers no assessment could be made as to how much of their cheese is exported by wholesalers. In the same submission p. 8-9 the notifying parties use data provided by the Dutch Dairy Board (Productschap Zuivel) on the aggregate amount of Dutch-type cheese exported by wholesalers (98 000 tonnes). It can reasonably be assumed that cheese produced by the notifying parties accounts for [50-60%]* of these exports since the notifying parties have a significant market position as regards the sales of Dutch-type cheese to specialised cheese wholesalers (see section 8.4.1) and since it can be inferred from figure 8-1 that specialized cheese wholesalers have exported 37% of all Dutch-type cheese sourced by them in the Netherlands ([30-40]*% of all Dutch-type cheese sourced from the notifying parties ([...]* tonnes) amounts to [...]* tonnes); the notifying parties have also applied this ratio in their submission of 27 October 2008 on remedies cheese, p. 5. This conclusion is also in line with the results of the market investigation (replies to questions 4 and 16 of the phase II questionnaire to specialised cheese wholesalers) and the general trend that two thirds of the Dutch production of Dutch-type cheese is exported (on the basis of [...]* tonnes direct exports and [...]* tonnes of indirect exports, [60-70%]* of the notifying parties' total production of Dutch-type cheese is exported).

430 Note that the notifying parties were unable to provide any indication how much of their cheese is exported by wholesalers, submission of the notifying parties of 11 September 2008, p. 12 and 13.

431 Reply of the notifying parties to the Statement of Objections, p. 19.

432 Reply of the notifying parties to the Statement of Objections, p. 20. 433 Approximately [60-70]*% of Friesland Foods' production and approximately [50-60]*% of Campina's production, see notifying parties' submission of 18 August 2008, annexes 7.1 and 7.2.

434 According to annex 6.I.7 to the Form CO, the average spare capacity at specialised cheese wholesalers is [...]* tonnes per year.

435 Replies of CUW-C-2-16, CUW-C-2-17 and CUW-C-2-18 to question 14 of the phase II questionnaire to specialised cheese wholesalers.

436 Reply of CUW-C-2-17 to question 25 of the phase II questionnaire to specialised cheese wholesalers.

437 Replies of CUW-C-2-14 and CUW-C-2-16 to question 25 of the phase II questionnaire to specialised cheese wholesalers.

438 For instance, reply of the notifying parties to the Statement of Objections p. 14 and p. 38; submission of the notifying parties of 7 November 2008, p. 11.

439 Submission of the notifying parties of 16 September 2008, annex 1.1.

440 Submission of the notifying parties of 16 September 2008, annex 1.1.

441 Submission of the notifying parties of 16 September 2008, annex1.1.

442 Submission of the notifying parties of 16 September 2008, annex 1.1.

443 Submission of the notifying parties of 16 September 2008, annex 1.1.

444 Form CO and submission of the notifying parties of 16 September 2008, annex 1.1.

445 Submission of the notifying parties of 16 September 2008, annex 1.1. Imports of 15-day-old cheese into the Netherlands are insignificant.

446 Replies of CUW-C-I-2, CUW-C-I-7, CUW-C-I-12, CUW-C-I-18, CUW-C-I-20, CUW-C-I-17, CUW-C-I-24, CUW-C-I-4 to question 62 of the phase I questionnaire to specialised cheese wholesalers. Replies to question 4 of the phase II questionnaire to specialised cheese wholesalers.

447 Submission of the notifying parties of 16 September 2008, annex 1.1.

448 Submission of the notifying parties of 16 September 2008, annex 5.1.

449 OOH and industry channel based on Friesland Foods' market analysis for the Benelux, submission of the notifying parties of 11 September 2008.

450 Submission of the notifying parties of 11 September 2008, reply to question 27.

451 The notifying parties explain that AC Nielsen data are imprecise since the volume of cheese sold over the counter ("OTC") is variable, some prepacked packages of cheese contains an extra free slice of cheese and since no account is taken of vaporisation, waste and theft of cheese (submission of the notifying parties of 18 September 2008, p. 21). AC Nielsen however covers the OTC segment (ZMP, Der Käsemarkt in den Niederlanden, 2007, p. 41). On 7 November 2008, the notifying parties explained that the Commission had misunderstood the AC Nielsen since those data only related to Dutch-type cheese. However, it is clear that the AC Nielsen data at issue cover both Dutch-type cheese (81% of the volume indicated by AC Nielsen) and non-Dutch-type cheese (18.7% of the volume indicated by AC Nielsen), see ZMP, Der Käsemarkt in den Niederlanden, 2007, p. 24.

The notifying parties themselves use AC Nielsen data on the Dutch cheese market in the Form CO Section 6.I.26 and Section 7.I.24.

452 For example, the notifying parties suggest that all 730 000 tonnes of cheese produced in the Netherlands are Dutch-type cheese (see presentation of the notifying parties of 18 September 2008 and figure 8-1) while according to the Dutch Dairy Board (Productschap Zuivel - e-mail of 15 September 2008) 71 000 tonnes of cheese produced in the Netherlands fall into the category of "others" (it can be assumed that this category, for example, comprises Emmenthal produced by Friesland Foods). Similar observations apply for the notifying parties' correction of import and export data. Also the data provided by the Dutch Dairy Board by email of 15 September 2008 on production, import and export of Dutch-type cheese (including varieties) differ considerably from the data prepared by the notifying parties. Furthermore, ZMP estimates the production (based on national statistics) in the Netherlands of Gouda, Maasdam and Edam to be 594 000 tonnes in 2005 (ZMP, der Käsemarkt in den Niederlanden, p. 28) which is in contrast to the notifying parties' estimate of the total production in 2005 of 670 000 tonnes (see figure 8-1 above).

453 Sumission of the notifying parties of 11 September 2008.

454 Notifying parties' response of 11 September 2008 to question 17 of the Commission's request of 1 September 2008.

455 All main modern types of retail in the Netherlands (who together represent 94% of all modern types of retail) as well as further modern types of retail have participated in the market investigation and indicated that they sold 113 000 tonnes of Dutch-type cheese in 2007. The figure of 120 000 tonnes reflects an addition in order to achieve a full coverage of the level of modern types of retail.

456 See ZMP, Der Käsemarkt in den Niederlanden, 2007, p. 42. The Dutch retailer Koopconsult is not included in this market volume; according to the notifying parties the inclusion of Koopconsult leads to a market volume of [...]* tonnes (submission of the notifying parties of 7 November 2008, annex 5).

457 Minutes of telephone call with CUR-CNL-2-8 of 29 August 2008, confirmed on 16 September 2008.

458 This has also been admitted by the notifying parties after the Statement of Objections. In their latest submission of 7 November 2008, p. 15, the notifying parties stated that the total sales to modern types of retail should be in a range of [...]* to [...]* tonnes. On the basis of the submission by the notifying parties that modern types of retail account for [80-90]* % of all Dutch-type cheese sold at retail level and that the retail level accounts for [70-80]*% of the downstream level, this would lead to a total volume sold at downstream level of [...]*- [...]* tonnes and cheese produced by the notifying parties ([...]* tonnes, see section 8.4) would account for [70-80]*% of these sales.

459 The Dutch Dairy Board (Productschap Zuivel) has upon the Commission's request further broken down the consumption of Dutch-type cheese in the Netherlands into the different downstream distribution channels (Email of Dutch Dairy Board of 15 September 2008). According to this breakdown (which is partly based on extrapolated GfK market data), approximately 122 000 tonnes of Dutch-type cheese have been sold in Dutch supermarkets in 2007. Information provided by the Dutch Dairy Organisation (Nederlandse Zuivel Organisatie -NZO) and submitted by the notifying parties in their reply to the Statement of Objections submits that the Dutch Dairy Board has overestimated the size of the OOH segment although no definitive conclusion can be drawn on the actual size of the OOH segment since the NZO estimates, on which the notifying parties rely, are based on a study of 1998 and consumption trends have changed since 1998 (in 2007 34% of the total food turnover in the Netherlands is obtained in OOH, see Persbericht FSIN Horecava 2008 "Marktaandeel van de buitenshuis kanalen groeit sneller dan de totale foodmarkt"). Furthermore, in the email of 15 September 2008, the Dutch Dairy Board estimated the total consumption of Dutch-type cheese in the Netherlands to be approximately 232 000 tonnes. It can indeed be reasonably assumed that this total consumption volume of Dutch-type cheese in the Netherlands can be divided into (i) modern types of retail (122 000 tonnes, 53%), (ii) other small retail outlets (24 000 tonnes, 10%) and (iii) OOH and industry (86 000 tonnes, 37%). Also the notifying parties admit that modern types of retail and small retail outlets together may be "a few percentage points" less than [70-80]*%, see submission of 11 September 2008, p. 3.

460 Submission of the notifying parties of 1 September 2008, annex 22.2 as compared to the replies to the phase II questionnaire to retailers (if compared to annex 27 of the submission of the notifying parties of 11 September 2008, the deviation is approximately [5-10]*%).

461 Reply of the notifying parties to the Statement of Objections, annex 3.9. The figure of [...]* to [...]* tonnes can be obtained by adding Lidl's sales (source: questionnaire reply) and by deducting spreadable cheese.

462 Submission of 11 September 2008, reply to question 27.

463 Submission of the notifying parties of 11 September 2008, table 13; replies to phase II questionnaire to retailers.

464 As explained above, this also includes sales of 15-day-old cheese to Albert Heijn/Bouter.

465 The sales of Cheese Partners to modern types of retail reported by the notifying parties are approximately [...]* tonnes in 2007 and approximately [...]* tonnes in 2008 (RBB submission retailers, p. 9 and 10; submission of the notifying parties of 7 November 2008, p. 3).

466 Replies to questions 4 and 6 of the phase II questionnaire to specialised cheese wholesalers.

467 Submission of the notifying parties of 11 September 2008 - table 13.

468 Annex 3.10 of reply of the notifying parties to the Statement of Objections.

469 Form CO Section 7.I.7. As explained above, DOC has a stake in the specialised cheese wholesaler Cheese Partners together with Westland and Rouveen. The sales of Cheese Partners to modern types of retail reported by the notifying parties are approximately [...]* tonnes in 2007 and approximately [...]* tonnes in 2008 (RBB submission retailers, p. 9 and 10; submission of the notifying parties of 7 November 2008, p. 3).

470 Reply of CUW-C-I-21 to questions 64 and 65 of the phase I questionnaire to specialised cheese wholesalers.

471 Replies of CUW-C-2-1, CUW-C-2-12, CUW-C-2-8, CUW-C-2-14, CUW-C-2-15, CUW-C-2-16, CUW-C-2-17, CUW-C-2-13, CUW-C-2-21 to question 17 of the phase II questionnaire to specialised cheese wholesalers.

472 The notifying parties themselves submit that pre- and aftersales services are non-existent or of little importance in the cheese sector, Form CO Section 8.I.7.

473 Replies of CUW-C-2-8, CUW-C-2-15, CUW-C-2-14 to question 17 of the phase II questionnaire to specialized cheese wholesalers, reply of CUW-C-I-22 to question 41 of the phase I questionnaire to specialised cheese wholesalers.

474 Reply of CUW-C-I-17 to questions 64 and 65 I of the phase I questionnaire to specialised cheese wholesalers.

475 Reply of CUW-C-I-18 to question 58 of the phase I questionnaire to specialised cheese wholesalers.

476 See replies of CUW-C-I-8, CUW-C-I-17, CUW-C-I-4, CUW-C-I-2 and CUW-C-I-18 to question 58 of the phase I questionnaire to specialised cheese wholesalers.

477 Reply of CUW-C-I-18 to questions 64 and 67 of the phase I questionnaire to specialised cheese wholesalers.

478 Reply of CUW-C-I-4 to question 64 and 65 of the phase I questionnaire to specialised cheese wholesalers.

479 Reply of CUW-C-I-21 to questions 66 of the phase I questionnaire to specialised cheese wholesalers.

480 CUW-C-I-8, reply to question 64 of the phase I questionnaire to specialised cheese wholesalers.

481 Reply of CUW-C-I-27 to questions 64 and 66 of the phase I questionnaire to specialised cheese wholesalers.

482 Replies of CUW-C-I-8 and CUW-C-I-20 to question 67 I of the phase I questionnaire to specialised cheese wholesalers; reply of CUW-C-2-15 to question 26 of the phase II questionnaire to specialised cheese wholesalers.

483 Reply of CUR-CNL-2-8 to question 24 of the phase II questionnaire to retailers.

484 Reply of CUR-C-I-22 to question 59 of the phase I questionnaire to retailers.

485 Reply of CUR-CNL-2-9 to question 24 of the phase II questionnaire to retailers.

486 RBB submission retailers, p. 9 et seq.

487 RBB submission retailers, p. 7.

488 RBB submission retailers, p. 9 et seq.

489 Replies to question 16 of the phase I questionnaire to specialised cheese wholesalers.

490 Reply of CUR-CNL-2-8 to question 22 of the phase II questionnaire to retailers.

491 Replies to question 3 of the phase II questionnaire to retailers.

492 Replies to question 11 of the phase II questionnaire to retailers

493 Replies to question 36 of the phase I questionnaire to retailers and to question 12 of the phase II questionnaire to retailers.

494 Replies to question 16 of the phase II questionnaire to retailers. Replies to question 22 of the phase II questionnaire to competitors.

495 Submission of the notifying parties of 7 November 2008, p. 16. Replies to question 13, 14 and 15 of the phase I questionnaire to retailers.

496 RBB submission retailers, p. 14.

497 Replies to question 37 of the phase I questionnaire to retailers.

498 Replies to question 16 of the phase II questionnaire to retailers; only two modern types of retail mentioned that "for nature there would be an effect but unclear which price increase would be necessary" and that switching might be possible in case of comparable quality.

499 Replies to question 52 of the phase I questionnaire to retailers. DOC was mentioned in two instances but DOC has explained that it sells its cheese only to wholesalers.

500 Replies to questions 11 and 12 of the phase II questionnaire to retailers.

501 Replies to question 8 of the phase II questionnaire to retailers.

502 Reply of CUR-C-I-12 to question 32 of the phase I questionnaire to retailers.

503 Reply of CUR-CNL-2-8 to question 17 of the phase II questionnaire to retailers.

504 Reply of CUR-CNL-2-8 to question 23 of the phase II questionnaire to retailers.

505 RBB submission wholesalers, p. 15.

506 Replies to question 12 of the phase II questionnaire to competitors outside the Netherlands.

507 Replies to question 16 of the phase II questionnaire to competitors outside the Netherlands.

508 Replies to question 13 of the phase II questionnaire to competitors outside the Netherlands.

509 Replies to question 15 of the phase II questionnaire to competitors outside the Netherlands; only one participant did not answer.

510 Reply of the notifying parties to the Statement of Objections, p. 53.

511 Submission RBB retailers, p. 15.

512 Minutes of telephone call with CUR-CNL-2-8 of 29 August 2008, confirmed on 16 September 2008, reply of CUR-CNL-2-8 to question 11 and 13 of the phase II questionnaire to retailers.

513 RBB submission retailers, p. 19.

516 Replies to question 23 of the phase I questionnaire to retailers.

517 Form CO Section 6.I.14.

518 According to Section 6.I.16 of the Form CO, the common age categories for Dutch-type cheeses are: mild (jong): not older than six weeks; mild mature (jong-belegen): between six and nine weeks; mature (belegen): between nine and fourteen weeks; extra mature (extra belegen): between fourteen and forty weeks; old (oud): between forty weeks and one year; overgrown (overjarig): one year and older. 519 Form CO Section 6.I.14.

520 Submission RBB retailers, p. 14, on the basis of the notifying parties' total sales to retailers of approximately [...]* tonnes. The notifying parties' sales of mild mature nature cheese account for approximately [20-30]*%.

521 Form CO Section 6.I.24. It is to be noted that in these forms of sales which constitute the clear majority of sales of Dutch-type cheese at the level of modern types of retail, consumers can clearly distinguish rindless from nature cheese. In this context and in line with the consumer preferences described above in section 8.2.2.3, the consumer perception is that nature cheese is a product of a better quality than rindless cheese. The effects of these forms of sales of nature cheese are therefore similar to those of a brand.

522 RBB submission retailers, p. 15.

523 Presentation of the notifying parties at the oral hearing, p. 19.

524 RBB submission retailers, p. 17 according to which "a significant part of sliced nature cheese is sold with the rind already removed".

525 Submission of the notifying parties of 11 September 2008, p. 23.

526 See section 8.4.1.3.

527 Form CO Section 6.I.15.

528 Submission of the notifying parties of 11 September 2008, p. 12 and 13.

529 This is a reasonable assumption on the basis of the general percentages of sales of nature cheese purchased by these wholesalers from the notifying parties.

530 Reply of the notifying parties to the Statement of Objections, p. 43.

531 RBB submission wholesalers, p. 17.

532 The notifying parties have submitted information on 24 September 2008 according to which during the shutdown part of the production of the Lutjewinkel plant was taken over by other production plants of Campina, and also DOC produced some cheese for Campina. It was Campina's sales policy to fulfil the outstanding contractual obligations during the shutdown period. Therefore, contracts with retailers in the Netherlands and abroad, except one abroad, were respected. Hence, the shutdown of Lutjewinkel had no effect at all on the cheese prices of modern types of retail.

533 Submission of the notifying parties of 11 September 2008, p. 13; email of the Dutch Dairy Board of 15 September 2008. As in the case of the market for the sale of Dutch-type cheese to modern types of retail (see section 8.4.2.1), the notifying parties mentioned Cono, Westland, Vergeer and Uniekaas as main competitors in addition to a category of "others" which comprises private label/unbranded sales.

534 Submission of the notifying parties of 16 September 2008, annex 5.1; email of the Dutch Dairy Board of 15 September 2008. As in the case of the market for the sale of Dutch-type cheese to modern types of retail (see section 8.4.2.1), the notifying parties mentioned Cono, Westland, Vergeer and Uniekaas as main competitors in addition to a category of "others" which comprises private label/unbranded sales.

535 Submission of the notifying parties of 16 September 2008, annex 5.1; email of the Dutch Dairy Board of 15 September 2008. As in the case of the market for the sale of Dutch-type cheese to modern types of retail (see section 8.4.2.1), the notifying parties mentioned Cono, Westland, Vergeer and Uniekaas as main competitors in addition to a category of "others" which comprises private label/unbranded sales.

536 Submission of the notifying parties of 16 September 2008, annex 5.1.

537 Submission of the notifying parties of 16 September 2008, annex 1.1; reply to question 3 of the phase II questionnaire to specialised wholesalers. It is to be noted that the notifying parties combined market share as regards sales of Maasdam cheese to modern types of retail is insignificant. See submission of the notifying parties of 11 September 2008, annex 22.1.

538 Submission of the notifying parties of 16 September 2008, annex 1.1; reply to question 3 of the phase II questionnaire to competitors.

539 Reply of the notifying parties to the Statement of Objections, in particular p. 15.

540 See Case No. COMP-M.3130 - Arla Foods/Express Dairies, Commission Decision of 10 June 2003.

541 See answers to phase I questionnaire for butter to customers, question 11.

542 See answers to phase I questionnaire for butter to competitors, question 14.

543 See answers to phase I questionnaire for butter to customers, questions 9-10.

544 See answers to phase I questionnaire for butter to customers, reply by CU-B-I-15 and CU-B-I-55.

545 See answers to phase I questionnaire for butter to customers, reply by CU-B-I-30.

546 See answers to phase I questionnaire for butter to customers, reply by CU-B-I-21, CU-B-I-44, CU-B-I-15, question 13.

547 See answers to phase I questionnaire for butter to competitors, question 15-16.

548 See replies by CO-B-I-1 and CO-B-I-21 to phase I questionnaire for butter to competitors.

549 See answers to phase II questionnaire for bulk butter to competitors, reply by CO-B-2-15 and CO-B-2-3.

550 See answers to phase I questionnaire for butter to competitors, question 24 and answers to phase I questionnaire to customers, question 20.

551 See reply to phase II questionnaire for bulk butter to competitors by CO-B-2-15, CO-B-2-17, CO-B-2-27 and CO-B-2-5.

552 See answers to phase II questionnaire for bulk butter to customers, question 8.

553 See answers to phase II questionnaire for bulk butter to competitors, question 14.

554 See answers to phase II questionnaire for bulk butter to customers, question 11.

555 See answers to phase II questionnaire for bulk butter to customers, question 13.

556 See reply to phase II questionnaire for bulk butter to customers by CUB-B-2-47.

557 See reply to phase II questionnaire for bulk butter to competitors by CO-B-2-15. This respondent argues that, in case production of butter becomes more profitable, they would be able to double the production. See also replies by CO-B-2-17, which could produce additional 10 000 tons at short notice and CO-B-2-1, which could produce additional 30 000 tons.

558 See reply to phase II questionnaire for bulk butter to customers by CUB-B-2-10.

559 See answers to phase II questionnaire for bulk butter to customers, question 10.

560 See answers to replies to phase I questionnaire for butter to customers, question 15.

561 See answers to phase I questionnaire for butter to competitors, questions 20 and 35-36.

562 See answers to phase I questionnaire for butter to customers, questions 16-18.

563 Commission decision of 8 August 2001, in Case No COMP-M.2399 - Friesland Coberco/Nutricia.

564 Commission decision of 28 September 2000, in Case No COMP-M.1990 - Unilever/Bestfoods, OJ C311, 31.10.2000, p.6.

565 See reply to phase II questionnaire for butter to competitors by CO-B-2-29.

566 See answers to phase I questionnaire for butter to customers, question 19. See, in particular, reply by CU-B-I-50 and CU-B-I-44. See answers to phase I questionnaire for butter to competitors, question 23.

567 See answers to phase I questionnaire for butter to customers, replies by CU-B-I-49, CU-B-I-38 and CU-B-I-44.

568 See answers to phase I questionnaire for butter to competitors, reply by CO-B-I-24 and CO-B-I-17.

569 See answers to phase I questionnaire for butter to competitors, reply by CO-B-I-16, CO-B-I-3 and CO-B-I-17.

570 See reply to phase II questionnaire for butter to customers by CUP-B-2-47.

571 See answers to phase I questionnaire for butter to customers, question 20.

572 See answers to phase I questionnaire for butter to customers, question 21.

573 See replies to phase I questionnaire for butter to customers by CU-B-I-21 and CU-B-I-50.

574 See answers to phase I questionnaire for butter to customers, question 23.

575 These figures also include the sales of the "Frau Antje" packet butter produced by Friesland Foods. In this respect, the notifying parties have explained that Campina is a licensee of the brand and retains the proceeds of such sales. Therefore, the notifying parties have attributed the relevant sales and turnover to Campina.

576 See answers to phase I questionnaire for butter to competitors, questions 51-52. 577 See reply to phase I questionnaire for butter to customers by CU-B-I-38, CU-B-I-49 and CU-B-I-41. See the reply to phase I questionnaire to competitors by CO-B-I-1, CO-B-I-3. More in general see replies to question 40.

578 See reply by CO-B-I-25.

579 See replies by CUP-B-2-39 and CUP-B-2-6 to phase II questionnaire for packet butter to customers. 580 See reply by CUP-B-2-42 to phase II questionnaire for packet butter to customers.

581 See replies by CUP-B-2-39, CUP-B-2-42, CUP-B-2-6 and CUP-B-2-10 to phase II questionnaire for packet butter to customers.

582 See answers to phase I questionnaire for butter to competitors, question 45. See answers to phase II questionnaire for packet butter to competitors, questions 14-15.

583 See replies by CUP-B-2-39, CUP-B-2-42 and CUP-B-2-10 to phase II questionnaire for packet butter to customers.

584 See minutes of conference calls with CO-B-2-25 and CO-B-2-29.

585 See Form CO Section 7.G.2.

586 See Form CO Section 7.G.3.

587 See Form CO Section 6.G.3.

588 See Form CO Section 6.G.3.

589 See reply to question 10 first phase questionnaire retailers value added yoghurt and quark and reply to question 9 first phase questionnaire competitors value added yoghurt and quark.

590 See reply to question 11 first phase questionnaire retailers value added yoghurt and quark.

591 See Form CO Section 6.G.

592 See reply to question 13 first phase questionnaire retailers value added yoghurt and quark.

593 See Form CO Section 6.G.10.

594 Since in the out of home segment private label products are not present, the discussion whether one has to distinguish private label from branded products does only apply to the retail segment.

595 First phase questionnaire value added yoghurt and quark to customers, question 16 and first phase questionnaire value added yoghurt and quark to competitors, question 13.

596 See submission of the notifying parties of 08 August 2008.

597 First phase questionnaire value added yoghurt and quark to customers, question 26 and first phase questionnaire value added yoghurt and quark to competitors, question 28.

598 See Form CO Section 7.G.18-20.

599 See Form CO Section 6.C.28 Footnote 17.

600 See Case No. COMP-M.2399-Friesland Coberco/Nutricia.

601 See for example minutes CUO-Y-2-7 or minutes CUO-Y-2-2.

602 See reply to question 10 in M5223173-1-20385846 of 4 September 2008.

603 See reply to first phase questionnaire customers OOH value added yoghurt and quark, question 19, and second phase questionnaire customers OOH value added yoghurt and quark, question 3.

604 See minutes CUO-Y-2-6 and minutes CUO-Y-2-7.

605 See minutes CUO-Y-2-2.

606 See minutes CUO-Y-2-6, CUO-Y-2-7 and CUO-Y-2-2 12/09/2008.

607 See minutes CO-BD-2-16, minutes CO-BD-2-20, similar minutes CUO-Y-2-6, CUO-Y-2-7 and CUO-Y-2-2 12/09/2008.

608 See reply to question 8 in M5223173/1/20385846 of 4 September 2008.

609 See second phase questionnaire competitors value added yoghurt and quark, question 4.

610 Calculations based on figures submitted by the notifying notifying parties in Annex D.3a and Annex D.3b in reply M5084549-1-20385846 of 08-08-2008.

611 See reply to question 9 in M5223173-1-20385846 of 4 September 2008.

612 See Form CO Section 6.G.19.

613 Because the assessment of the relevant geographic market does within each distribution channel apply as well to all potential submarkets (health, indulgence, branded, private label) no further separation is necessary.

614 First phase questionnaire value added yoghurt and quark to competitors, question 4 and first phase questionnaire value added yoghurt and quark to customers, question 18 and 19.

615 First phase questionnaire value added yoghurt and quark to competitors, questions 18 and 19.

616 First phase questionnaire value added yoghurt and quark to competitors, question 24 and first phase questionnaire value added yoghurt and quark to customers, question 22.

617 See first phase reply of CUR-Y-1-14. Similar CUR-Y-1-16 who stated that "in dit segment wordt naast de nationale leveranciers ook de leveranciers uit Duitsland en Belgie meegnomen in de sourcing".

618 First phase questionnaire value added yoghurt and quark to customers, question 26.

619 First phase questionnaire value added yoghurt and quark to customers, question 28.

620 See second phase questionnaire competitors value added yoghurt and quark, question 4.

621 See Form CO Section 8.E.13.

622 See first phase questionnaire OOH value added yoghurt and quark, question 26.

623 See minutes CUO-Y-2-6 and CUO-Y-2-7.

624 See minutes CUO-Y-2-2.

625 See for example reply CUR-Y-1-8 to first phase questionnaire; email CUR-Y-1-14 or agreed minutes CUR-Y-1-15.

626 First phase questionnaire value added yoghurt and quark to customers, questions 6 and 7.

627 First phase questionnaire value added yoghurt and quark to customers, question 28.

628 First phase questionnaire value added yoghurt and quark to competitors, questions 39 and 40.

629 See reply CO-Y-1-25.

630 See reply CO-Y-1-7.

631 The figures are based on Form CO Section 7.G.15 and 7.G.8. Since the market shares do not differ significantly between value added yoghurt and quark, in what follows only the overall market will be described. Moreover, the health segment in which Friesland Foods is not active in, represents only 5% of the overall market, thus, the market shares of the potential indulgence market are almost identical to the overall market comprising health and indulgence.

632 See minutes CUO-Y-2-6 and CUO-Y-2-7 and CUO-Y-2-2.

633 See minutes CUO-Y-2-6 and CUO-Y-2-7.

634 See Form CO Section 7.G.27.

635 See for example reply of CO-Y-2-6 where it is stated that "the most important aspect to enter this market is that one can offer a complete range of dairy products."

636 See CO-Y-2-7: "Geografische Entfernung zu einem Geschaeftsfeld, dass preisagressive ist (Kostennachteil Logistik) und damit zeitlich geringe Flexibilitaet zu Wettbewerbern, die im Markt agieren."

637 See section 10.2.

638 Case M.2399 - Friesland Coberco / Nutricia, decision of 8 August 2001.

639 Case M.3130 - Arla Foods / Express Dairies, decision of 10 June 2003.

640 First phase questionnaire to competitors fresh-flavoured dairy drinks, question 13. First phase questionnaire to customers fresh-flavoured dairy drinks, question 8.

641 Second phase questionnaire to customers fresh flavoured dairy drinks.

642 See Commission Decision of 27.7.2007 in Case No COMP/M.4688 Nestlé/Gerber, OJ C 203, 31.8.2007, p.1 and Commission Decision of 31.10.2007 in Case No COMP-M.4842 Danone/Numico, OJ C284, 27.11.2007, p.1.

643 Prices are weighted with respect to volume sales. Weighting with respect to value sales does not alter the results but leads to the "weight management" price series to fall below the "kind" an "regular" price series.

644 See reply to question 4 in M5129511-1-20385846-SC of 22-08-2008.

645 See Form CO Section 6.E.10.

646 See Form CO Section 6.E.10.

647 See Form CO Section 6.E.11.

648 See for example Case No. COMP-M.4533 - SCA-P&G.

649 Since in the out of home segment private label products are not present, the discussion whether one has to distinguish private label from branded products does only apply to the retail segment.

650 See reply to first phase questionnaire fresh flavoured dairy drinks to customers, question 15 and first phase questionnaire fresh flavoured dairy drinks to competitors, question 18.

651 See reply to first phase questionnaire fresh flavoured dairy drinks to customers, question 15.

652 See reply CUR-FD-I-1.

653 See Form CO Section 7.E.12.

654 See reply to first phase questionnaire fresh flavoured dairy drinks to customers, question 34 and 36 and first phase questionnaire fresh flavoured dairy drinks to competitors, question 38.

655 See Form CO Section 7.E.29.

656 All figures based on Form CO Section 7.E.12.

(657) Marketing and Advertisement figures based on Annex III - V in reply to the notifying parties M5129511-1-20385846-SC of 22-08-2008.

658 See Form CO Section 6.E.13 Footnote 10.

659 See Case No. COMP-M.2399-Friesland Coberco/Nutricia.

660 See Form CO Annex 6.E.10.

661 See Form CO Section 6.E.18.

662 See Form CO Section 6.E.19.

663 See Form CO Section 6.E.23.

664 See Form CO Section 7.E.10.

665 See Form CO Section 7.E.

666 See reply CO-FD-2-2.

667 See reply CUR-FD-I-1.

668 The figures are based on Form CO Section 7.E.18-22.

669 See reply of CUR-FD-I-1 and CUR-FD-2-3 670 RBB Economics, Case COMP-M.5046 Campina/Friesland Foods - Response to the Statement of Objections - Long Life Flavoured Dairy Drinks, 17 October 2008.

671 All modeling assumptions, the main results and conclusions of the Statement of Objections, summary of the notifying parties' replies to the Statement of Objections as well as details regarding the additional arguments and modeling are contained in Annex 1 and its attached Appendices.

672 CUR-FD-I-1 and CUR-FD-I-4.

673 See reply to question 10 in M5129511-1-20385846-SC of 22-08-2008.

674 See Memorandum Case M.5046 - Friesland Foods/Campina, Comments on the Article 6.1c decision of 17 July 2008, 12 August 2008.

675 See reply CUR-FD-I-5.

676 CUR-FD-I-4: "Nee omdat wij op dit moment alleen A-merken verkopen en deze aangeboden worden vanuit Nederland."

677 See replies of CO-FD-2-1, CO-FD-2-2 and CO-FD-2-3.

678 CUR-FD-I-1.

679 See Memorandum Case M.5046 - Friesland Foods/Campina, Comments on the Article 6.1c decision of 17 July 2008, 12 August 2008, page 2.

680 See reply CO-FD-2-1.

681 See reply CUR-FD-I-1.

682 Form CO, Annex 6.F.4.

683 Form CO, Annex 6.F.5 " Friesland Foods : Identiteit zuivel, drinkyoghurt en chocoladedranken".

684 "Memorandum case COMP M.5046 Friesland Foods/Campina comments on the 6(1)c decision", dated 12 August 2008, page 22.

685 Questionnaire to customers long-life flavoured dairy drinks sent on 19 June 2008, question 11.

686 Form CO, Annex 6.F.8.

687 See for example reply from CUR-LLFD-I-2 to 1st phase questionnaire to customers long-life flavoured dairy drinks sent on 19 June 2008, question 11 a) "Given the particular difference in taste between chocolate and, for example, fruit, it is very doubtful whether consumers of other flavoured dairy drinks would switch to chocolate drinks or vice versa should the price of those other drinks increase permanently by 5-10%".

688 Form CO, Section 6.F, paragraphs 16 and 17.

689 Annex 4-2 to the reply to the Statement of Objections.

690 RBB Economics, Case COMP-M.5046 Campina/Friesland Foods - Response to the Statement of Objections - Long Life Flavoured Dairy Drinks, 17 October 2008.

691 All modeling assumptions, the main results and conclusions of the Statement of Objections, summary of the notifying parties' replies to the Statement of Objections as well as details regarding the additional arguments and modeling are contained in Annex 1 and its attached Appendices.

692 See Form CO Section 6.L, paragraph 31.

693 Form CO, Section 6.F, paragraphs 37 and 41.

694 See Memorandum "Case M.5046 - Friesland Foods/Campina, Comments on the Article 6.1c decision of 17 July 2008, RE Distinction between branded and PL products", 28 July 2008.

695 See Memorandum "Case M.5046 - Friesland Foods/Campina, Comments on the Article 6.1c decision of 17 July 2008, RE Distinction between branded and PL products", 28 July 2008.

696 Reply to the Statement of Objections, pages 89 to 91.

697 Contrary to IRI GFK data includes discounters (Aldi and Lidl)

698 "Exclusive buyers" are buyers who have not purchased LLDDs from other suppliers. See reply to the Statement of Objections, annex 4-1 "Response to the statement of objections - long-life flavoured dairy drinks", page 9.

699 Reply to the Statement of Objections, paragraph 251.

700 Friesland Foods sells a negligible volume of fruit-flavoured dairy drinks under PL in the Netherlands (EUR 400 000).

701 Questionnaire to customers long-life flavoured dairy drinks sent on 19 June 2008, question 14.

702 Reply to the Statement of Objections, paragraph 249.

703 See reply dated 25 August 2008 from the notifying parties to Article 11 request of the Commission with regard to long-life dairy drinks, Question 5, annexes V and VI.

704 Reply to the Statement of Objections, paragraph 250.

705 Form CO, Section 7.F, tables 10 et 11. These figures are for Retail + Out of home.

706 Form CO, Section 7.F, tables 28 and 29. These figures are for Retail + Out of home.

707 Form CO, Section 6F, paragraph 34. "Already in 1997 there was a significant volume (45.3%) and value (29.9%) of PL on the market for chocolate flavoured long-life dairy drinks in Belgium and this has not significantly changed during the last ten years". These figures are based on Nielsen data (see form CO, Annex 6.F.12) for the retail segment.

708 Reply to the Statement of Objections, annex 4-1, page 10.

709 Form CO, Section 7.F, paragraph 6. "As for the Netherlands, IRI Infoscan data has been used for the retail segment. However, the data of IRI Infoscan do not include all Dutch retailers (Aldi, Lidl and Koopconsult are not included). In order to make a best effort estimate of the size of the market including Aldi, Lidl and Koopconsult, the notifying parties have used GFK data. In Belgium and Germany, the notifying parties have made use of market data of Nielsen".

710 Questionnaire to competitors long-life flavoured dairy drinks sent on 19 June 2008, question 13.

711 See Form CO Section 6.F, paragraphs 45 and 46.

712 Reply to the Statement of Objections, paragraph 258.

713 Case M.3130-Arla Foods/Express Dairies. Long-life flavoured milk corresponds to what is here called long-life flavoured dairy drinks or LLDDs

714 Respectively [90-100]*% in the OOH segment with an overlap of [0-5]*% and [90-100]*% in the retail segment with an overlap of [0-5]*%.

715 Respectively [90-100]*% in the OOH segment with an overlap of [5-10]*% (Campina) and [90-100]*% in the retail segment with an overlap of [40-50]*% (Friesland Foods).

716 Questionnaire to competitors long-life flavoured dairy drinks sent on 19 June 2008, questions 35 and 36. Questionnaire to customers long-life flavoured dairy drinks sent on 19 June 2008, question 31 and 32.

717 Reply from CU-LLFD-2-1 to 2nd phase questionnaire on long-life flavoured dairy drinks sent to customers on 19 August 2008, Question 8.

718 Reply from CUO-LLFD-I-11 to 2nd phase questionnaire on long-life flavoured dairy drinks sent on customers on 19 August 2008, Question 9.

719 Answer from CU-LLFD-2-2 to 2nd phase questionnaire on long-life flavoured dairy drinks sent on customers on 19 August 2008, Question 8.

720 Reply from CU-LLFD-2-2 to 2nd phase questionnaire on long-life flavoured dairy drinks sent on customers on 19 August 2008, Question 8.

721 RBB Economics, Case COMP/M.5046 Campina/Friesland Foods - Response to the Statement of Objections - Long Life Flavoured Dairy Drinks, 17 October 2008.

722 All modeling assumptions, the main results and conclusions of the Statement of Objections, summary of the notifying parties' replies to the Statement of Objections as well as details regarding the additional arguments and modeling are contained in Annex 1 and its attached Appendices.

723 Reply to the Statement of Objections, paragraph 264.

724 "Memorandum case COMP M.5046 Friesland Foods/Campina comments on the 6(1)c decision", dated 12 August 2008, page 23.

725 Reply to the Statement of Objections, paragraph 272.

726 Generally A-brands are priced substantially above PL, are supported through substantial brand-related marketingefforts, enjoy a significant geographic spread and relatively high market shares.

727 Reply from CU-LLFD-2-2 to 2nd phase questionnaire sent on customers on 19 August 2008, Question 9.

728 2nd phase questionnaire sent to customers on 19 August 2008, Question 8.

729 Reply to the Statement of Objections, paragraph 269 and 274.

730 Answer from CU-LLFD-2-2 to 2nd phase questionnaire sent on customers on 19 August 2008, question 11 "Some A-brand products ("must have- brands", such as Chocomel by FF) are particularly strong. These brands are almost impossible to constrain by offering private label alternatives". CU-LLFD-II-1 underlined that "For Fristi and Yogho Yogho it will not be able to switch : it has very loyal customers"

731 Form CO, Section 6.F.13, paragraph 35.

732 See e-mail from CO-LLFD-I-4 of 18 September 2008, 12:07 (non-confidential).

733 Questionnaire to competitors long-life flavoured dairy drinks sent on 19 June 2008, questions 35 and 36. Questionnaire to customers long-life flavoured dairy drinks sent on 19 June 2008, question 31 and 32.

734 Minutes of a teleconference with CUR-LLFD-2-4 of 15 September 2008. Minutes of a teleconference with CUR-LLFD-2-5 on 16 September 2008

735 Minutes of a teleconference with CUR-LLFD-2-4 of 15 September 2008. Minutes of a teleconference with CUR-LLFD-2-5 on 16 September 2008.

736 Minutes of a teleconference with CUR-LLFD-2-5 on 16 September 2008.

737 Reply to the Statement of Objections, paragraph 269 and 274.

738 Minutes of a teleconference with CUR-LLFD-2-5 on 16 September 2008.

739 Form CO, Section 6.F.13, paragraph 35.

740 Form CO, Section 7.F, table 59.

741 Form CO, Section 7.F, table 53.

742 Reply to the Statement of Objections, paragraph 259.

743 Reply to the Statement of Objections, annex 4-1, page 11.

744 See http://www.frieslandfoods.com/nl/frieslandfoods/PressReleases/Pages/pers.aspx?guid=17f388eb-890f-4e79- a4c1-c3cf6fbff410&iid=113

745 See reply from CO-LLFD-I-4 by e-mail dated 18 September 2008, 15:50 : "Eine Direktvermarktung in den gefragten Ländern kam für uns bisher nicht in Frage" or reply from CO-LLFD-I-2 by e-mail dated 22 September 2008, 11:31 : "(...) erwägt keinen Markteintritt in den Niederlanden und Belgien mit langlebigen Molkereigetränke-Marken"

746 See Form CO Section 6.H.1.

747 See Form CO Section 6.H.8-10.

748 See Form CO Section 6.H.7.

749 See Form CO Section 6.H.11.

750 First phase questionnaire fresh dairy desserts to customers, question 9 and first phase questionnaire fresh dairy desserts to competitors, question 13.

751 See reply CUR-D-2-7. 752 See reply CUR-D-2-12.

753 Original citation:" Wij zien Vla meer als een basis product dat niet direct concurreert met de de afzet van het luxere dessert segment".

754 Second phase questionnaire fresh dairy desserts to retailers and OOH wholesalers.

755 First phase questionnaire fresh dairy desserts to customers, question 9 and first phase questionnaire fresh dairy desserts to competitors, question 14.

756 First phase questionnaire fresh dairy desserts to customers, question 11 and first phase questionnaire fresh dairy desserts to competitors, question 12.

757 CUR-D-I-7 for example stated that "Although porridge is consumed at breakfast, as a snack and at dessert, (...) considers porridge to be mostly related to the fresh dairy dessert market."

758 See Form CO Section 6.H.11.

759 See Form CO Section 6.H.11.

760 See notifying parties' reply to the Article 6.(1)(c) Decision "Memorandum Case M.5046 - Friesland Foods/Campina, Comments on the Article 6.1c decision of 17 July 2008", 12 August 2008.

761 See for example Case No. COMP-M.4533 - SCA/P&G.

762 Since in the out of home segment private label products are not present, the discussion whether one has to distinguish private label from branded products does only apply to the retail segment.

763 See reply to first phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to customers, and first phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to competitors, question 22.

764 See Form CO Section 7.C.29. The figures refer to the overall market including sales by retailers and OOH wholesalers. Private label is only present in the retail segment, its share there would be 49.3% based on then notifying parties estimates.

765 Results are similar in terms of sales value.

766 According to IRI scanner data the share of private label in 2007 has been 30%, below the figure of 49% submitted by the notifying parties. However, since IRI data do not account for the deep discounter Aldi, Lidl and Koopconsult, they underestimate the share of private label significantly.

767 In that respect the present case is different from M.4533 where the private label supply was supplied by a completely different set of producers compared to the branded market, thereby leading to asymmetric conditions of competitions - while the branded producers could constrain the private label producers, the opposite was not possible as a private label producer would have to invest into a brand and marketing before being able to compete. The equally strong presence of Campina as well as Friesland Foods in private labels and branded products ensures symmetry.

768 See reply to Questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to customers, sent on 18/06/2008, question 9 and 25.

769 See for example reply CUR-D-2-7.

770 First phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to competitors, question 37 and first phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to customers, questions 44 and 45.

771 For the exact market share data see the section 12.3.

772 See Form CO Section 6.C.28 Footnote 17.

773 See Case No. COMP/M.2399-Friesland Coberco/Nutricia.

774 Since this section repeats essentially structural characteristics of the OOH market, which have already been explained in previous sections in more detail (see for more details 7.2.1.3), the references are not repeated again.

775 See reply to question 10 in M5223173-1-20385846 of 4 September 2008.

776 See Form CO Section 6.H.23.

777 See Form CO Section 6.H.23.

778 See Form CO Section 6.H.8.

779 See reply to question 7, Questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to competitors, sent on 18/06/2008.

780 See reply CO-D-2-1.

781 See reply CO-D-2-3: "verfuegt ueber keine Abfuellanlagen, die fuer Vla geeignet waeren. Vla passt nicht in das [...] Konzept".

782 See reply to first phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to customers, question 44.

783 See reply to first phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to competitors, question 7.

784 See for example reply CUR-D-2-7 where it is stated that "there are no other suppliers for porridge".

785 See Form CO Section 6.H.10.

786 The figures are based on Form CO Section 7.H.5-7.

787 See Form CO Section 7.

788 See Form CO Section 7.

789 See reply to first phase questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to competitors, question 7.

790 See for example CU-BD-1-1 who stated that "Pre merger, [...] has always been able to negotiate with either the merging parties knowing that it could go to the other if unfavourable terms were offered. Post merger, [...] will no longer be in a position to play off one cooperative society against the other and will likely be confronted with significant price increases." 791 A view shared in general by the majority of respondents to the market investigation, which considered Friesland Foods and Campina as the closest competitors with respect to product range, price, ability to supply large quantities and brands - Questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to customers, sent on 18-06-2008, question 56 and 57 and Questionnaire fresh milk, fresh buttermilk, plain yoghurt and custard to competitors, sent on 18-06-2008, question 53 and 54.

792 RBB Economics, Case COMP-M.5046 Campina/Friesland Foods - Response to the Statement of Objections _Long Life Flavoured Dairy Drinks, 17 October 2008.

793 All modeling assumptions, the main results and conclusions of the Statement of Objections, summary of the notifying parties' replies to the Staement of Objections as well as details regarding the additional arguments and modeling are contained in Annex 1 and its attached Appendices.

794 See reply to question 15 in M5155217/1/20385846 of 25/08/2008.

795 See replyCU-BD-1-1.

796 See reply CUR-D-2-13.

797 See reply CUR-D-2-12.

798 See reply CU-BD-1-1.

799 See Form CO Section 7.H.22.

800 See reply of CO-D-2-15.

801 See reply of CO-D-2-5." beschaeftigt sich prinzipiell nicht mit diesem Produktangebot".

802 See Form CO Section 8.H.3-4.

803 See Form CO Section 6.H.10.

804 See reply to question 36c in first phase questionnaire retailers dessert.

805 See reply CUR-D-1-7.

806 See answers to phase I questionnaire for cream to retailers, question 9, phase I questionnaire to competitors, question 11.

807 See answers to phase I questionnaire for liquid cream to retailers, to OOH customers and to industrial customers, question 7.

808 See answers to phase II questionnaire for liquid cream to retailers, question 5, phase II questionnaire to OOH customers, question 3.

809 See answers to phase I questionnaire for liquid cream to competitors, question 10.

810 See answers to phase II questionnaire for liquid cream to competitors, question 3.

811 See reply to phase I questionnaire for cream to competitorsby CO-Cr-I-34.

812 See answers to phase I questionnaire for cream to retailers, question 6 and 8.

813 See answers to phase II questionnaire for liquid cream to competitors, questions 6-7.

814 See the notifying parties' letter of 25 August 2008. in reply to a Commission Article 11 request. The notifying parties explained that, when long-life cream was introduced in the Netherlands, the current technology available was not such to ensure a good taste to long-life cream compared to fresh cream. Therefore, Dutch consumers developed the conviction that fresh cream has better taste and quality than long-life cream.

815 See answers to phase II questionnaire for liquid cream to customers, question 4-5.

816 See CUR-Cr-2-7, CUR-Cr-2-24, CUR-Cr-2-2 question 7; CUR-Cr-2-11 question 8.

817 See CUR-Cr-2-2 and CUR-Cr-2-11.

818 See CUO-Cr-2-4, CUO-Cr-2-9.

819 See CO-Cr-2-10, question 8.

820 See CO-Cr-2-32, CO-Cr-2-31, CO-Cr-2-20 question 8.

821 See the notifying parties' electronic submission of 26 September 2008, in reply to the Commission request for information.

822 See answers to phase I questionnaire for cream to retailers, question 23. See answers to phase I questionnaire for cream to competitors, questions 33-34.

823 See answers to phase I questionnaire for cream to retailers, questions 10-12 and answers to phase I questionnaire for cream to competitors, questions 12-14.

824 See answers to phase II questionnaire for cream to retailers, question 20.

825 See answers to phase I questionnaire for liquid cream to competitors, question 15.

826 See answers to phase I questionnaire for cream to retailers, question 14 b).

827 See answers to phase I questionnaire for liquid cream to industrial customers, questions 10.

828 See reply of CO-Cr-2-27, CO-Cr-2-10 to phase II questionnaire for liquid cream to competitors.

829 See answers to phase I questionnaire for cream to retailers, question 15.

830 See reply of CO-Cr-2-6 and CO-Cr-2-13 to phase II questionnaire for liquid cream to competitors.

831 See answers to phase I questionnaire for liquid cream to industrial customers, questions 10 and 12.

832 See replies to phase I questionnaire for liquid cream to retail customers, question 16.

833 See replies to phase I questionnaire for liquid cream to retail customers, questions 13 b), 15.

834 See answers to phase II questionnaire for liquid cream to industrial customers, question 8.

835 Figures based on the notifying parties' electronic submission of 26.9.2008, in reply to the Commission Article 11 request.

836 Figures based on the notifying parties' electronic submission of 26.9.2008, in reply to the Commission Article 11 request.

837 Figures based on the notifying parties' electronic submission of 26.9.2008, in reply to the Commission Article 11 request.

838 Figures based on the notifying parties' electronic submission of 26.9.2008, in reply to the Commission Article 11 request.

839 Figures based on the notifying parties' electronic submission of 16.9.2008, in reply to the Commission Article 11 request.

840 See the notifying parties' letter of 25 August 2008, in reply to an Article 11 letter from the Commission.

841 See CUI-Cr-I-26, CUI-Cr-I-16.

842 See CUI-Cr-I-26, CUI-Cr-I-24.

843 CUI-Cr-I-26 argues that they enter into contracts lasting only 1 month, but that 40% of their liquid cream supplì is negotiated weekly. See also CUI-Cr-I-16, CUI-Cr-I-24.

844 See answer to phase II questionnaire for cream to industrial customers by CUI-Cr-2-2.

845 See answer to phase II questionnaire for cream to industrial customers by CUI-Cr-2-1.

846 See CO-Cr-I-19, CO-Cr-2-45.

847 See phase II questionnaire for liquid cream to competitors, questions 9 and 26. See, in particular, replies by COCr-2-10 which has excess capacity in the formats exceeding 1Kg, CO-Cr-2-42, CO-Cr-2-25, CO-Cr-2-35 and CO-Cr-I-7.

848 See answers to phase II questionnaire for cream to retailers, question 23.

849 See answers to phase II questionnaire for cream to OOH customers, question 16.

850 See answers to phase II questionnaire for cream to competitors, question 3.

851 See minutes of conference call with CO-Cr-2-47.

852 See answers to phase I questionnaire for cream to retailers, questions 10-12; answers to phase I questionnaire for cream to competitors, questions 12-14.

853 See statement by CO-Cr-2-39 that "With a PL operation the [retail] customer can create its own image to the public". CUR-Cr-2-23 noted in its reply to the market investigation that shelf space for brands is decreasing. See also replies by CO-Cr-2-3 and CO-Cr-2-20, question 5. In Germany, in the retail segment, private labels cover approximately 72% of the market in volume, in the market including the Netherlands, Belgium and Germany the share of private labels is approximately 64% in volume.

854 See CUR-Cr-2-21.

855 See minutes of the conference call with CO-Cr-2-47.

856 See answers to phase I questionnaire for cream to competitors, question 15.

857 See answers to phase I questionnaire for cream to retailers, questions 17-21; see reply to phase I questionnaire for cream to competitors by CO-Cr-I-7 and CO-Cr-I-34.

858 See answers to phase II questionnaire for cream to retailers, question 28; phase II questionnaire for cream to OOH customers, question 19.

859 See reply to phase I questionnaire for cream to competitors by CO-CR-I-30 and CO-Cr-I-34.

860 See reply to phase II questionnaire for cream to competitors by CO-Cr-2-47 and CO-Cr-I-7.

861 See answers to phase I questionnaire for cream to retailers, questions 13-14.

862 See minutes of conference call with CO-Cr-2-47.

863 See answers to phase II questionnaire for spray cream to OOH customers, questions 17-19.

864 See replies by CO-Cr-I-30, CO-Cr-I-7 and CO-Cr-I-34 to phase I questionnaire for cream to competitors, question 16.

865 See replies to phase II questionnaire for spray cream to competitors by CO-Cr-2-44 and CO-Cr-I-7.

866 See reply to phase II questionnaire for spray cream to competitors by CO-Cr-2-47.

867 See answers to phase I questionnaire for cream to retail customers, questions 34-35; answers to phase II questionnaire for cream to retail customers, question 37.

868 See answers to phase II questionnaire for spray cream to retail customers, questions 31-32.

869 See answers to phase II questionnaire for spray cream to OOH customers, questions 17-19.

870 See answers to phase II questionnaire for spray cream to OOH customers, questions 22-25.

871 Questionnaire customers liquid coffee whiteners sent on 19 and 23 June 2008, question 8 and questionnaire competitors liquid coffee whiteners sent on 19 June 2008, question 9.

872 First phase questionnaire customers liquid coffee whiteners sent on 19 and 23 June 2008, question 7 and questionnaire competitors liquid coffee whiteners sent on 19 June 2008, question 8. In particular, see reply from CUR-CW-I-12 "To consumers, the use of either liquid coffee whiteners vs powder is the same as the difference between black and white".

873 Form CO, Section 6.L, paragraphs 10 to 13.

874 Form CO, Section 6.L, paragraph 15.

875 Form CO, Section 6.L, paragraphs 16 and 17.

876 Form CO, Section 6.L, paragraph 37.

877 Form CO, Section 6.L, paragraph 17. See also Campina's internal document "Strategic plan - liquid coffee whiteners 2007-2009" available in Annex 6.L.6.

878 Decision M. 2399 Friesland Coberco/Nutricia of 8 August 2001

879 See reply from CO-CW-2-4 to 2nd phase questionnaire to competitors liquid coffee whiteners sent on 13 August 2008, question 4.

880 See reply from CO-CW-2-15 to 2nd phase questionnaire to competitors liquid coffee whiteners sent on 13 August 2008, question 4.

881 See reply from CUR-CW-2-3 to 2nd phase questionnaire to customers retailers liquid coffee whiteners sent on 13 August 2008, question 2.

882 See reply from CO-CW-2-7 to 2nd questionnaire to competitors liquid coffee whiteners sent on 13 August 2008, question 7.

883 Form CO, Section 6.L, paragraph 17.

884 An example is the Friesche Vlag (Friesland Foods) range of coffee milk in the Netherlands, these products being offered in three varieties, each having a different fat content (0 gr., 4.3 gr. and 8.6 gr. of fat per 100 ml of coffee milk). See "Memorandum - comments on the Article 6(1)(c) of 17 July 2008", dated 12 August 2008".

885 See for example Case No. COMP/M.4533 - SCA/P&G

886 See Form CO Section 6.L, paragraph 40.

887 Form CO, Section 6.L, paragraph 41.

888 Ibidem.

889 The assessment in this section with regard to a potential distinction between branded and PL products is valid for coffee milk and coffee cream (hereinafter "liquid coffee whiteners").

890 Case No. COMP/M.4533 - SCA/P&G

891 See questionnaire customers liquid coffee whiteners sent on 19 and 23 June 2008, question 10 and questionnaire competitors liquid coffee whiteners sent on 19 June 2008, question 12.

892 See Form CO Annex 7.L.1. These figures are somewhat lower in value terms: 42.2% in the Netherlands (+1.8%), 28.6% in Belgium (+2.1%), 43.9% in Germany (+0.9%).

893 See Form CO, Annex 7.L.1.

894 See questionnaire customers liquid coffee whiteners sent on 19 and 23 June 2008, question 12 and questionnaire competitors liquid coffee whiteners sent on 19 June 2008, question 14.

895 See reply from CUR-CW-I-3 to 1st phase questionnaire customers liquid coffee whiteners sent on 19 and 23 June 2008, question 27.

896 The notifying parties also argue that contrary to the retail channel final consumers in the OOH channel do not buy coffee and LCWs separately. LCWs are provided by the restaurants and cafes as a complementary service and the price of a LCW forms only a small share of the overall coffee price. However, this argument is valid at the downstream level but not at the upstream level where both notifying parties operate and the merger is likely to have an impact.

897 Form CO, Section 6.L, paragraph 44. 898 See for an example of these promotional campaigns targeted at the OOH channel Form CO, Annex 6.L.11.

899 Form CO, Section 6.L, paragraphs 45 to 56.

900 Decision M2339 Friesland Coberco/Nutricia of 8 August 2001

901 See Memorandum "Comments on the Article 6(1)(c) decision of 17 July 2008" dated 12 August 2008 pp 56-57.

902 See questionnaire to competitors liquid coffee whiteners sent on 13 August 2008, question 3.

903 Ibidem

904 The assessment in the Friesland Coberco/Nutricia decision focused on the branded segments of the downstream markets.

905 See questionnaire customers liquid coffee whiteners sent on 19 and 23 June 2008, questions 16 to 20 and questionnaire competitors liquid coffee whiteners sent on 19 June 2008, question 19 to 23.

906 See 2nd phase questionnaire to competitors liquid coffee whiteners sent on 13 August 2008, question 13. CURCW- 2-3 explained in that regard that "Although the packaging of the products are said to be typically Dutch, next to the brand names Friesche Vlag and Campina being typically Dutch, it is not perceived as explicitly referring to the Dutch origin of the product".

907 See Form CO Annex 7.L.1.

908 No specific distinction was made in the 6(1)(c) decision between coffee milk and coffee cream.

909 See reply from CO-CW-2-15 to 2nd phase questionnaire competitors liquid coffee whiteners sent on 13 August 2008 question 13 "In the Netherlands, we consider private label products to be the closest competitor to Friesland Foods then Becel (Unilver) followed by Campina. In Belgium we consider private label products to be the closest competitor to Friesland Foods to Friesland Foods followed by Becel and Nestlé".

910 See reply from CUR-CWA-2-1 to 2nd phase questionnaire customers liquid coffee whiteners sent on 13 August 2008, question 20.

911 See reply from CUR-CW-I-3 to 1st phase questionnaire customers liquid coffee whiteners sent on 19 June 2008, question 29.

912 Form CO, Section 6.L paragraphs 34 and 35 and Section 8.L, paragraph 10.

913 The notifying parties have provided a graph showing that the consumption of liquid coffee whiteners (coffee milk and coffee cream) in the Netherlands has plummeted from 21 kg per person per year to 6 kg between 1980 and 2004.

914 See reply from CO-CW-2-7 to 2nd phase questionnaire competitors liquid coffee whiteners sent on 13 August 2008, question 10.

915 See replies from inter alia CO-CW-2-3 and CO-CW-2-7 to 2nd phase questionnaire competitors liquid coffee whiteners sent on 13 August 2008, question 9.

916 See Form CO Annex 7.L.1.

917 There is no private label coffee milk in the OOH segment in Germany.

918 See reply from CUO-CW-2-5 to 2nd phase questionnaire OOH wholesalers liquid coffee whiteners sent on 13 August 2008, questions 7 and 5.

919 See reply from CUO-CW-2-12 to 2nd phase questionnaire OOH wholesalers liquid coffee whiteners sent on 13 August 2008, question 4.

920 See reply from CUO-CW-2-5 to 2nd phase questionnaire OOH wholesalers liquid coffee whiteners sent on 13 August 2008, question 13.

921 See reply from CUO-CW-2-5 to 2nd phase questionnaire OOH wholesalers liquid coffee whiteners sent on 13 August 2008, questions 8, 9, 10 and 12.

922 See reply from CUO-CW-2-8 to 2nd phase questionnaire OOH wholesalers liquid coffee whiteners sent on 13 August 2008, questions 7 and 10.

923 Form CO, Annex 7.L.1.

924 The notifying parties have identified only five segments since they consider that toppings and batter stabilizers belong to the same product market

925 Campina acquired Satro in 2007.

926 Form CO, Section 6.M, paragraphs 20 to 24.

927 Form CO, Section 6.M, paragraphs 6 to 12.

929 See reply from CU-SDE-I-30 to 1st phase questionnaire to customers SDE sent on 17 June 2008, Question 5.

930 Except, to a limited extent, between fat concentrates and creamers.

931 Form CO, Section 6.N, paragraph 37.

932 Questionnaire to customers SDEs sent on 17 June 2008, question 7.

933 See reply of CO-SDE-I-5 to 1st phase questionnaire sent to competitors SDEs on 17 June 2008, question 8.

934 See reply of CO-SDE-I-56to 1st phase questionnaire sent to competitors SDEs on 17 June 2008, question 7.

935 Questionnaire to competitors SDE sent on 17 June 2008, question 8.

936 Form CO, Section 6.M, paragraph 40.

937 See reply from CU-SDE-2-27 to 2nd phase questionnaire to competitors SDE sent on 6 August 2008, Q.11 "The approval process would not be viable in terms of costs and resources (it is already very complex to manage in the EU). Import tariffs, shipping costs, use of a special palm oil (would have to be exported to EU from supplier) plus approval process would prove to be prohibitive". See reply from CU-SDE-2-26 to the same questionnaire "Import/customs duties, transport and/or quality restraints qualified sourcing overseas not feasible".

938 See reply from CU-SDE-2-27 to 2nd phase questionnaire to competitors SDE sent on 6 August 2008, Q.12 "We are not aware of any quality suppliers in Asia. It would require a lot of efforts for such a complex and sensitive ingredient to be sourced from Asia. The savoury creamer suppliers are considered as strategic suppliers, which we need to have at close proximity to establish the required close relationship".

939 See reply from CU-SDE-2-29 to 2nd phase questionnaire to competitors SDE sent on 6 August 2008, Q.13 "We have noticed a big difference in quality. Furthermore their service level differs regarding ability to get customized recipes. In addition the long (and costly) transport times require stock building (expensive and notpossible for us due to absence of physical space) and additional complexity in terms of planning (risk of less service to end-customers regarding delivery times)"

940 Source: Notifying parties' estimates broadly confirmed by the market investigation.

941 Answer from the notifying parties to the Commission's request for information dated 14 August 2008, Question 1, annex 1.

942 Answer from the notifying parties to the Commission's request for information dated 14 August 2008, page 11. See also Campina's internal documents: "Meeting management food systems" dated 30 January 2007, available in Annex VIII and "Campina : Update GIP SP 2007-2009" dated May 2007 available in annex IX.

943 Form CO, Section 7.M, paragraph 65.

944 This has been confirmed by an e-mail from the notifying parties to the Commission dated 11 September 2008, 17:26.

945 See reply to 2nd phase questionnaire to customers SDE sent on 6 August 2008, Q.9.

946 Answer from the notifying parties to Article 11 to the Commission's request for information, dated 14 August 2008, page 4. See also See also Campina's internal documents: "Meeting management food systems" dated 30 January 2007, available in Annex VIII and "Campina: Update GIP SP 2007-2009" dated May 2007 available in annex IX.

947 See reply to 2nd phase questionnaire to customers SDE sent on 6 August 2008, Q.16.

948 See reply to 2nd phase questionnaire to customers SDE sent on 6 August 2008, Q.12.

949 Source: Notifying parties' estimates broadly confirmed by the market investigation

950 Answer from the notifying parties to Article 11 to the Commission's request for information, dated 14 August 2008, page 11. See also Campina's internal document: "Supplying the ingredients of success to key players worldwide in food, nutrition and pharma" available in Annex VII.

951 See Campina's internal documents "Supplying the ingredients of success to key players worldwide in food, nutrition and pharma" page 11: "Due to limited focus on this product group in the past years, we don't have clear insights in the customer needs" and page 13 "The market is showing attractive growth (10%) but also serious competition. The fact that our portfolio consists of only one product that is not even meeting customers' needs creates opportunities for the competition"

952 Source: Notifying parties' estimates sent in an answer to the Commission's request for information, dated 11 July 2008, page 7.

953 Commission Decision of 2.6.2006 in Case No COMP/M.4207 Campina/Fonterra Cooperative Group JV, OJ C194, 18.8.2006, p.40. 954 Questionnaire to Customers pharma lactose, 17 June 2008 questions 7 and 8.

955 Reply of CU-PL-I-35.

956 The notifying parties have provided more examples of alleged switching form pharmaceutical grade to food grade lactose, notably in South East Asia and India.

957 Reply of CO-PL-I-6.

958 Replies CU-PL-I-47, CU-PL-I-9, CU-PL-I-41, CU-PL-I-43, CU-PL-I-48, CU-PL-I-34, CU-PL-I-32, CU-PL-I-33, CU-PL-I-2 Questions 7, 8 and 9.

959 Annex 34 Notifying parties' Response to article 11 5 September 2008.

960 Response of the notifying parties to the 6(1)(c) Decision.

961 Cf. paragraphs 20 - 24 of the Commission Notice on the definition of Relevant Markets for the purpose of Community Competition Law, 97-C 372-03.

962 Cf. paragraph 22 of the Commission Notice on the definition of Relevant Markets for the purpose of Community Competition Law, 97-C 372-03.

963 Replies CO-L-2-1, CO-L-2-1, CO-L-2-3, CO-L-2-6, CO-L-2-8 and CO-L-2-10.

964 Reply CO-L-2-8.

965 Reply CO-PL-I-10.

966 Reply CO-L-2-1.

967 Reply CO-PL-I-6.

968 Reply CO-L-2-8.

969 Cf. paragraphs 20 - 24 of the Commission Notice on the definition of Relevant Markets for the purpose of Community Competition Law, 97-C 372-03.

970 Form CO, Section 6.N.5/11

971 Form CO, Section 6.N.5/11

972 Form CO Section 6.N.6/11

973 Form CO Section 6.O.17/25.

974 Form CO Section 6.O.17/25.

975 Response of the notifying parties to the 6(1)(c) Decision.

976 Response of the notifying parties to the 6(1)(c) Decision.

977 Commission Notice on the definition of the relevant market for the purposes of Community competition law, [1997] OJ C372/5, [1998] 4 CMLR 177, para.13.

978 Commission Notice on the definition of the relevant market for the purposes of Community competition law, [1997] OJ C372/5, [1998] 4 CMLR 177, para.13.

979 Case COMP M. 4207 CAMPINA / FONTERRA CO-OPERATIVE GROUP / JV.

980 On 8 September 2008 the notifying parties have submitted two opinions that discuss the possibilities to exchange the use of different excipients with filler binder properties. It is considered that ad hoc documents for the purpose of merger control proceedings should be scrutinized with utmost care. Unlike pre-existing studies they have not been prepared in the normal course of business for the adoption of business decisions. This view is similar to the Relevant Market Notice, para 41.

981 Email CU-L-2-43 of 12-09-2008; Email CO-L-2-13 of 17 September 2008.

982 Email CU-L-2-48 of 04 September 2008.

983 Annex 33, Notifying parties' Response to Article 11 letter of 5 September 2008.

984 Email CU-L-2-48 of 04 September-2008.

985 Notifying parties' Response to Article 11 letter of 5 September 2008.

986 Reply CO-L-2-6.

987 Email CU-L-2-9 of 08-09-2008 and Email CO-L-2-12 of 11 September 2008.

988 The notifying parties have provided examples that in certain formulations under certain conditions it would be possible to use starch 1500 instead of lactose for direct compressed tablets.

989 CO-L-2-13 of 17 September 2008.

990 Email CU-L-2-43 of 11 September 2008.

991 Notifying parties' Response to article 11 5 September 2008.

992 Annex 34 Notifying parties' Response to article 11 5 September 2008.

993 Email CU-L-2-48 of 11 September 2008.

994 The price differences between the different products in Asia and the Americas are similar.

995 Questionnaire to Customers pharma lactose, 17 June 2008 question 6.

996 Reply CU-PL-I-41.

997 Email CO-L-2-22 of 3 September 2008.

998 Questionnaire to Customers pharma lactose, 17 June 2008 questions 5 and 6.

999 Replies CU-PL-I-41, CU-PL-I-12, CU-PL-I-2, CU-PL-I-20, CU-PL-I-43, CU-PL-I-48, CU-PL-I-34, CU-PL-I-5.

1000 On 8 September 2008 the notifying parties have submitted two opinions that discuss the possibilities to exchange the use of different excipients with filler binder properties. It is considered that ad hoc documents for the purpose of merger control proceedings should be scrutinized with utmost care. Unlike pre-existing studies they have not been prepared in the normal course of business for the adoption of business decisions. This view is similar to Commission Notice on the definition of the relevant market for the purposes of Community competition law, [1997] OJ C372/5, [1998] 4 CMLR 177, para 41.

1001 Replies CU-PL-I-41, CU-PL-I-12, CU-PL-I-2, CU-PL-I-20, CU-PL-I-43, CU-PL-I-48, CU-PL-I-34 and CU-PLI-5.

1002 Reply CU-PL-I-2.

1003 For example CU-PL-I-34, CU-PL-I-48 and CU-PL-I-5.

1004 Reply CU-PL-I-12.

1005 Cf. Commission Notice on the definition of the relevant market for the purposes of Community competition law, [1997] OJ C372/5, [1998] 4 CMLR 177, para 42.

1006 Email CO-L-2-12 of 11 September 2008.

1007 Commission Notice on the definition of the relevant market for the purposes of Community competition law, [1997] OJ C372/5, [1998] 4 CMLR 177, paragraph 13.

1008 See for example Nestle Perrier and Metso Svedala paras 69-72.

1009 Form CO Section 6.O.1/25.

1010 Questionnaire to Customers pharma lactose, 17 June 2008 questions 11 and 12.

1011 Notifying parties' response to Article 6(1)(c) Decision page 69.

1012 Cf. paragraphs 20 - 24 of the Commission Notice on the definition of Relevant Markets for the purpose of Community Competition Law, 97-C 372-03.

1013 Questionnaire to Competitors pharma lactose, 17 June 2008 question 14.

1014 Reply CO-L-2-6.

1015 Reply CO-L-2-6.

1016 Reply CO-L-2-9.

1017 Questionnaire to Competitors pharma lactose, 17 June 2008 question 14.

1018 Cf. paragraphs 20 - 24 of the Commission Notice on the definition of Relevant Markets for the purpose of Community Competition Law, 97-C 372-03.

1019 Questionnaire to Customers food lactose, 17 June 2008 questions 14.

1020 Questionnaire to Competitors pharma lactose, 17 June 2008 question 21and Questionnaire to Customers pharma lactose, 17 June 2008 question 19.

1021 For example replies CU-PL-I-6 and CU-PL-I-7.

1022 Annex 7.01 to the Form CO.

1023 Questionnaire to Customers pharma lactose, 17 June 2008 question 20. (je n'ai pas trouvé la note 1024 ?? - isabell)

1025 BCC market research report "Excipients in pharmaceuticals" of Norma Corbitt from BCC research from Washington

1026 Email of 12 November 2008 by competitor Hilmar and Email of competitor Alpavit of 6 November 2008

1027 Reply CO-PL1-6 and CO-PL-1-9.

1028 Reply CO-L-2-1.

1029 Replies CO-L-2-10 and CO-L-2-1.

1030 CO-L-2-10.

1031 Guidelines on the assessment of horizontal mergers under the Council Regulation on the control of concentrations between undertakings Official Journal C 31, 05.02.2004.

1032 This production capacity can only produce one single type of pharmaceutical lactose (200 mesh). However, this particular type of product 200 mesh is 60% of the total market volume for pharmaceutical grade lactose, according to the Notifying parties..

1033 Reply CU-L-I-9.

1034 Replies CU-L-I-41, CU-L-I-48, CU-L-I-33, CU-L-I-9 and CU-L-I-27.

1035 Form CO Annex 7.O2.

1036 Submission of the notifying parties upon request of the Commission on 31st of October 2008

1037 The market shares of Campina and Friesland Foods are based on the notifying parties own estimates in the Form ,CO. As the market investigation did not confirm that any other supplier is active in DPI lactose the remaining market value / volume was attributed to the market share of Meggle, for the purpose of this decision.

1038 Reply CU-PL-I-9.

1039 The black out period is a period during which the merged entity will abstain from any use of the brand in the fresh dairy products markets. In this case, this black out period will be perpetual, which means that the parties will not use the Friesche Vlag brand in the fresh dairy products markets during an indefinite period of time.

1040 OJ C 267, 22.10.2008, p.1.

1041 The raw milk supply agreement as well as the Dutch Milk Fund are discussed in more detail in section 17-4.

1042 See reply to questions 2-4 Questionnaire to competitors on remedies and questions 2-5 Questionnaire to customers on remedies.

1043 See Remedies Notice, point 41. The Notice also argues (point 40) that "a re-branding remedy may be acceptable in circumstances where the brand at stake is widely used and a high proportion of its turnover is generated in markets outside those in which competition concerns have been identified". The circumstances apply in the case of Friesche Vlag, which according to the notifying parties generates [80-90]*% of its turnover outside the fresh dairy markets in the Netherlands.

1044 As explained in section 8.4.3, no competition concerns exist in the markets for the sale of rindless Dutch-type cheese (including narrower segmentations). It is therefore not necessary to require the overlap between the notifying parties in the rindless segment of the Dutch-type cheese markets to be divested.

1045 The "total" overlap between the notifying parties is lower where the nature Dutch-type cheese markets are defined more narrowly as only comprising Gouda or 15-day-old cheese.

1046 If it was assumed that this cheese sold to specialised cheese wholesalers will follow the same ratio as cheese procured by wholesalers in general (according to figure 8-1, 98 000 tonnes of the 263 000 tonnes of Dutch-type cheese procured by wholesalers from producers, i.e. 37%, were exported), it follows that 63% of this cheese, i.e. approximately 28 000 tonnes, will be sold to downstream distribution channels in the Netherlands. As modern types of retail account for approximately 55% of all downstream consumption and as in the OOH and industry segment more rindless cheese is used than at retail level, it can reasonably be assumed that approximately [...]* tonnes of nature Dutch-type cheese produced by the Cheese Divestment Business will be passed on to the level of modern types of retail in the Netherlands.

1047 See replies to questions 20 and 21 of the first questionnaire on remedies sent on 6 November 2008 to customers and replies to questions 23 and 24 of the first questionnaire on remedies sent on 6 November 2008 to competitors.

1048 See replies to questions 41 and 42 of the first questionnaire on remedies sent on 6 November 2008 to customers.

1049 See replies to questions 22 and 23 of the first questionnaire on remedies sent on 6 November 2008 to customers and replies to questions 26 and 27 of the first questionnaire on remedies sent on 6 November 2008 to competitors.

1050 See replies to question 29 of the first questionnaire on remedies sent on 6 November 2008 to customers and replies to questions 34 and 35 of the first questionnaire on remedies sent on 6 November 2008 to competitors.

1051 See replies to questions 23 et seq. of the first questionnaire on remedies sent on 6 November 2008 to competitors.

1052 See replies to question 25 of the first questionnaire on remedies sent on 6 November 2008 to competitors.

1053 See replies to question 36. of the first questionnaire on remedies sent on 6 November 2008 to customers and replies to questions 42 et seq. of the first questionnaire on remedies sent on 6 November 2008 to competitors.

1054 See replies to questions 37 of the first questionnaire on remedies sent on 6 November 2008 to customers and replies to question 44 of the first questionnaire on remedies sent on 6 November 2008 to competitors.

1055 See replies to question 44 of the first questionnaire on remedies sent on 6 November 2008 to customers and replies to question 47 of the first questionnaire on remedies sent on 6 November 2008 to competitors.

1056 See recitals6 and 97.

1057 For the first year, an appropriate supply arrangement may be negotiated between FrieslandCampina and the purchaser.

1058 See replies to first questionnaire on remedies sent on 6 November 2008, questions 12 and 37.

1059 Ibidem.

1060 See replies to first questionnaire on remedies sent on 6 November 2008, question 13 and 38.

1061 Ibidem.

1062 See replies to first questionnaire on remedies sent on 6 November 2008, question 71 and 7.

1063 See replies to second questionnaire on remedies sent on 19 November 2008, question 2 c).

1064 See replies to second questionnaire on remedies sent on 19 November 2008, question 5 a).

1065 See replies to second questionnaire on remedies sent on 19 November 2008, question 4 a).

1066 See replies to questionnaire on remedies sent on 6 November 2008, question 53.

1067 See replies to questionnaire competitors on remedies sent on 6 November 2008, question 57.

Notes II

1 See for example EDIM, cf. Consortium INRA-Wageningen, 2002; FAPRI Europe, cf. FAPRI, 2004.

2 See Bouamra-Mechemach et al. (Nov 2007) for a comprehensive survey as well as additional results.

3 Note that when a product is sold under promotion, it has its own ean.

4 Measure Guide, version 3 - December 2004.

5 See e-mail from RBB to Commission dated 03/09/2008 and titled "Retail structure in IRI data"

6 Example: if vendor = CAMPINA NL, it seems reasonable to assume that the major brand is Campina, so the observation "no brand" is transformed into "CAMPINA".

7 For example, comparing the level of prices between fresh and long lasting milk.

8 The special pack promotion does not appear explicitly in our dataset. On the basis of the information available, however, it should be possible to conclude that a product is sold under this promotion if a positive value is observed in correspondence of any of the promotion variables, except "any promotion" (which is the only one that could include this information).

9 On this and related issues concerning demand estimation see Hosken et al (2002)

10 For a more refined presentation of this simple argument see Sheffman, D. 1991. Statistical Measures of Market Power: Uses and Abuses. In The Cutting Edge of Antitrust: Market Power, American Bar Association, An Econometric Analysis for RTE Cereal Section of Antitrust, Willard Intercontinental Hotel, Washington, D.C., October 17-18

11 For a full discussion see for example WERDEN, G.J., 1992. "Four Suggestions on Market Delineation", Antitrust Bulletin, Vol. 37, No. 1, (Spring), pp. 107-121.

12 KASERMAN, D.L., and H. ZEISEL, 1996. "Market Definition: Implementing the Department of Justice Merger Guidelines", Antitrust Bulletin, Vol. 41, No. 3 (Fall), pp. 665-690.

13 The introduction to this section largely follows Hauman & Leonard (2004)

14 In case demand elasticities were to be used to estimate the price effects of a merger it should be noted that imprecision in the latter naturally translates into imprecision in the former. Choice of functional form can have major implications for the magnitude of predicted prices. For example, if the researcher assumes that the demand functions exhibit constant elasticities, the predicted post-merger price increases will be much larger than if linear or logit demands are assumed. See Crooke et al (1999) for an analysis of the importance of the "curvature" of different specifications of demand for extrapolating post-merger prices. Here "curvature" refers to how quickly elasticity increases as price rises. A rapid increase in elasticity will quickly curtail price increases post-merger.

15 See, for example, Hausman et al. (1994) or Cotteril & Haller (1997)

16 See Hausman and Leonard (1997) or Werden (1997)

17 See Green & Alston, (1990) or Chalfant (1987).

18 By duality: E(p, V(p, M)) = M and g(p, M) = h(p, V(p, M)). By Shephards lemma the partial derivatives of the expenditure function gives the Hicksian demand function: ( , ) ( , ) i i i E u p h u p q p ? = = ?

19 Adding-up requires that 1 ( , ) N i i i pq p M M = S = and homogeneity that ( , ) ( , ) i i i q ?p?M =q p M?i

20 Note that the elasticity appears also on the right hand side and the solution thus comes from a system ofsimultaneous equations, which solves for all the elasticities jointly (see Church et al, page 5 for the reduced form version).

21 Hausman & Leonard's approach is implemented as follows: After allowing for the brand-city fixed effects, we use the prices from one city as instruments for other cities. The intuition is that prices in each city reflect both underlying product costs and cityspecific factors that vary over time as supermarkets run promotions on a particular product. To the extent that the stochastic city-specific factors are independent of each other, prices from one city can serve as instruments for another city. This approach assumes that unobserved shocks to product costs affect all cities equally (i.e., there are no cityspecific cost shocks) but that there are city-specific demand shocks (and not nationwide demand shocks). In that case, the prices in other cities can serve as valid instruments for the price in any particular city (i.e., the prices of fresh milk in (say) Brussels, or Dusseldorf, can serve as instruments for the price of milk in Amsterdam). In the context of the present case, there is no reason to expect that with respect to the consumption of dairy products there are region-specific shocks that would render the Hausman assumption appropriate. On the contrary, it would seem that not only supply shocks but also demand shocks are likely to affect all regions equally. This assumption is also not sustainable when for example advertising campaign are launched to reach all local markets (i.e. national advertising campaign), or when news about the reputation or the quality of the product spread across local markets. Bresnahan (1997) maintains that the identification strategy proposed by Hausman et al. is not convincing because there is little evidence that demand shocks across markets are not correlated. As, a result, the estimates are biased. Arguably, seasonality also affect demand across local markets, though this may be tackled by introducing dummy variables.

22 A further complication arises when using the Stone price index as a proxy for P. Buse (1994) shows that the share weight of the Stone price are in fact endogeneous. As a result, SUR cannot be the appropriate estimation ,procedure, but instead IV should be applied. However, Buse (1994) also shows that the IV procedure would also fail to remedy the problem as the instruments are automatically correlated with the disturbance term. As a result, both the SUR and the IV estimators are inconsistent. Does that mean that LAIDS should not be estimated? Buse (1994) shows that the bias of the elasticity estimates shrinks significantly in the presence of high multi-collinearity and when the model fits well the data.

23 Note that this reasoning assumes that the (common) demand shocks (e.g., from national ad campaigns) cannot be foreseen when retail prices are set, which is would appear unlikely in the present context.

24 Greene (2003, p207-212 and 340-351).

25 The random effects (RE) specification is another possibility. It is not considered though as its implementation is more involved in a system context. RE amounts to specifying a GLS structure within each equation. The specification relies on assuming the panel fixed effect is not correlated with the error term. If this assumption holds RE provides a more efficient estimator as the FE or FD estimators. However, if the assumption does not hold the estimator based on the RE specification is biased. This is not true for the FE or FD approaches: they are consistent irrespectively of the assumption's validity. That is, the FE and FD specifications are more robust.

26 Wooldridge (2002, p307-314).

27 It can often occur, particularly in differentiated product markets that a SSNIP can not be imposed on a particular product since customers would switch to a large number of other alternatives. However none of these alternatives dominates and it is thus arbitrary which alternative is aggregated next.

28 More technically, the relevant question is the magnitude of the cross elasticities between fruit and choco flavored LLDDs and not between fruit LLDDs and non-LLDD products or between choco LLDDs and non- LLDD products.

29 "Some have argued that RESET is a very general test for the model misspesification, including unobserved omitted variables and heteroscedasticity. Unfortunately, such use of RESET is largely misguided. [...]* The bottom line is that RESET is a functional form test, and nothing more." Wooldridge (2003) p.194.

30 See also Feleke and Liu (2005).

31 More detailed Augmented Dickey-Fuller tests reveal that in the case of the fresh flavored segment, only one time series (Friesland at Albert Heijn) in the problematic equation is non-stationary, while in the case of the vla segment all four of them - revealing a further, unresolved specification problem.

32 The SO's results are presented in Appendices 1-7 and its main conclusions in Section 5.2.

Notes III

1 Eight Elopak gable top lines, four tetrapak lines, two cup lines, two doypack lines, one PET line and one BIB (bags-in-box) line

2 In Mainstream 3 Elopak packaging lines are leased until 2012. In Specialties Elopak packaging lines are leased until 2009