EC, March 11, 1998, No 98-531

THE COMMISSION OF THE EUROPEAN COMMUNITIES,

Having regard to the Treaty establishing the European Community, Having regard to Council Regulation No 17 of 6 February 1962, first Regulation implementing Articles 85 and 86 of the Treaty (1), as last amended by the Act of Accession of Austria, Finland and Sweden, and in particular Article 3(1) thereof, Having regard to the applications lodged under Article 3(2) of Regulation No 17 by Masterfoods Limited and Valley Ice Cream (Ireland) Limited against Van den Bergh Foods Limited, formerly HB Ice Cream Limited, alleging that they are being hindered, in restraint of competition, in selling their ice-cream products in Ireland, Having regard to the Commission Decision of 29 July 1993 to initiate proceedings in this case, Having regard to the application for negative clearance/exemption lodged pursuant to Article 4(1) of Regulation No 17 by Van den Bergh Foods Ltd on 9 March 1995 for its distribution arrangements for impulse ice-cream products in Ireland, Having given the undertaking concerned, pursuant to Article 19(1) of Regulation No 17 in conjunction with Commission Regulation No 99-63-EEC of 25 July 1963 on the hearings provided for in Article 19(1) and (2) of Council Regulation No 17 (2), the opportunity of being heard on the matters to which the Commission has taken objection, Having consulted the Advisory Committee on Restrictive Practices and Dominant Positions, Whereas:

I. INTRODUCTION

(1) This Decision relates to the arrangements by Van den Bergh Foods Limited, formerly known as HB Ice Cream Limited, (hereinafter 'HB`) for the distribution of its impulse ice-cream products in Ireland. HB pursues a policy under which it makes freezer cabinets available to retail outlets stocking its ice-cream products on the basis of cabinet exclusivity.

(2) On 18 September 1991, Master Foods Limited, trading as Mars Ireland, (hereinafter 'Mars`) lodged with the Commission a complaint (3) under Article 3 of Regulation No 17 against HB, in respect of the latter's provision to a large number of retailers of freezer cabinets on the basis of cabinet exclusivity. Mars claims that it is being hindered, in restraint of competition, from gaining access to retail outlets for the sale of its impulse ice-cream products in Ireland by reason of this practice.

(3) Mars entered the Irish ice-cream market in 1989. In April 1990, the Irish High Court granted an interlocutory injunction to HB which prevented Mars products from being placed in HB freezer cabinets. In May 1992, the High Court (4) granted HB a permanent order restraining Mars from inducing retailers to store its ice-cream products in HB's cabinets. The court further held that HB's policy of freezer exclusivity was not caught by the prohibition contained in Article 85(1). It also found that Article 86 had not been infringed, either by HB's exclusivity or by its pricing policies. The court did, however, consider that HB held a dominant position on the Irish impulse ice-cream market. Mars has appealed against these judgments to the Irish Supreme Court.

(4) On 22 July 1992, Valley Ice Cream (Ireland) Limited ('Valley`) lodged with the Commission a complaint (5), against HB similar in terms to the Mars complaint. Valley was a manufacturer of ice cream in Ireland, but went into liquidation in 1997. It had, until shortly before that, been acting as principal distributor for Mars ice-cream products in Ireland.

(5) On 29 July 1993, the Commission provisionally concluded that the system of distribution at that time operated by HB infringed both Article 85 and Article 86, addressing a Statement of Objections against HB to that effect. HB contested the Commission's view of the legality of its distribution arrangements.

(6) In its Statement of Objections, the Commission did accept that HB's arrangements bring about some improvement in distribution, but found that this was outweighed by the harmful effects of restricted competition. It was none the less made clear that HB remained at liberty to propose an alteration to its distribution arrangements for impulse ice cream in Ireland which might be capable of qualifying for an exemption pursuant to Article 85(3). While maintaining its view as to the legality of its current distribution arrangements, and following discussions with the Commission, HB came forward with such proposals. On 8 March 1995, these were notified to the Commission (6), which then issued a statement (7) expressing the prima facie view that the changes would merit exemption.

(7) On the basis of these proposals and in the light of HB's expressed expectations regarding their effects in the market, on 15 August 1995 (8) the Commission announced its intention to take a favourable view towards HB's distribution arrangements as notified. The changes did not, however, achieve the expected results in terms of opened outlets. In view of this and of the situation as it currently stands in the market, the Commission has revised its expressed intention (see paragraph 247 et seq.). Accordingly, on 22 January 1997 the Commission sent a new Statement of Objections to HB. HB submitted its written response thereto on 24 April 1997, and put forward its arguments at an oral hearing held on 11 June 1997.

II. THE FACTS

1. The product

(i) Ice cream generally

(8) The essential raw materials for the production of ice cream include milk (including skimmed milk), fat (milk or vegetable) and sugar. Other additions may include, inter alia, emulsifying and stabilising agents, colouring and flavouring substances, nuts, chocolate, and fruit concentrate. Legislation concerning product definition, labelling, and permitted contents (air coefficient etc.) varies somewhat between the different Member States. The contents of frozen yoghurt, 'water ices`/'ice pops` (water-based products) and 'soft`/'whipped` ice cream (which is made with vegetable fat and has no dairy content) differ substantially from ice- cream products generally.

(9) Ice-cream products must be kept at a low temperature during all stages of production and distribution before ultimate consumption. An exception to this is soft ice cream which is delivered to a retailer in the form of an ambient soft mix (sold to retailers either as a 'fresh` or 'sterile` product), and is only transformed into a final product by being processed through special dispensing equipment on the retailer's premises.

(10) Ice cream can be categorised according to how it is manufactured and distributed. A distinction is usually made between 'artisan` ice cream on the one hand, which is generally produced, distributed and consumed locally on a small scale, and industrial ice cream on the other, which is produced for wide-scale distribution. Almost all of the ice cream produced and sold in Ireland is of the industrial kind.

(11) Ice cream can also be categorised according to where it is intended to be consumed. This generally includes three categories:

- 'impulse` ice cream, consisting of single-wrapped items (often stick products), and individual portions of 'scooped` ice cream, designed for immediate consumption at or near the place of purchase,

- 'take-home` ice cream, including multipacks of single items, blocks, tubs, dessert products and so on, intended for consumption at home,

- catering ice cream, which is sold in bulk to hotels, restaurants, cafés, etc., for consumption, as part of a catering service, in those places.

(ii) Impulse ice cream

(12) Single-wrapped items of impulse ice cream are generally stored in self-service freezer cabinets installed in or just outside the retailer's premises. The cabinets are usually prominently positioned. As with all products relying on impulse purchasing decisions, availability and visibility at point of sale are important factors in the sale of impulse ice cream. If the product is not available or visible, that is a sales opportunity which may be permanently lost (9), consumers of impulse ice cream will in general not travel to buy, or defer a purchasing decision. Access to wide-scale distribution is particularly crucial for the sampling by consumers of new products, thereby stimulating demand for those products. Single-wrapped impulse ice-cream products are almost invariably branded products, supported by considerable investment in advertising and other promotional activity.

(13) Seasonal factors have a considerable effect on the consumption of impulse ice cream, with most sales being made during the summer months. In 1992, for example, approximately two thirds of HB's impulse sales were made in the four-month period from May to August. Sales also depend to a substantial degree on the weather.

(14) While children are particularly important as purchasers of single item ice-cream products, the trend in new industrial impulse lines is towards 'premium` or 'super premium` items (terms generally denoting products with a high dairy fat content and a low proportion of air), which appeal primarily to adult consumers. These are more highly priced products aimed at a wider market which includes adults, and represent the fastest growing segment of the impulse market.

(15) Scooping ice cream is delivered in bulk form to the retailer, who generally stores it in a specially-designed freezer cabinet and serves it to a customer on a cone. In Ireland, scooping ice cream is of only marginal importance in terms of sales.

(16) Soft ice cream is delivered in the form of an ambient soft mix to the retailer, who must then further process the mix by the addition of further ingredients and by the use of special dispensing equipment which transforms it into the low-temperature product which is served to the customer. Soft ice cream is usually served on a wafer cone.

(17) From the manufacturer's point of view, large-scale business in impulse ice cream is capital intensive and involves considerable business risks due to its dependence on unpredictable weather conditions. It requires high levels of working capital in order to finance manufacture in advance of the season and to hold stocks. Considerable logistical and planning demands are imposed on the business over the annual cycle in order to support the relatively short campaign season.

2. The ice-cream market in the Community

(18) Within the Community, Member States' ice-cream markets demonstrate different characteristics. The annual consumption of 'artisan` ice cream, for example, ranges from over six litres per capita in Italy to less than half a litre per capita in Spain, and even less in Ireland. Industrial supply accounts for most of the products sold in Europe. Production of industrial ice cream is highest in the UK and, in descending order, in Germany, Italy and France. Sales of impulse ice cream as a proportion of the total ice-cream market also vary considerably between Member States.

(19) Companies in the Unilever group are the largest ice-cream suppliers in most Member States. In the impulse sector, Unilever companies are the market leaders in every Member State with the exception of Spain, Greece and Finland. The Nestlé, Artic/Beatrice, Grand Metropolitan and Schöller groups all sell ice cream in more than one Member State. Some of these groups of companies. Unilever and Nestlé for example, trade in different Member States under different names.

(20) National legislation on the content of ice cream is not harmonised in the Community. The differences that exist, especially those relating to the permitted types of dairy and vegetable fat, may give rise to material differences in the products available in different Member States and in their cost. Average retail prices for similar and even identical ice-cream products can also vary significantly from one Member State to another. An analysis, however, of the range of products on offer for sale by suppliers operating in several markets shows that similar or identical products are often sold under different product or brand names. There is, moreover, a considerable volume of trade between Member States in ice-cream products.

(21) The practice of supplying freezer cabinets to retailers subject to a condition of exclusivity is widely employed by ice-cream manufacturers and distributors throughout Europe, both by companies of the Unilever group and by many of its competitors.

3. The ice-cream market in Ireland

(i) General

(22) Per capita consumption of ice cream in Ireland is one of the highest in the Community. The total size, in volume and value terms, of industrial ice cream marketed in Ireland is outlined in the table below for 1995 and 1996 (10):

<EMPLACEMENT TABLEAU>

During the summer of 1995, sales (in particular of impulse products) were boosted by good weather conditions. Comparatively poorer weather conditions during the summer of 1996 explain the relative decline in sales for that year. The total value of the soft ice-cream market (11) in Ireland is estimated by HB as standing at approximately IEP [. . .] annually.

(ii) HB

(23) HB is Ireland's principal manufacturer and distributor of ice-cream products. The company traces its position as the predominant supplier of ice-cream products in Ireland back to its acquisition in 1968 of the ice-cream business of its principal rival at the time, Premier Dairies(12). It has, since 1974, been a part of the Unilever group of companies. The company was reformed in 1993 as Van den Bergh Foods Ltd, following the merger of HB Ice Cream Limited with other companies in the Unilever group. In 1996, it achieved a turnover of some IEP [. . .]. The corresponding turnover figure for 1995 was IEP [. . .], up from IEP [. . .] in 1994 and IEP [. . .] in 1993.

(24) The multinational Unilever group of companies is one of the world's leading consumer-goods manufacturers. The two parent companies, Unilever NV and Unilever PLC, achieved a worldwide consolidated turnover in 1996 of some GBP 33,5 billion. The turnover of the Unilever group in Ireland for the same period amounted to some IEP 231 million. Companies belonging to the Unilever group produce and distribute ice-cream products in all Member States, in most of which Unilever is the market leader. Unilever's policy in the various national ice-cream markets has been to acquire local ice-cream companies and develop brands on a national basis. More recently, however, it has developed some products internationally, notably the highly-successful 'Magnum` stick bar.

(25) HB supplies a complete range of ice-cream products, including a full range of impulse items (including premium products), many of which enjoy very great popularity with consumers. The majority of the products are manufactured in Ireland, but some items are made by other Unilever ice-cream companies and imported by HB. The ice cream is marketed under the 'HB` brand, consumer awareness of which is very high in Ireland. HB also sells ambient soft mix to retailers for transformation by them into soft ice cream.

(26) The total volume of ice cream marketed by HB in Ireland is outlined in the table below (1 000 litres):

<EMPLACEMENT TABLEAU>

(27) In 1996, HB marketed a total in impulse ice cream, in value terms, of IEP [. . .] at GSV (gross sales value), compared with a marketed total in 1995 of some IEP [. . .], and in 1994 of some IEP [. . .].

(28) HB achieved a share of the market (13), by value, for impulse ice cream in Ireland of about 85 % during the four-month period from June to September of 1996, up from 83 % for the same period in 1995, from 77 % for the summer of 1994 and from 76 % for the summer of 1993. In volume terms, HB's market share in the summer of 1996 amounted to some 85 %, up from 83 % for the same period in 1995, and from 80 % in 1994 and 1993. In the June/July period of 1997, HB's volume share amounted to 89 % and its value share was also 89 %. HB attributes this increase in its market share, at least in part, to the financial difficulties experienced by Valley, and to the termination of the joint-distribution arrangement between Mars and Valley. HB estimates its share of the soft ice-cream market in Ireland as standing at about [. . .] %.

(iii) Other manufacturers

(29) The remainder of the Irish ice-cream market is shared principally between Mars, Valley, Nestlé, Leadmore (Ireland) Ltd (hereinafter 'Leadmore`), Dale Farm Dairy Group Ltd (hereinafter 'Dale Farm`), and Häagen Dazs.

(a) Mars

(30) Mars is a wholly-owned subsidiary of the privately-owned US-multinational Mars Incorporated. As such, it is part of an international group with subsidiaries in most Member States. Mars first entered the ice-cream market when it purchased Dove, a US ice-cream company. Building on the renown of its ambient confectionery brands, Mars developed ice-cream versions of some of those brands, thereby extending them into another product category. Mars manufactures all of its ice cream for the European market at its production facilities in France (Doveurope SA) and uses the same product names in each Member State. It first launched its ice-cream products in Europe in 1988.

(31) Mars began selling its ice cream in Ireland, in single-item form as well as in multipacks, in 1989. The brands' already high recognition among consumers reduced the costs of market launch. The range of ice-cream products offered by Mars is limited to a relatively small number of items, most of which are in the premium (high-price) segment of the market and appeal primarily to adult consumers.

(32) Mars' market share in value terms for impulse ice cream in Ireland during the period from June to September 1996 was about 6 %, down from 7 % for the same period in 1995, and down from about 10 % for the summer of 1994 and 14 % for 1993. Mars' shares in volume terms are somewhat lower, reflecting the premium nature of the range of products which it offers. In the summers of 1995 and 1996, it achieved some 4 %, down from about 6 % in summer 1994, and from 8 % in 1993. In the June/July period of 1997, Mars' volume share amounted to 4 % and its value share was 5 %.

(b) Valley

(33) Until its liquidation in October 1997, Valley supplied a wide range of impulse ice-cream products, though it did not offer products in the premium category. Valley had a history of financial difficulty, including its liquidation and subsequent reconstitution in 1985. It had a market share (value) in the summer of 1996 of about 6 %, compared with a share for 1994 and 1995 of 7 %, and for 1993 of 5 %. Its share in volume terms amounted to about 8 % in 1996, compared to 9 % the previous year. In the June/July period of 1997, Valley's volume share amounted to 3 % and its value share was 2 %. From 1989 until early 1997, Valley was the primary distributor of Mars' impulse ice-cream products.

(c) Nestlé

(34) Nestlé is a multinational company with ice-cream subsidiaries in most Member States. Nestlé entered the Irish market in 1994, and achieved a market share by value and volume of about 2 % in 1995 and 1996, up from about 1 % in 1994. In the June/July period of 1997, Nestlé's volume and value shares both amounted to 2 %.

(d) Häagen Dazs

(35) Häagen Dazs (part of the Grand Metropolitan group of companies) entered the Irish impulse ice-cream market in 1993 with a limited range of premium-priced luxury ice creams. Its share of the impulse ice-cream market is very small and is not separately calculated by Nielsen.

(e) Dale Farm

(36) Dale Farm, a subsidiary of the UK dairy company Northern Foods, is a company primarily based in Northern Ireland and the market leader there; it distributes its ice-cream products via independent franchisees and agents. It secured a 1 % volume and value market share in the summers of 1995 and 1996, up from 0,5 % in 1994. In the June/July period of 1997, Dale Farm's volume and value shares both amounted to 1,5 %. Dale Farm also distributes Mars' products, though these are supplied to it by the UK, rather than the Irish, Mars subsidiary.

(f) Leadmore

(37) Leadmore is a regional producer of ice-cream products. In December 1994, Leadmore Ice Cream Ltd went into receivership. It was purchased as a going concern from the receiver in May 1995 and reconstituted as Leadmore (Ireland) Ltd. Leadmore achieved a value market share of about 0,5 % in the summers of 1995, 1996 and June/July 1997, down from 3 % in the summer of 1994. Leadmore largely operates within a limited geographic area.

(g) Soft-mix suppliers

(38) Apart from HB (see above), the other principal suppliers of soft mix in Ireland are: R& F Martin (trading under the name 'Angelito's`), with an estimated 35 % market share; Pritchards, with an estimated 20 % share; other suppliers include Leepatrick Dairies and Dawn Dairies, with even smaller shares.

4. Distribution channels for impulse ice cream in Ireland

(i) General

(39) In Ireland, industrial impulse ice cream is distributed to consumers through a wide array of retail and other outlets ranging from grocery stores, the 'traditional` retail trade (often referred to as TSNs - tobacconists, sweetshops, newsagents), garage forecourts, and kiosks to places of entertainment and leisure. By far the most important outlets for impulse sales are the traditional retail trade, grocery stores (including symbol/franchise groups) and garages. This is illustrated by AC Nielsen surveys (14) of the ice-cream market in Ireland for the summer of 1992 and April/May 1995 which gave the following breakdown (in value terms) of impulse ice cream sold through the different retail channels:

<EMPLACEMENT TABLEAU>

(40) The AC Nielsen retail censuses for 1991 and 1994 show the following retail outlet structure for the food and confectionery trade (numbers of outlets):

<EMPLACEMENT TABLEAU>

(41) As can be seen, the structure has changed little over the three-year period, except for a decline in the number of grocery outlets. Most of these outlets sell impulse ice cream. In 1994, the proportion was 95 %. HB estimates the number of retail outlets selling impulse ice cream as standing (in 1997) at 9 454, on the basis of the AC Nielsen figures.

(42) Multiples (a category of outlets essentially comprising supermarkets) sell principally take-home ice cream, including multipacks, and these are often stored, as are frozen foods, in the retailers' own freezer cabinets. It should also be noted that in some of the smaller grocery outlets in Ireland, impulse and take-home ice-cream products can be sold from the same outlets and even from the same freezer cabinet.

(43) Normal retail outlets clearly have limited space for the siting of freezer cabinets. The outlets which are the most important for the sale of impulse ice cream are generally small in area, and are consequently under particular space constraints. Impulse ice cream is a marginal product for most retailers, in that it will only account for a limited percentage of an outlet's turnover and profits, and competes for sales space with a range of other (impulse and non-impulse) products, many of which have the advantages that their sales are not seasonally variable and so annually may provide higher sales per unit of area (15). According to information provided by HB, ice cream amounts to only 8,5 % in value of total sales in Ireland of 'impulse` products (confectionery, soft drinks, savoury snacks and ice cream). Moreover, the space which is available in an outlet is not all of equal value for the sale of impulse products: prominence of display is crucial (see footnote 9).

(44) Access to distribution is normally measured in terms of: (a) 'unweighted (numeric) distribution`, namely the percentage of outlets offering impulse ice cream, in numeric terms, where a manufacturer's products are available; and (b) 'weighted distribution`, namely the percentage of outlets offering impulse ice cream where a manufacturer's products are available, weighted for turnover in impulse ice cream. Although weighted distribution is generally considered the more important measure of a company's distribution levels, in the case of impulse products numeric distribution is of particular importance: it is the most accurate indication of the widespread availability (and visibility) of the product to consumers. The importance of high levels of numeric distribution is most pronounced where impulse products are being launched in the market for the first time, thereby increasing the chance that consumers will try the product at least once.

(ii) HB

(45) HB ice-cream products are available in each category of distribution channel referred to at paragraph 38. HB sells its impulse ice-cream products to [. . .] customers in total. [. . .] ([. . .] %) of HB's customers have an annual turnover in HB impulse ice cream of less than IEP 1 500.

(46) According to AC Nielsen surveys conducted in Ireland over the last few years, HB achieved about 79 % numeric distribution and 94 % weighted distribution for its impulse ice- cream products in the August/September period of 1995. This means that its products were being sold from 79 % of outlets by number (16), and that HB impulse products were being sold from outlets accounting for 94 % of total impulse ice-cream sales in Ireland. The figures were almost the same in the June/July period of that year, when the weighted share was 95 %. In June/July 1994, the numeric and weighted distribution figures for HB were 78 % and 92 % respectively. In August/September of that year, the figures were 79 % and 93 %. In the June/July period of 1996, HB's numeric distribution was 82 % and its weighted share 96 %; in August/September 1996, the figures remained unchanged, but in June/July 1997, HB's numeric distribution was 85 % and its weighted share 97 %.

(iii) Mars

(47) According to AC Nielsen, Mars achieved about 35 % numeric distribution and 48 % weighted distribution for its impulse ice-cream products in the August/September period of 1995. The figures were the same in the June/July period of that year. In June/July 1994, the numeric and weighted distribution figures for Mars were 36 % and 56 % respectively. In August/September of that year, the figures were 34 % and 52 %. In the June/July period of 1996, Mars' numeric distribution was 32 % and its weighted share 43 %; in August/September 1996, the numeric figure was 31 % and the weighted share remained unchanged. In June/July 1997, its numeric distribution was 30 % and the weighted share 40 %.

(48) Mars estimates (on the basis of a survey carried out by its salesforce at the time) that, prior to the granting of the interlocutory injunction by the Irish High Court in 1990 (subsequently made permanent, see paragraph 3), preventing Mars from inducing retailers to stock Mars ice-cream products in HB cabinets, it had attained a level of some 42 % numeric distribution in the impulse ice-cream market in Ireland. This level was reached within the first few months of the launch of its products in the Irish market and was achieved, in large part, by the placing of Mars products in HB freezer cabinets. Following the injunction, Mars estimates that its level of distribution fell to below 20 %.

(iv) Other manufacturers

(49) Valley achieved numeric distribution in August/September 1995 of 29 % and weighted distribution of 36 % in the same period. In June/July 1995 the figures were 30 % and 37 % respectively. In June/July 1994, they were 28 % and 39 % respectively; in August/September 1994, they were 29 % and 39 % respectively. In the June/July period of 1996, Valley's numeric distribution was 26 % and its weighted share 31 %; in August/September 1996, the weighted figure was 30 % and the numeric share remained unchanged. In June/July 1997, its numeric distribution fell to 20 % and its weighted share to 19 %.

(50) Nestlé achieved numeric distribution in AugustSeptember 1995 of 20 %, and weighted distribution of 25 % in the same period. In June/July 1995 the figures were 20 % and 26 % respectively. In June/July 1994, they were 19 % and 18 % respectively; in August/September 1994, they were 20 % and 19 % respectively. Most of Nestlé's distribution figures in 1994 were accounted for by sales of just one of its range of products, 'Fruit Pastilles`. In the June/July period of 1996, Nestlés numeric distribution was 19 % and its weighted share 24 %; in August/September 1996, the numeric figure was 18 % and the weighted share 25 %. In June/July 1997, its numeric distribution increased to 20 % and its weighted share declined to 21 %.

(51) Dale Farm achieved numeric distribution in August/September 1995 of 11 %, and weighted distribution of 12 % in the same period. In June/July 1995 the figures were 11 % and 11 % respectively. In June/July 1994, they were 5 % and 6 % respectively; in August/September 1994, they were 4 % and 6 % respectively. In the June/July period of 1996, Dale Farm's numeric distribution was 9 % and its weighted share 8 %; in August/September 1996, the numeric figure was 8 % and the weighted share 7 %. In June/July 1997, the numeric share was 11 % and the weighted share 11 %.

(52) Prior to going into receivership at the end of 1994, Leadmore had secured in June/July 1994 a 17 % numeric distribution and a 15 % weighted figure. Its numeric distribution figure in the same period of 1995 was some 10 %. In the June/July period of 1996, Leadmore's numeric distribution was 7 % and its weighted share 6 %. In June/July 1997, the numeric share was at 4 % and the weighted share 7 %.

5. Distribution arrangements for impulse ice cream in Ireland generally

(53) The wide-scale distribution of industrial ice cream requires, among other things, frequent delivery using refrigerated transport from the factory to the sales outlet or catering establishment, which must in turn have cold-storage facilities (usually freezer cabinets) on the premises. A manufacturer of ice cream needs either to build up his own distribution network for frozen products, or to conclude an agreement allowing him access to the distribution network of another ice-cream manufacturer or wholesaler. A distributor needs to be able to ensure frequent 'drops` and to respond to orders at short notice, particularly in summer.

(54) The custom in Ireland has been for manufacturers supplying impulse ice cream to lend freezer cabinets to the retailer free on loan (i.e. with no direct charge) and subject to freezer- cabinet agreements, under which the retailer undertakes that freezers are to be used exclusively for the storage of the supplier's products (cabinet exclusivity); this is illustrated in the market research detailed below (see paragraphs 90 and 91). The supplier usually provides a maintenance service for the freezer cabinets, also free of charge (with no direct charge). Manufacturers often supply most of their products to retailers direct, including via agents, without selling to independent intermediaries; wholesale business in ice cream is small-scale in Ireland.

(55) Until the disposal by HB of a portion of its cabinet fleet in 1996 (see paragraphs 72 and 73), there appear to have been very few retailer-owned cabinets in place in Ireland for the storage and sale of impulse ice-cream products, as is illustrated by the market research outlined at paragraph 85 et seq.

6. HB's distribution arrangements for impulse ice cream in Ireland

(i) General

(56) HB operates a direct distribution system to 80 % of its ice-cream outlets in Ireland. Delivery is made direct to the cabinet, whether it is owned by HB or otherwise. HB has regional depots, from which local delivery in its own fleet of refrigerated vans is carried out. The remaining 20 % of its outlets receive deliveries from eight different local agents who distribute on behalf of HB; the arrangements with these distributors are long-standing and stem from logistical difficulties for HB in otherwise achieving geographical coverage by direct distribution to all outlets. Within the Irish market, HB alone has a nationwide, vertically integrated distribution network.

(57) HB provides turnover-related discounts for purchases of its ice cream (not applicable to soft-mix ice cream). These are given in accordance with a graduated scale ranging from (for 1997) a 2 % discount for purchases of IEP 1 350 to IEP 2 750 to an 11 % discount for purchases of IEP 26 601 or more. The discounts are made available to all retailers of HB ice cream irrespective of the question of cabinet ownership.

(ii) Cabinet provision

(a) HB-exclusive cabinets

(58) HB has for many years made available to retailers freezer cabinets for the storage and display of impulse ice cream at the point of sale (17). The cabinet is either loaned to the retailer with no direct charge (see paragraphs 76 to 79) or leased for a nominal annual rent (IEP 1) which is not collected. The maintenance and repair of the cabinets is also performed by HB. HB does not provide dispensing machinery to retailers for the processing of soft ice cream.

(59) The cabinet is supplied in accordance with a standard form contract concluded between HB on the one hand and the retailer on the other; the main provisions of the cabinet agreement are that:

- HB agrees to make a freezer cabinet available to the retailer; the ownership and property in the cabinet remains that of HB, and HB undertakes to maintain it at its own cost and expense (save where damage is due to the retailer's misuse or neglect of the cabinet),

- the cabinet is to be used exclusively for storing products for sale which are supplied by HB; accordingly no product manufactured or supplied by a third party is to be sold or offered for sale from the freezer cabinet, or stored therein,

- the agreement is terminable at any time by either party giving two months' notice (18),

- the retailer undertakes to keep the cabinet in a prominent position on his premises. Only HB advertising material may be displayed on top of or on the sides of the cabinet.

(60) The number of retailers which whom HB has such cabinet agreements (those supplied subject to a condition as to exclusivity) amounts (in 1997) to 7 907. HB has given a breakdown of its freezer-cabinet agreements as follows:

<EMPLACEMENT TABLEAU>

(61) HB has at present (in 1997) [. . .] cabinets installed in outlets in Ireland as 'front-of-shop` (19) cabinets. In 1995, HB marketed IEP [. . .] in outlets with HB freezer-cabinet agreements, representing [. . . more than 95 % . . .] of the total impulse ice cream marketed by HB in Ireland in that year (see paragraph 27). In 1994, HB marketed IEP [. . .] in outlets with HB freezer-cabinet agreements, which also represented [. . . more than 95 % . . .] of the total impulse ice cream marketed by HB in Ireland in that year.

(62) It has not been HB's official policy to provide a cabinet to any outlet which offers to stock ice cream. Instead, both financial and trading criteria (20) are applied. The financial criteria are based on the marginal cost to HB involved in taking on a new outlet, coupled with an acceptable return on that investment. This includes an assessment therefore of (a) the net cost to HB of the cabinet, and (b) its variable margin on the associated ice-cream sales through the outlet. The trading criteria are expressed as an annual sales target, based on a minimum estimated turnover in the outlet in question.

(63) Substantial overall economies are achieved where HB, as opposed to individual retailers or even large buying groups, bears the cost of purchasing and maintaining freezer cabinets. As part of the Unilever Group (which buys a very large number of cabinets per year) HB obtains significant discounts for such scale purchases. By means of illustration, HB estimates that the capital cost (including delivery, storage, handling and so on) to a retailer of purchasing a standard 1 m freezer cabinet would be very considerably higher than the capital cost to HB of providing the same cabinet. HB employs its own refrigeration engineers for cabinet maintenance and repair, and also uses the services of local refrigeration companies. During 1994, 1995 and 1996, HB's expenditure on freezer cabinets (including purchase, installation, maintenance, cabinet 'control` and depreciation) amounted to [. . .] %, [. . .] % and [. . .] % of its turnover in impulse ice cream, respectively, for each of the three years.

(64) The importance attributed by the Unilever group of companies to the policy of supplying freezer cabinets subject to exclusivity became particularly evident following the entry of Mars on the European market in the late 1980s. The following extracts (paragraphs 65 to 68) from company documents are an illustration of how crucial Unilever saw the maintenance of freezer exclusivity at that time.

(65) In a Unilever group document dated 3 September 1990 and entitled 'Mars' position in Europe and the FPC (21) response, under the heading 'Strategic approach`, one of the points made is '. . . to make every effort to maintain cabinet exclusivity`.

(66) In another FPC document dated April 1989 and entitled 'European ice cream marketing strategy`, reference is made to the importance of freezer exclusivity retaining ownership of the freezer cabinet: 'We must retain ownership of the cabinet, particularly where distribution is performed by third parties, in order to retain, as far as possible through exclusivity contracts, sole brand supply to the fridge, and de facto, to the outlet`.

(67) The 'Minutes of the marketing managers meeting` held in Rotterdam (an FPC document) in November 1989 contain the following statements: 'HB has driven Mars down to 400 HB outlets (4,2 %) from their August peak of 1 920 cabinets (29,7 %) by establishing a Special Task Force . . . In certain cases, they [HB] have been able to buy Mars out of outlets. Diversion to other refrigeration in store and very selective cabinet withdrawal is also being used . . .`; under the heading 'Mars, What next?`, it is stated that, 'To achieve their economies of scale in production Mars needs to achieve significant volume increases. They need to break down the distribution barrier . . . . Denying Mars distribution is the key immediate response to their entry . . .`.

(68) In a Unilever internal strategy document entitled 'Frozen products coordination, Sales Directors Conference, June 1990, Vienna, highlights and follow-up action`, the importance of freezer exclusivity is underlined. Under the heading 'Exclusivity` it is stated: 'We are all too aware that in many countries one of the foundations of our success, EXCLUSIVITY, is now under threat:

- Exclusivity at concessionaire and distributor level,

- Exclusivity in the outlets and cabinets`.

(b) Revision of cabinet policy

(69) In the light of the Commission's Statement of Objections of 29 July 1993 (see paragraph 5), and following discussion with the Commission, HB has revised its distribution arrangements. These have been notified to the Commission, as an application for negative clearance or, failing that, exemption (see paragraph 6). The revised arrangements include the standard-form agreement under which HB continues, on request, to supply HB-exclusive freezer cabinets to retailers as just described (paragraphs 58 and 59), as well as HB's standard conditions of sale and discount schedule (see paragraph 57); these now include a new pricing scheme for its ice-cream products (see paragraphs 76 to 79) and the introduction of a hire- purchase scheme (see paragraphs 70 and 71) designed to facilitate the acquisition of freezer cabinets by retailers.

(70) HB has introduced, as an optional alternative to the acceptance of an HB cabinet under the standard-cabinet agreement, a freezer cabinet hire-purchase scheme designed to enable retainers to buy their own new cabinet. The cabinets are offered at the wholesale price at which HB purchased them, as is evidenced by an invoice from the independent cabinet supplier. Under the scheme, retailers are offered a 1 m 'visi-top` cabinet, available in respect of one cabinet per outlet at any time. Details of the scheme are published on HB price lists and relevant promotional material. Repayment is made over a maximum period of five years; the terms for early repayment without penalty are made clear to the retailer. The credit rate should be the prevailing hire-purchase rate in Ireland (22). The cabinet is to be exclusively used for the stocking of HB products during the period of repayment only. Maintenance is performed by Unilever throughout the repayment period, during which period hire-purchasers will qualify for the differential pricing lump sum (see paragraphs 76 to 79), minus the maintenance element (23).

(71) Unilever informed the Commission that it believed that the scheme would 'prove attractive to retailers and incentivise them to buy their own cabinets`. Since the introduction of the hire-purchase scheme in Ireland in early 1995, not a single retailer has availed himself of the offer.

(72) As a means of achieving a substantial freeing up of outlets in the short term, Unilever undertook to the Commission to ensure the sale of an in situ 'front-of-shop` freezer cabinet with 'a useful remaining life` to retailers in a significant number (some 20 %) of the outlets in which its cabinets were then (December 1994) installed; in total, 1 750 cabinets were to be sold. No more than 400 of such cabinets were to be sold in outlets with an annual turnover in HB impulse ice cream of below IEP 650, and none were to be in outlets with a turnover of less than IEP 400 p.a. in those products. From the moment of sale, a retailer was to be eligible for an annual lump sum under the differential pricing scheme (see paragraphs 76 to 79). Sale was to be made to retailers with current possession of a sole HB cabinet. The price was to be pitched at a level which would ensure achievement of these results. Sale was to be carried out on a 'one-off` basis during 1995 and 1996; sale was to take place equally in 1995 and 1996, and a reasonable geographic spread was to be ensured. The disposal of cabinets was intended to 'kick start` a movement towards freezer cabinet ownership by retailers. Unilever stated that the effect of the measure would be 'to provide, for the 1995 season, a visible and material change to the relevant market. This (would) provide a 'kick start` to the process, the future development of which is underwritten by the dual price structure and the hire-purchase scheme`.

(73) In accordance with this undertaking, HB has disposed of [. . .] cabinets in [. . .] outlets since the beginning of 1995; almost all of these disposals were carried out during the first half of 1996 (24) and [. . .]. According to HB, 1 750 of the disposals comply with the conditions cited in paragraph 72. Almost all of the cabinets disposed of were models and types of cabinet which have not been newly installed by HB in outlets over at least the last [. . .] years. The average age of the cabinets disposed of is just under [. . .] years (25). Not all of the cabinets disposed of had been installed in retail outlets; some ([. . .], according to HB) were installed in schools, staff canteens, swimming pools and the like. Over [. . .] customers acquired more than one cabinet under the disposal scheme. HB concentrated its disposals in what it describes as the 'small-shop sector`. According to HB, the total turnover in impulse ice cream of all those outlets which have acquired cabinets in this way is approximately IEP [. . .].

(c) Recent cabinet installation and product proliferation

(74) In 1994, HB intensified its modernisation of its freezer-cabinet stock, replacing many of the older cabinet models (26). In 1993, [. . .] new cabinets were installed in outlets; in 1994, [. . .]; in 1995, [. . .]; in 1996 to August, [. . .]. In general, the new cabinets take up more floor space in retail outlets (though they are, in some instances, smaller in volume), not only in that they are larger in area but also in that many are 'island-type` cabinets which should not adjoin other store fixtures or the wall of the store. HB talks of changing patterns in its cabinet fleet, which it identifies as the 'increasing popularity of the island cabinet model, largely at the expense of visitop and closed top models`, 'a gradual increase in the proportion of cabinets with a length in excess of 1 m`, and 'a gradual reduction in the overall size of cabinets taking into account length, breadth and height`. According to HB, these patterns are explained by 'two principal factors`: '(i) the decline in the proportion of total outlets represented by TSNs and small independent grocers means that the proportion of the total cabinet fleet represented by the 1 m visitop typical of such outlets has shown a corresponding decrease; and (ii) the introduction by manufacturers of new cabinets designed to be more user friendly for the self- serving customer and, to that end, longer and shallower`.

(75) Mars points out that 'Unilever's general product proliferation, and range expansion has the aim of filling these freezers . . . so that retailers can be persuaded that they have a 'full range` of ice-cream products and therefore have no need to sell competing manufacturers' products`.

(d) Pricing arrangements

(76) Until March 1995, HB supplied its ice-cream products and freezer cabinets at an 'inclusive` price. This price included the capital costs of the cabinet, its maintenance costs and the value of the ice-cream products. All retailers were charged the same price irrespective of the ownership of the cabinets in which the products are stocked. Thus, a retail outlet with its own freezer cabinet paid the same price for HB ice cream as a retailer accepting the offer of an HB freezer. In the 1993 Statement of Objections (see paragraph 5), the Commission reached the provisional conclusion that this pricing policy infringed Article 86 of the EC Treaty in that it discriminated against retailers who had not taken an HB freezer cabinet, but who none the less purchased HB ice cream. The Commission provisionally concluded that the inclusive pricing policy not only served as an inducement to grant exclusivity, to the detriment of competing suppliers of impulse ice cream, but that it also gave rise to discrimination between trading partners, by treating dissimilar situations in a similar fashion. Retailers with their own freezer cabinets effectively paid for a service which they did not receive and, in so doing, were forced to subsidise cabinet provision to those taking HB cabinets; the former retailers thereby placed themselves at a competitive disadvantage vis-à-vis the latter ones.

(77) In view of the Commission's objection, HB abandoned the policy of 'inclusive pricing` in 1995, and introduced a 'differential` or 'dual` pricing scheme. The scheme allows for the payment of a lump sum to retailers stocking HB ice cream but not taking an HB freezer cabinet, provided that retailer achieves a minimum annual turnover in HB ice cream (sales from the outlet) of IEP 650 GSV (gross sales value). This lump sum is currently set at IEP 78 p.a. and reflects the purchase and maintenance cost savings to HB in not supplying and servicing a standard 1 m cabinet to the retailer. The retail turnover threshold is set at a level which, according to HB, allows it to secure a minimum return on the sale of ice cream to a retailer.

(78) Unilever stated that the new scheme was 'designed to remove any economic incentive on the retailer to take the HB cabinet`. Unilever went on to say that the scheme would 'have a significant effect on the alleged foreclosure of the market, as more retailers realise that the bonus can fund the purchase of their own freezer`.

(79) The scheme has been operational since the beginning of 1995 and its terms are published on HB price lists; for 1995, retailers holding their own cabinets for a part year were to receive a part bonus. The lump sum can be periodically adjusted to reflect any changes in HB's costs. Similarly, the turnover threshold may also be varied from time to time. All variations are to be carried out in a transparent manner. In March/April 1996, [. . .] retailers qualified for, and received, the dual pricing bonus relating to 1995 purchases of impulse ice cream from HB. In 1997, [. . .] retailers qualified for, and received, the dual pricing bonus relating to 1996 purchases of impulse ice cream from HB.

7. Other manufacturers' distribution arrangements for impulse ice cream in Ireland

(i) Mars and Valley

(80) In 1989, Mars appointed Valley to distribute its newly developed ice-cream products. Prior to the granting of an injunction against Mars by the Irish High Court in 1991 (see paragraph 3), Mars products had also been placed by some retailers in HB-owned cabinets. Initially, Mars supplied Valley who then supplied the Mars products into Valley or Mars-owned freezer cabinets, as well as into cabinets jointly purchased by Mars and Valley. Mars has, however, been doing some direct distribution since 1993. Following the 1992 High Court judgment rendering permanent the injunction against Mars, Mars and Valley amended their cabinet agreements to make them supplier-exclusive. Mars states that its 'total expenditure on freezers alone in 1994 was 22 % of its turnover in impulse ice cream` in that year. In March 1997, Mars terminated its distribution arrangement with Valley.

(ii) Nestlé

(81) Since Nestlé entered the Irish impulse ice-cream market in 1994, it has been installing freezer cabinets in retail outlets subject to a condition of exclusivity. Nestlé distributes its ice cream in Ireland solely through distributors. On entering the ice-cream market in 1994, Nestlé entered into a cooperation arrangement with Leadmore regarding distribution; this was not continued beyond the first ice-cream season.

(iii) Dale Farm

(82) Dale Farm provides freezer cabinets to retail outlets on free-on-loan terms, for the exclusive storage of Dale Farm ice cream, Mars ice cream and other Northern Foods frozen food products.

(iv) Häagen Dazs

(83) Häagen Dazs installs freezer cabinets in retail outlets for the storage of its impulse and other ice-cream products; the cabinets are mainly small and are subject to a condition of exclusivity. All of its ice-cream products are sold through distributors.

(v) Leadmore

(84) Leadmore supplies freezer cabinets to retail outlets subject to a condition of exclusivity.

8. Market research and quantification of foreclosure

(85) A general point should first be made in relation to market research: there is always a certain margin of error when data are obtained by surveying only a sample, however carefully chosen for its representativeness, of the total market in question. It is, however, the only effective and affordable means of obtaining many kinds of market information. The reliability of such data can, moreover, be significantly enhanced where more than one survey of a particular market tend substantially to corroborate each other's findings, as is the case with the three market research surveys described below.

(86) While HB acknowledges a broad correlation between the results of the Lansdowne and B& A surveys, it criticises the representativeness of the Rosslyn survey for being 'based on revisiting outlets surveyed some five years previously` in a similar survey, which original survey was allegedly 'based on a random sample with no attempt at stratification` according to outlet type and size. The Rosslyn survey, according to HB, is 'disproportionately tilted towards TSNs at the expense of independent grocers`, in contrast with the Lansdowne and B& A surveys, which are stratified in accordance with the Nielsen Universe of outlets. It is also the only one of the three surveys to have been conducted before the disposal by HB of some of its freezer cabinets under the 'kick-start` scheme. The Commission has taken due note of these criticisms in any reliance which it has placed on that survey. The Commission has also noted that the B& A survey is weighted significantly more toward large outlets than the other two surveys.

(i) The Lansdowne survey

(87) The survey was carried out on behalf of the Commission by Lansdowne Market Research Limited in July/August 1996, the height of the ice-cream season. A random sample of 501 outlets was chosen, and it was ensured that this would be representative of the relevant retail trade (27). A geographical spread, including a clear urban/rural divide was ensured. The data was acquired by face-to-face interviews with owners/managers in all outlets. The survey's principal findings are detailed below.

(88) Of all cabinets in the outlets surveyed 84 % were owned by an ice-cream manufacturer/supplier: 61 % of all supplier-provided cabinets were HB-owned (28); 11 % of them were Mars-owned; 9 % were Valley-owned; 8 % were Nestlé-owned; 4 % Dale Farm- owned (29); 2 % Leadmore-owned; 1 % Häagen Dazs-owned.

(89) Of all cabinets 12 % were found to be retailer-owned; 2 % were on hire purchase from an ice-cream manufacturer; 1 % were owned by a retail group. 38 % of retailer-owned cabinets were purchased from HB or HB agents.

(90) The proportion of outlets with only one freezer cabinet for impulse ice cream was 58 %; the proportion with two such freezers was 35 %; the proportion with three or more was 7 %. The average (mean) number of cabinets per outlet was 1,50.

(91) Of all outlets selling impulse ice cream 85 % had at least one supplier-provided cabinet; 72 % (30) had at least one HB cabinet; 14 % had at least one Mars cabinet; 10 % Nestlé; 10 % Valley; 5 % (31) Dale Farm; 2 % Leadmore; 1 % Häagen Dazs; 5 % other suppliers. Of all outlets 17 % had at least one cabinet owned by the retailer himself.

(92) Of all outlets 56 % had a cabinet, or cabinets, from one manufacturer/supplier only (and are consequently only in a position to sell the ice-cream products of that single manufacturersupplier); 27 % had cabinets owned by more than one manufacturer/supplier (and no retailer-owned cabinets): consequently, 83 % of all outlets have only manufacturer/supplier-provided cabinets; 17 % of all outlets had at least one cabinet owned by the retailer himself, this 17 % being composed of 12 % of all outlets which only have a retailer-owned cabinet and 5 % of all outlets which have a retailer-owned cabinet/s, as well as a manufacturer/supplier-owned cabinet/s.

(93) This information was also broken down according to the turnover of each outlet surveyed in impulse ice cream. There were three categories: (i) outlets with a turnover of under IEP 1 000 (22 % of the outlets), (ii) outlets with a turnover of between IEP 1 000 and IEP 2 000 (23 % of the outlets), and (iii) outlets with a turnover of IEP 2 000 or more (40 % of the outlets) (32). 28 % of outlets in category (i) had at least one cabinet owned by the retailer himself (22 % only had a retailer-owned cabinet/s, 6 % had a retailer-owned cabinet/s as well as a manufacturer-supplier-owned cabinet/s); 19 % of outlets in category (ii) had at least one cabinet owned by the retailer himself (15 % only had a retailer-owned cabinet/s, 4 % had a retailer-owned cabinet/s as well as a manufacturersupplier-owned cabinet/s); however, only 10 % of outlets in category (iii) had at least one cabinet owned by the retailer himself (4 % only had a retailer-owned cabinet/s, 6 % had a retailer-owned cabinet/s as well as a manufacturer/supplier-owned cabinet/s). 58 % of outlets in categories (i) and (ii) had a cabinet/s from one manufacturer/supplier only (and are consequently only in a position to sell the ice-cream products of that single manufacturer/supplier); 53 % of outlets in category (iii) are only in a position to sell the ice-cream products of that single manufacturersupplier.

(94) 95 % of HB cabinets are set up in outlets which have only got manufacturer-supplied cabinets; 58 % of HB's cabinets are in outlets where only HB's cabinets are in place. The level of sales of impulse ice cream in outlets in the latter category of outlet is almost identical to that in any outlets with HB cabinets.

(95) 47 % of all outlets are only in a position to sell the ice-cream products of a single manufacturersupplier, with a single cabinet only in each outlet; 8 % are only in a position to sell the ice-cream products of a single manufacturer/supplier, with only two cabinets. 41 % of all outlets are only in a position to sell the ice-cream products of HB, 35 % of all outlets are only in a position to sell the ice-cream products of HB, with one HB cabinet only, 6 % with two or more HB cabinets only. 15 % of outlets are only in a position to sell the ice-cream products of a single manufacturersupplier other than HB. None of these other brands is significant in terms of the number of outlets tied to the sale of its products alone.

(96) The average age of the cabinets surveyed was 3,02 years; for retailer-owned cabinets, the average age was 4,59 years; for HB cabinets, it was 2,92; for Mars cabinets, 2,06; for Valley cabinets, 3,53; for Nestlé cabinets, 1,65. Retailers were unable to date one in six of the cabinets surveyed.

(97) When retailers were asked if it would be a 'viable option` for them to use up further space to install another freezer cabinet for impulse ice cream, 87 % of all outlets said that it would not; 11 % said that it would. Of all outlets 53 % considered that one freezer cabinet would be the optimal number of cabinets to have in the store during the summer (ice cream) season; 36 % said two would be optimal; 8 % said three; 1 % four or more. The average optimal number was 1,57. Lansdowne have commented that 'it is evident that the number of ice-cream cabinets currently in place in retail outlets around the country is perceived as approaching the maximum viable`.

(98) Of the cabinets surveyed 97 % were permanently in place in the store (i.e. they had not just been brought in for the summer season). Of all outlets 87 % said that they never placed additional cabinets in the store in order to cope with the increased demand during the summer; 6 % said they sometimes did; a further 6 % said they always did.

(99) Of all cabinets surveyed 90 % are used exclusively for ice-cream storage (50 % are only used for storing impulse products, 40 % for storing both impulse and non-impulse products); 94 % of all manufacturer-owned cabinets surveyed are used exclusively for ice-cream storage (54 % are only used for storing impulse products, 40 % for storing both impulse and non- impulse products); however only 67 % of all retailer-owned cabinets surveyed are used exclusively for ice-cream storage (26 % are only used for storing impulse products, 41 % for storing both impulse and non-impulse products, 34 % for storing both ice cream and non-ice-cream products).

(100) Those retailers who did not own a freezer cabinet were asked for their reasons for not owning such cabinets: 54 % indicated that it did not make commercial sense; 25 % that the cost was prohibitive; 9 % said they only had demand for one brand of ice cream; 4 % said it was inconvenient.

(101) All retailers who did not own a freezer cabinet or were not hire purchasing a freezer from HB were asked if HB's hire purchase offer interested them. 8 % said it did, while 78 % were not interested by it. 54 % said they were not at all interested; 24 % said they were not very interested; 5 % were indifferent; 6 % were quite interested; 2 % very interested. The overwhelming reason cited for this lack of interest was contentment with current freezer cabinet arrangements.

(102) All outlets only in a position to sell the ice-cream products of HB were asked if there was demand for brands other than HB: 53 % responded that there was no demand; 40 % that there was demand (31 % reported a little demand, 9 % a big demand). The same category of outlets was further asked if they were interested in stocking brands other than HB: 62 % (33) said they were not interested (32 % (34) not at all interested, 30 % (35) not very interested); 9 % were indifferent; 25 % (36) were interested (17 % (37) quite interested, 8 % (38) very interested).

(103) Of those interested in stocking brands other than HB (25 % of 41 % of all outlets, hence 10 % of all outlets), 53 % said they would not be prepared to install an additional cabinet for the storage of such brands; 40 % said they would be prepared to do so. Almost all of those not prepared cited space constraints as the main reason. A majority of those prepared to install a further cabinet expressed a preference for a supplier-exclusive one.

(104) Those interested in stocking brands other than HB were also asked if they would be prepared to replace or exchange ('swap`) one of their existing HB cabinets with one from another ice-cream supplier: 82 % said they were not prepared to do so; 11 % said they were. The principal reasons cited for this unwillingness to replace HB cabinets were the popularity and leadership position of HB and contentment with the present arrangement. The same group were further asked if they would be prepared to replace or exchange one of their existing HB cabinets with a cabinet hired or purchased by the retailer himself: 76 % said they would not be prepared to do so; 18 % said they would. Happiness with the current arrangement and cost considerations were the principal reasons given for the lack of preparedness. Finally, the same group was asked if they would be prepared to exchange an HB cabinet for two smaller ones, one not owned by HB: 49 % said no; 44 % yes.

(ii) The Rosslyn survey

(105) Mars also commissioned a survey of the Irish impulse ice-cream market. The survey was carried out by the London-based firm Rosslyn Research Ltd (39), though the fieldwork was done by the locally-based Irish Marketing Surveys. The Rosslyn survey covers 408 outlets spread geographically throughout Ireland, and with a spread of outlet size and type. Some of the Rosslyn survey's main findings are outlined below, together with some comparative figures from the Rosslyn survey of 1991. On the basis of the results yielded by this survey, Mars further commissioned an economic analysis to be carried out, also discussed below.

(106) The proportion of outlets with only one 'front-of store` freezer cabinet for impulse ice cream was 64 % (40); the proportion with two such freezers was 31 % (41); the proportion with three or more was 4 % (42). The average (mean) number of cabinets per outlet was 1,42 (43).

(107) Of the freezer cabinets 92 % (44) were owned by the manufacturer of the product stocked in it. 6 % (45) of the cabinets were owned by the retailer himself. 64 % (46) of cabinets were owned by HB (47), 14 % (48) by Mars, 4 % by Valley, 2 % by Dale Farm, 1 % by Leadmore.

(108) 50 % of outlets had one HB freezer only; 5 % had two HB freezers only; 2 % had one Mars freezer only; 1 % each had one Valley, Dale Farm and Leadmore freezer only; 14 % had one HB and one Mars freezer; 7 % had an HB freezer and one from any other manufacturer except Mars.

(109) Of the manufacturer-supplied freezers 97 % were on loan from that manufacturer; of these, 99 % were provided free of charge, and 88 % were provided subject to an exclusivity condition (49). Of retailers supplied with freezers by manufacturers 27 % (50) said they would like to stock other manufacturers' products in the freezers (68 % (51) were not interested).

(110) Of retailers with manufacturer-exclusive cabinets 87 % would have to install another freezer if they wished to sell another brand of ice cream; of these, 67 % said they did not 'have room` to do so. Of those who said they did have room, 15 % said they had chosen not to install one because it 'would not be economical in terms of running costs` and 13 % because they had 'better use for the space`; 43 % of those with room said they would be prepared to install another cabinet 'if the demand was high for another product`.

(111) Of those freezer cabinets replaced (returned, exchanged or changed) by another freezer from the same or different ice-cream manufacturer over the previous five years, 78 % of them belonged to HB; 55 % of the freezers were returned, exchanged or changed in order to be replaced by a more modern freezer; a further 25 % of them were returned, exchanged or changed in order to be replaced by a larger freezer; 74 % of the new freezers belong to HB; freezers returned, exchanged or changed were almost invariably replaced by freezers from the same manufacturer - 214 of the 234 freezers replaced were HB freezers replacing HB freezers.

(112) Of all outlets 72 % said they would not be prepared to install a replacement or additional freezer; of those who would be prepared to do so (26 % of the total sample), 82 % would not be prepared to buy a cabinet themselves; 94 % of them would not be prepared to take a cabinet subject to exclusivity on a loan reflecting the costs of cabinet provision; 80 % would not take a non-exclusive cabinet on rent; 84 % would, however, be prepared to take a free on loan cabinet on supplier-exclusive terms. When the same 26 % of the sample were asked if they could install a further cabinet (in addition to another supplier-exclusive one), 77 % said they could not; of these, 51 % cited lack of space and better use of space as reasons, while a further 22 % said it would not be economical in terms of running costs. Of those who said they could (20 % of 26 % of total sample), only 17 % were prepared to do so.

(113) The survey was carried out in mid-April 1996. Only 5 % of retailers had been offered a freezer under the 'kick-start` disposal (see paragraphs 72 and 73). Only 13 % of retailers were aware of the existence of the HB hire-purchase scheme.

(114) The economic analysis was carried out by the firm of economists, Case Associates, and is entitled 'Estimating foreclosure levels in the Irish ice-cream market`. This analysis has drawn a number of conclusions from the survey. The analysis purports, inter alia, to calculate what proportion of all outlets are 'foreclosed` in the sense that a new cabinet would have to be installed in order for a new entrant to be able to sell its ice cream there. It concludes, by one means of calculation, that at least 89 % (52) of all outlets selling impulse ice cream are foreclosed in this way. Using an alternative means of calculation, the analysis arrives at a foreclosure level (in the sense just described) of 84 % (53). Yet another means of calculation puts foreclosure at the higher level of 93 % (54).

(iii) The B& A surveys

(115) HB also commissioned a market survey, which was conducted in August/September 1996 by Behaviour & Attitudes Limited, a market research firm. The results of this survey were contained in HB's written response to the Statement of Objections. The sample for the survey comprised 507 retail outlets, spread geographically throughout Ireland, and with a spread of outlet size and type (55). The survey's principal findings are outlined below.

(116) [. . .] % of outlets surveyed had one cabinet in front of the store, [. . .] % two cabinets, and [. . .] % three or more. HB products were sold from [. . .] % of the outlets surveyed, Mars from [. . .] %, Valley from [. . .] %, Nestlé from [. . .] %, and Häagen Dazs from [. . .] %. [. . .] % of all the outlets surveyed stocked HB ice-cream products together with the products of no other ice-cream manufacturer. HB faces direct competition from other ice-cream brands in [. . .] % of the outlets surveyed, less than[. . .] % of the outlets in which its products are sold.

(117) Of those outlets with only one cabinet, [. . .] % of these were HB cabinets, [. . .] % were Mars cabinets, [. . .] % Valley cabinets, [. . .] % Nestlé cabinets and were [. . .] % retailer cabinets. Of those with two or more cabinets, [. . .] % had at least one HB cabinet, [. . .] % a Mars cabinet, [. . .] % a Valley cabinet, [. . .] % a Nestlé cabinet, [. . .] % a Häagen Dazs cabinet, and [. . .] % a retailer cabinet. [. . .] % of all outlets surveyed had a single HB cabinet alone in front of store.

(118) B& A asked retailers how often they were asked for brands not currently stocked. [. . .] % replied that they were never asked, [. . .] % said rarely, [. . .] % occasionally and [. . .] % frequently. Of those who were at some stage asked for non-stocked brands, [. . .] % took no action, [. . .] % 'added a manufacturer`, and none 'changed manufacturer`. The survey found that, over the previous seven years, [. . .] % of the outlets had at some stage stocked new brands of ice cream, and that [. . .] % of outlets had at some stage de-stocked brands of ice cream; the principal reasons cited for such de-stocking were lack of customer demand ([. . .] %), poor delivery standards ([. . .] %), and poor cabinet maintenance ([. . .] %).

(119) Retailers with only one cabinet in their outlet were asked if they were prepared to install a second cabinet, [. . .] % were not, [. . .] % were, and the remainder did not know. The main reason cited for this absence of preparedness was lack of space. The same outlets were also asked about their preparedness to replace the existing cabinet, [. . .] % were not prepared, [. . .] % were and the remainder did not know. The principal reasons given for such lack of preparedness were satisfaction with the current supplier's product range and cabinet. The same retailers were further asked if they would be prepared to replace the existing freezer with two smaller ones, [. . .] % were not, [. . .] % were and the remainder did not know.

(120) Retailers were asked whether, in the event that manufacturers no longer imposed freezer exclusivity and instead charged a separate cabinet rental, they would stock a wider range as a result. [. . .] % said they would stock a wider range; [. . .] % said they would not; [. . .] % said they would stock the same range; [. . .] % said they would cease stocking altogether. The [. . .] % who expected to stock a wider range were further asked if they would expect to buy/lease a cabinet rather than paying the rental, [. . .] % said they would, [. . .] % that they would not. Finally, the B& A survey registered very high levels of retailer satisfaction with the product range, delivery arrangements and maintenance service of all ice-cream suppliers to the outlets surveyed.

(121) HB also submitted, together with its response to the Statement of Objections, another B& A survey, conducted in 1994, and entitled 'New HB cabinets: the views of retailers`. This survey was conducted with the purpose of obtaining retailers' and consumers' views on newly- installed HB cabinets. The survey was accordingly targeted towards larger outlets and, as such, its usefulness for the purposes of this Decision is somewhat limited. The survey found, inter alia, that HB was rated more highly by retailers than other ice-cream suppliers in terms of its performance under a number of different headings. It also found that retailers were more likely than not to dislike the idea of accepting the installation of two HB cabinets in an outlet, the principal reasons for this being lack of space and cost.

(122) HB furthermore submitted, together with its response to the Statement of Objections, a 'Qualitative study` entitled 'Retailers' stocking policy`, carried out by B& A for HB in March/April 1997 on the basis of 25 interviews. The purpose of the research was 'to gain some qualitative insight into how retailers manage their stores`, though accepting that the findings could only be viewed as 'broad indicators`. These findings include the following.

(123) Most of the interviewed retailers 'do not go so far as to work out profit per square foot calculations`; rather, 'they work to their own instinctive estimate of that calculation`. There was also some concern expressed about the fact that nothing can be stocked above ice-cream cabinets; this means that 'they tend to be placed in the middle of the floor`, which 'obviously leads to some concern about underutilisation of space`. Retailers also indicated that re- stocking a freezer 'more frequently to maximise throughput during the summer months` was preferable to 'putting in extra fridges to cope with peak summer demands for ice cream`. The study also found that 'space reallocation` in order to meet 'changing consumer demands` was 'not on a major scale because the stores themselves are, in the main, relatively small`.

(iv) Other sources

(124) HB estimates (1997) that [. . .] outlets are supplied with one 'front-of-store` freezer cabinet only by HB, that almost [. . .] outlets are supplied with more than one such cabinet and that, of that total of [. . .], approximately [. . .] stock exclusively HB impulse products. On the basis of HB's estimate of [. . .] for the total number of outlets (see paragraph 41), [. . . more than 14 % . . .] of all retail outlets in Ireland are in a position to sell only HB's impulse ice-cream products.

(125) In a paper entitled 'Sales of cabinets in situ` presented to the Commission by HB on 29 November 1994, HB stated that it 'has [. . .] outlets including agents outlets. Of these, [. . .] (or [. . .] %) are shared with other suppliers. Total Universe (Nielsen) is 9 669 outlets`. On this basis, [. . .] outlets (or [. . .] % of HB's outlets) were HB-only outlets at that time; the [. . .] HB-only outlets amounts to [. . . more than 60 % . . .] of all outlets (on the basis of the Nielsen Universe).

9. The ice-cream market in Northern Ireland

(126) The structure of the ice-cream industry in Northern Ireland is substantially different from that in Ireland. Whereas HB is the only Unilever-owned ice-cream producer operating in Ireland, in Northern Ireland both HB and Unilever's United Kingdom ice-cream subsidiary, Walls Ice cream Ltd, are both engaged in the supply of ice cream. Walls handles the multiple trade, while the traditional trade is now principally handled by HB. Dale Farm was until recently the leading supplier of ice-cream products in Northern Ireland (56) but, since 1996, HBWalls have assumed that leadership position. Nevertheless, consumer brand awareness varies considerably between Ireland and Northern Ireland, as is demonstrated by the fact that Dale Farm continues to hold a market share of some 40 % in value terms (in contrast with its negligible share of the Irish market, see paragraph 36); prices for similar products also differ to some extent.

(127) Whereas in Ireland HB usually distributes its ice-cream products direct to retailers, in Northern Ireland it only uses agents to distribute its products. With regard to freezer cabinet arrangements in Northern Ireland, HB charges retailers an annual rent for the loan of its freezer cabinets, and applies a price reduction to products which it supplies to retailers with their own cabinets.

(128) Regulations governing the permitted content of ice cream differ between Ireland and Northern Ireland. In Ireland, ice cream is required to have a minimum butterfat content of 5 % whereas UK legislation contains no such requirement. VAT rates also vary somewhat.

III. LEGAL ASSESSMENT

(129) The object of this legal assessment is to determine whether the arrangements made by HB for the distribution of its impulse ice-cream products in Ireland, whereby it pursues a policy under which it makes available to outlets stocking its products freezer cabinets on the basis of cabinet exclusivity, are compatible with Articles 85 and 86 of the EC Treaty. The assessment of the compatibility with Article 85 of the exclusivity agreements in question is made in the context of the legal and economic circumstances in which they are found (57).

A. ARTICLE 85(1)

1. The relevant market

(i) The relevant product market

(130) The relevant product market includes, in principle, all goods which are perceived by the consumer, on the grounds of their characteristics, price or intended purpose, as being reasonably interchangeable with each other (58). Using these criteria, a distinction should in the first instance be drawn, in accordance with how ice cream is manufactured and distributed, between industrial and 'artisan` ice cream (see paragraph 10); only industrial ice cream is found to any significant extent in Ireland. Secondly, a distinction should be made between the following three ice-cream (59) product categories (60), determined in accordance with the place of consumption: impulse ice cream, take-home ice cream, and catering ice cream (see paragraph 11).

(131) Ice cream which is offered to the consumer as part of catering services forms a distinct product market (61). This market is essentially (62) made up of industrial ice cream for bulk- buying customers, which they may serve, for example, as desserts; value is often added to the ice cream by the caterer.

(132) The inherent nature of the product dictates that the place of consumption is crucial to the determination of the relevant market. Owing to the need for ice cream to be maintained at a low temperature, a consumer purchasing such a product in a retail outlet has in essence two options for consumption of the product: the ice cream purchased by the consumer may either be consumed immediately at or near the place of sale in order to satisfy an impulse, or be stored as soon as possible in a freezer cabinet, probably in the consumer's own home, for consumption at a later time. This distinction on the basis of the consumer's intended purpose in purchasing the ice cream in turn determines the differences in characteristics and price between impulse and take-home products. Take-home products usually take the form of blocks or cakes of ice cream, as well as multipacks of single items, whereas impulse products are invariably offered as single items. It is clear, therefore, that impulse and take-home ice cream form separate and distinct product markets (63).

(133) From the consumer's point of view single wrapped items may be perceived as being reasonably interchangeable with individual portions of scooping ice cream and with soft ice cream. In the case of soft ice cream, this may be a matter of dispute, given the non-dairy nature of its contents. The consumer's point of view is, however, not in every instance the sole criterion in the determination of a product market; nor is an examination limited only to the objective characteristics of the products in question sufficient. The competitive conditions and the structure of supply and demand on the market must also be taken into consideration (64). In the case of impulse ice cream, a further distinction is therefore necessary because of the different conditions applying to the production and distribution of these categories of impulse ice cream. On account of the marginal importance of scooping ice cream in Ireland, its inclusion in the relevant product market does not need to be considered. Soft ice cream, on the other hand, is quite widely sold in Ireland.

(134) Single wrapped items of ice cream are purchased by the retail trade in the form in which they are resold to the consumer. They are specifically designed for consumer self-service (see paragraph 12). In the case of soft ice cream, value is added by the retailer who must further process the ambient soft ice cream mix before the final product can be served to the consumer. The added-value thus created by the retailer is reflected in generally higher trading margins for soft ice cream than in the case of single wrapped items of ice cream. Moreover, soft ice cream requires installations which are different from the freezer cabinets in which single wrapped items of ice cream are stored. For soft ice cream, special processing machinery must be installed in the outlet for the processing and dispensing of the product. The allocation of particular staff (for customer service and maintenance) and space resources are also involved.

(135) As a result of the foregoing differences in the conditions of distribution for soft and wrapped single items of ice cream, the competitive conditions under which they are offered to the retail trade are different and distinct. As regards production, the differing characteristics of the products (soft ice-cream mix as opposed to ice cream) and hence of the technology used in their manufacture provide a further impediment to substitution on the supply side. Wrapped items are, moreover, characterised by a high level of branding whereas, in general, soft ice cream carries no brand. These supply-side differences are illustrated by the fact that the producers of the two categories of ice cream differ (see paragraph 38): only HB produces both types, and even then its share of the market for soft ice cream is much smaller than for wrapped items (see paragraphs 27 and 38). They therefore form separate product markets.

(136) Even if the two categories just described were regarded as forming a single product market, this would not affect the present assessment in any material way. The reason for this is that soft ice cream is of relatively minor importance compared to the value of the impulse ice cream market as a whole. As a result, the positions of the various producers on the market would not be substantially affected. Indeed, of the producers of wrapped items, HB's position would be the least weakened by the addition of its share of the soft market (see paragraph 38).

(137) The competitive conditions in non-retail channels (such as schools, company canteens, sports and leisure venues) often have characteristics which are distinct from those in normal retail outlets, most particularly the fact that the products are in many instances not being sold to the general public but rather to a captive pool of consumers. Considerations such as space management may often be quite different in such venues, and price levels can also vary. Impulse sales via such non-retail venues are, however, relatively insignificant, and it is therefore not necessary to decide whether ice cream sold through such outlets should be excluded from the relevant market. Only 4 %, for example, of HB's cabinet agreements relate to cabinets installed in non-retail outlets of this kind (see paragraph 60).

(138) The product market that is relevant to the present proceeding therefore comprises single wrapped items of impulse ice cream. Hereafter, all references to 'impulse ice cream` shall only include such single wrapped items.

(ii) The relevant geographic market

(139) The objective competitive conditions of supply and demand for impulse ice cream vary considerably in the different parts of the Community. Although the production of industrial ice cream has shown a clear trend towards internationalisation, distribution is still largely organised on a national basis. Distribution arrangements, including freezer-cabinet agreements and analogous contracts granting exclusivity, are concluded at national level; this is also true for Ireland. National peculiarities are reflected in varying market structures, product ranges and prices. Consumer preferences for product types and brands can also vary; this is partly due to the fact that some ice-cream manufacturers operating in several Member States have a policy of using different brand names in different countries (paragraph 20), a clear illustration of the national character of markets. Moreover, the regulations governing the manufacture of ice cream are not harmonised at European level.

(140) In particular, the impulse ice cream market in Ireland differs from that in Northern Ireland in several important respects, as has been described in paragraphs 126, 127 and 128. The conditions of competition are therefore not sufficiently homogeneous for manufacturers and suppliers to treat Ireland and Northern Ireland as constituting a single market. The relevant geographic market is consequently Ireland (65).

(iii) The position of HB on the relevant market

(141) The position of HB on the market for single wrapped items of impulse ice cream in Ireland is one of particular strength, as is exhibited, inter alia, by its market share over many years (see paragraphs 28 and 255 et seq.). This position of strength is further illustrated by the degree of both numeric and weighted distribution enjoyed by HB's impulse products (see paragraph 46), as well as by the strength of the brand and the breadth and popularity of its range of products (see paragraphs 24 and 25). HB's position on the impulse market is further reinforced by the strength of Unilever's position, not only on the other ice cream markets in Ireland (take-home and catering), but also in international ice cream markets and in the markets for frozen foods and consumer products generally.

2. Restriction of competition

(142) HB supplies freezer cabinets to retailers on the basis of standard form cabinet agreements containing provisions requiring the cabinets to be used exclusively for the storing of products for sale which are supplied by HB (see paragraph 56); accordingly, no product manufactured or supplied by a third party is to be sold or offered for sale from those freezer cabinets, or stored therein. These contracts between HB and individual retailers constitute agreements between undertakings within the meaning of Article 85(1). HB has concluded a network of such freezer-cabinet agreements, relating to cabinets installed in outlets throughout the relevant geographic market.

(143) These contractual provisions have as their effect the restriction of the ability of the contracting retailers to stock and offer for sale in their outlets impulse products from competing suppliers, in circumstances where the only freezer cabinet, or cabinets, for the storage of impulse ice cream in place in the outlet has, or have, been provided by HB, where the incumbent HB freezer cabinet/s is/are unlikely to be replaced by a retailer-owned or by a competitor's freezer cabinet/s, and where it is not economically viable to allocate space to the installation of an additional cabinet. This restriction has the consequence that those competing suppliers are precluded from selling their products to those outlets, thereby restricting competition between suppliers in the relevant market.

(144) Having identified the origin of the restriction of competition brought about by the network of HB's cabinet agreements (paragraph 142), and the circumstances in which it produces restrictive effects on retailers and suppliers in the relevant market within the meaning of Article 85(1) (paragraph 143), the existence of these circumstances falls to be demonstrated. In so doing, consideration has not been given to the restrictive effect of each individual cabinet agreement, but rather to the effect produced by the category of agreements fulfilling the stated conditions and constituting an identifiable part of the network of HB's cabinet agreements as a whole (66); the assessment of the restrictive effect of that part of HB's network then applies equally to each of the agreements comprising that part (67).

(145) The assessment of this restrictive effect has been made against the background of the effect of all similar networks of freezer-cabinet agreements operated by the other ice-cream suppliers in the relevant market, as well as in the light of any further relevant market conditions. The restrictive effect has moreover been quantified so as to demonstrate its appreciable nature.

3. Outlet profile

(146) In determining the precise extent to which Article 85(1) applies to HB's cabinet agreements, it is first necessary to analyse the outlet profile in order to identify those outlets in which only an HB cabinet/s is/are found.

(147) Networks of supplier-exclusive cabinet agreements (that is, agreements whereby a cabinet is provided by a supplier on condition that only products supplied by him may be stored therein) clearly have the effect of restricting the ability of retailers to stock and sell the impulse ice-cream products of a number of competing suppliers. This restriction results from the inevitable space constraints experienced by retail outlets (see paragraph 43). The average number of cabinets in place in outlets has been calculated at 1,5 by the Lansdowne survey and at 1,42 according to the Rosslyn survey, clear illustration of these constraints (see paragraphs 89 and 105). Retailers are, moreover, conscious of these constraints, as can be seen from their perception of the optimal number of freezer cabinets to have in place in an outlet at the height of the season: 1,57 (see paragraph 97).

(148) Only a small proportion of retail outlets in Ireland contains non-exclusive cabinets (68): these can be referred to as 'open` outlets, in the sense that retailers are free to stock the impulse ice cream of any supplier. The Lansdowne survey, conducted in July 1996 (69), found (see paragraph 92) that only 17 % of all outlets had retailer-owned cabinets and thus could be described as 'open` in this way, and that the remaining 83 % had only supplier- provided cabinets. In April 1996 (70), the Rosslyn survey found (see paragraph 114) that no more than 11 % of outlets could have in place at least one cabinet which was not supplier- exclusive; only 6 % of all cabinets in the outlets surveyed were retailer-owned. These are the only outlets to which impulse ice-cream suppliers can have direct access, without offering the retailer a freezer cabinet for the storage of the new supplier's products, and without persuading the retailer to accept that offer.

(149) The remaining outlets, without non-exclusive cabinets, contain one or more supplier- exclusive cabinets. The Lansdowne survey found that 83 % of all outlets in the relevant market fell into this category (see paragraph 92). The Rosslyn survey found (see paragraphs 108 and 114) that a total of 84 % of all outlets offering impulse ice cream contain only supplier-exclusive cabinets. These are the outlets to which suppliers whose cabinets are not already set up therein cannot have direct access for the sale of their products without first overcoming the substantial barriers to entry described below. In this way, such newcomers to the outlet are foreclosed therefrom. Although this foreclosure is not absolute, in the sense that the retailer is not contractually precluded from selling other suppliers' products, the outlet can be said to be foreclosed in so far as entry thereto by competing suppliers is rendered very difficult.

(150) It is also clear that most sales of impulse ice cream are made through outlets which have only supplier-exclusive cabinets. This is illustrated, not only by the sheer numerical superiority of such outlets in the relevant market, but also by the much lower proportion of high turnover (in impulse ice cream) outlets which are open in the manner described at paragraph 148: only 10 % of the higher turnover category of outlets are open in this way, in contrast to 28 % of the lower turnover category of outlets (see paragraph 93).

(151) A majority of these foreclosed outlets are, moreover, only in a position to sell the impulse ice-cream products of just one supplier, in the sense that they only have an exclusive cabinet/s provided by one ice-cream supplier: according to Lansdowne, 56 % of all outlets are in this position (see paragraph 92).

(152) The majority of the supplier-provided cabinets in the relevant market have been provided by HB: 61 % of them, according to Lansdowne (see paragraph 88); 64 % of all cabinets, according to Rosslyn (see paragraph 107). According to Lansdowne, 72 % of all outlets have at least one HB cabinet (76 %, according to Rosslyn, see paragraph 108). HB's own estimate (see paragraph 124) is that [. . .] % of all retail outlets in the relevant market have at least one HB cabinet.

(153) The Lansdowne survey also found that 41 % of all outlets are only in a position to sell the impulse ice-cream products of HB, in that they only have a freezer cabinet, or cabinets, provided by HB, 35 % of all outlets having only a single HB cabinet and a further 6 % having more than one. According to Rosslyn, in 50 % (see paragraph 108) of all outlets, an HB cabinet is the only cabinet set up in the outlet; in a further 5 % of all outlets, there is more than one HB cabinet and these are the only cabinets in the outlet: this means that 55 % of the sample of outlets were only in a position to sell the impulse ice-cream products of HB (solus HB stockists) in the sense just described. According to B& A, in [. . .] % ([. . .] % of [. . .] %, see paragraphs 116 and 117) of all outlets, an HB cabinet is the only cabinet in place in the outlet; [. . .] % of the sample of outlets were solus HB stockists in the sense described above (see paragraph 116). [. . . more than 40 % . . .] of all outlets, according to HB's own estimate, are solus HB stockists in this sense (see HB's estimate in November 1994 of [. . . more than 60 % . . .] see paragraph 125).

(154) Almost all of HB's sales of impulse ice cream are made through outlets which have only supplier-exclusive cabinets. This can be seen from the fact that [. . . more than 95 % . . .] of HB's ice cream sales are made through outlets in which HB cabinets are in place (see paragraph 61). Given that only 5 % of all outlets are open and also have a supplier-exclusive cabinet/s (see paragraph 92), and that 95 % of HB cabinets are in place in outlets which have only such supplier-exclusive cabinets (see paragraph 94), the conclusion that almost all of HB's sales are made through such outlets is unavoidable (71).

(155) The Lansdowne survey demonstrates that the level of HB's sales through outlets with only an HB cabinet/s, that is to say, those which are solus HB stockists in the manner described above, is almost identical to the level of impulse ice-cream sales made through other outlets in which HB has provided a freezer cabinet for installation (see paragraph 94). Moreover, HB's levels of numeric and weighted distribution illustrate that the level of sales in the outlets where it is present is generally higher than in those where it is not (see paragraph 46). The level of HB's sales through outlets with only an HB cabinet/s is therefore at least as high as that made through all other outlets in the relevant market.

(156) It can thus be concluded that in some 40 % (72) of all outlets in the relevant market, the only freezer cabinet/s for the storage of impulse ice cream in place in the outlet has/have been provided by HB (see paragraph 153). HB has expressly stated in its response to the Statement of Objections that it does not dispute this conclusion. It has further been demonstrated that some 40 % of all impulse ice cream sales in the relevant market are made through this category of outlets (see paragraph 155). These outlets are solus HB stockists in the sense described above. The agreements relating to the cabinets installed in this category of outlets constitute an identifiable part of HB's network of cabinet agreements.

4. Conditions for finding that the identified part of HB's network of freezer-cabinet agreements produces a restrictive effect on the relevant market

(157) A supplier who wishes to gain access for the sale of his impulse ice-cream products to a retail outlet (that is, a new entrant to the outlet) in which at least one supplier-exclusive freezer cabinet is in place can only do so if that outlet has a non-exclusive cabinet/s (and thus is an open outlet in the sense described at paragraph 148) or if he can persuade the retailer either to replace an in situ supplier-exclusive freezer cabinet or to install an additional freezer cabinet alongside the in situ supplier-exclusive cabinet/s. These two alternatives are discussed at points (i) and (ii) below in relation to the category of outlets identified in paragraph 156.

(i) Likelihood of persuading a retailer to replace an HB freezer cabinet

(158) If a competing supplier wishes to gain access to an outlet in which the only freezer cabinet for the storage of impulse ice cream in place in the outlet has been provided by HB, he may attempt to persuade the retailer to replace one of the HB cabinets with either a cabinet purchased or leased by the retailer himself from a source other than an ice-cream supplier or with a cabinet supplied by the competing supplier (the new entrant to the outlet).

(a) Replacement by retailer's own cabinet

(159) A retailer contemplating the replacement of a supplier-exclusive cabinet will only do so if he expects to increase his profits as a result. In making such a calculation, he must take into account the strength of the position of the incumbent, or incumbents, on the relevant market. A retailer contemplating the replacement of an HB cabinet must therefore take into account the overwhelming strength of HB's position on the relevant market (see paragraph 141). This is particularly so where the HB cabinet is the only one in the outlet: where the retailer envisages replacing this one HB cabinet with his own freezer cabinet, it can be expected that the ice-cream products stocked in the retailer-owned freezer cabinet will in general reflect the respective market shares of the various different manufacturers in the relevant market. Given the strength of HB's position, the range of products offered in the outlet following the replacement is not likely to change to any great extent. The retailer is not therefore likely to envisage that his profits will be much enhanced by the possibility of offering a combination of manufacturers' products.

(160) Nor should the risk-aversion of retailers be discounted as a factor rendering it unlikely that a retailer will choose to acquire his own cabinet: uncertainty about future profits, as well as the general inconvenience of having to acquire and maintain one's own cabinet, amount to not insignificant disincentives, particularly to small retailers. It should be borne in mind that uncertainty concerning future profits is likely to particularly discourage small retailers for whom the risk inherent in such an investment (see below) could have serious consequences and so would be less attractive.

(161) The investment necessary for the purchase of a retailer-owned freezer cabinet, as well as the cost of its maintenance, amounts to a disincentive to retailers who would thereby forego the provision of a cabinet which is provided on 'free on loan` terms, with maintenance assured by the supplier, HB: even though the retailer pays indirectly for the service (see paragraphs 76 to 79), the economies of scale in purchasing, installing and maintaining cabinets which are possible for large-scale ice-cream suppliers such as HB are not available to small retailers and therefore amount to an incentive to accept supplier-provision. Before the introduction of dual pricing in 1995, HB's policy of inclusive pricing (see above) meant that this disincentive was even more marked. As a result of the inclusion of this element in the ice-cream price charged across the board, a retailer with his own cabinet was indirectly financing HB's cabinet fleet without benefiting from the provision of a cabinet. Such an investment is moreover a 'sunk` cost for the retailer if he cannot easily resell the cabinet. Such cost considerations are particularly pertinent in relation to retailers with cash or credit constraints, for whatever reason, and so will be particularly relevant to smaller outlets. The Lansdowne survey found that these commercialcost concerns for retailers were unmistakable disincentives to cabinet purchase by them: see paragraphs 100 and 104.

(162) The need to rupture relations with the supplier providing the cabinet to be replaced is a further impediment, particularly where that supplier enjoys a market position as strong as that held by HB. The very fact that there are so few retailers who have elected to purchase or lease freezer cabinets for the stocking of impulse ice cream in Ireland tends to confirm this analysis. Prior to HB's cabinet disposal in 1996, very few retailers in Ireland had opted to purchase their own cabinets (see Rosslyn survey at paragraph 107, carried out in April 1996, prior to the disposal). Most of the cabinet disposals to retailers by HB in 1996 were made [. . .], and account for a large proportion of the retailer-owned cabinets in the relevant market. HB's cabinet hire-purchase scheme, introduced in 1995, has not attracted the interest of any retailers (see paragraphs 70 and 71) in spite of HB's expressed conviction that its terms would prove attractive to them; only 11 % of those retailers interviewed by Lansdowne who did not already own a cabinet or have one on hire-purchase expressed interest in the HB scheme. The market research carried out by Rosslyn found that, of the 26 % of all outlets interviewed who said they would be prepared to install a replacement or additional freezer, 82 % would not be prepared to buy a cabinet themselves and 80 % would not take a non-exclusive cabinet on rent.

(163) HB have argued that in outlets with more than one HB freezer cabinet, there is a greater likelihood of the retailer replacing one of them with his own cabinet while retaining the other, or others. It should be pointed out that most of the outlets in the relevant category (i.e. those in which there are only HB cabinets) have only a single HB freezer cabinet for the sale of impulse ice cream (see paragraph 153); in most of the remainder there are only two freezer cabinets. For those outlets with two or more HB freezer cabinets, a retailer is likely to replace one or more of them only if he anticipates that the newly installed freezer cabinet will yield at least the same turnover in impulse ice cream as achieved through that, or those, displaced. While he may reach such a conclusion, the evidence indicates that retailers very rarely choose to take this course of action (see below); although the market strength of HB must be the principal reason for this reluctance to replace HB cabinets, it is likely to at least in part be on account of the disincentives to cabinet acquisition just described.

(164) The market research relied on by the Commission testifies to the fact that replacement of supplier-provided cabinets by either retailer-owned cabinets or by the supplier-provided cabinets of the incumbent's competitors is very unusual (see paragraph 111), and that retailer unwillingness to do so is particularly marked (see paragraphs 104, 112 and 119). The market research moreover indicates that replacement of HB-provided cabinets by cabinets not provided by HB is very rare: over the last five years, 78 % of all the cabinets replaced in outlets were HB cabinets, and 74 % of the ones being put in their place were also HB cabinets (see paragraph 111) and, in most cases, larger ones (see paragraph 74).

(165) The research also indicates that retailers are generally uninterested in the possibility of such replacement. The Lansdowne survey specially asked certain questions of retailers who had only an HB cabinet, and who were consequently in a position to offer for sale only HB impulse ice-cream products. These retailers were further asked if they were interested in stocking brands other than HB. Only 25 % of them were. Of those retailers (those with only HB cabinets and an interest in stocking other brands) 76 % said they would not consider replacing the HB cabinet (or one of them) with a cabinet hired or purchased by themselves (see paragraphs 102, 103 and 104).

(b) Replacement by a supplier-provided cabinet

(166) A retailer contemplating the replacement of a supplier-provided freezer cabinet with a cabinet supplied by another ice-cream supplier will only do so if he forecasts that higher profits are likely to result from such a substitution (73): he must in general be satisfied that there is more demand for the new entrant's products and that the newcomer's cabinet will therefore prove to be a better profit-earner than the incumbent's. If the entrant to the outlet is a less well-known ice-cream producer than the supplier, or suppliers, already present, if he has a less extensive or less popular product range, if the cabinet provided is smaller or the terms less attractive, the retailer will have no incentive to make the replacement. Where the supplier-exclusive cabinet is the only one in the outlet, those retailers would effectively give up the option of offering the displaced supplier's impulse ice-cream products altogether, as manufacturers appear almost invariably to supply freezer cabinets subject to a condition of exclusivity.

(167) For those retailers contemplating the replacement of an HB freezer cabinet by a cabinet supplied by another ice-cream supplier, the strength of HB's position on the relevant market amounts to an even greater disincentive than in the case of replacement by the retailer's own cabinet. This is particularly so where the HB cabinet is the only one in the outlet: those retailers would effectively give up the option of offering HB impulse ice-cream products altogether (74). The retailer will only choose this option if he forecasts that higher profits are likely to result from such a substitution and so is almost certain not to do so where the HB cabinet is the only one in the outlet. None of HB's competitors' products are as well known or as comprehensive in range. As a result, they do not in general provide a satisfactory alternative, from a retailer's point of view, to HB's products.

(168) The risk aversion of many retailers, inconvenience, and established relations with current supplies are also factors tending to make it more unlikely that a retailer will choose to replace one supplier's cabinet with another's. Where an existing, supplier's cabinet is replaced, not by the retailer's own cabinet but by the cabinet of another supplier, this amounts almost invariably to an effective refusal to deal further with the incumbent. These considerations are all the more pertinent in relation to the replacement of an HB cabinet, in view of that company's market strength. Even where an outlet has two or more HB cabinets only (meaning that it is a solus HB stockist), the replacement of one or other of these HB cabinets is unlikely, given the market strength enjoyed by HB: the likelihood of replacement is conditioned by the retailer's expectation of a higher return from the new cabinet. The market research described above confirms the unlikelihood of such replacement.

(169) The question of whether an HB freezer cabinet is likely to be replaced by a freezer cabinet supplied by a competitor has also to be considered from the point of view of that other supplier. As mentioned above, such a supplier can only be expected to make the investment necessary if he forecasts a satisfactory return on that investment in the form of increased sales. Given HB's strength in the relevant market, an ice-cream supplier other than HB is likely to have a lower return on such an investment. This competitive disadvantage vis-à-vis the leading supplier in the market makes it in some instances less likely that a retailer will be offered a cabinet by such a competitor.

(170) It should be recalled that the empirical evidence, in the form of the market research relied on by the Commission, indicates that replacement of supplier-provided cabinets generally, and of HB cabinets in particular, almost invariably results in the installation of new cabinets coming from the same supplier (see paragraph 111). Of the retailers surveyed by Lansdowne who had only an HB cabinet/s and expressed an interest in stocking other brands 82 % said they would not consider replacing the HB cabinet (or one of them) with a cabinet supplied by another ice-cream supplier (see paragraph 104), the leadership position and popularity of HB's products being cited by many retailers as the principal reasons.

(171) The HB freezer-cabinet agreements are concluded for an indefinite period of time. Each such agreement continues in force until either HB or the retailer gives the other two calendar months' notice in writing of its desire to terminate it (75). In spite of the fact that retailers are generally aware that the agreements can be terminated at short notice, the indefinite nature of the agreements is none the less another factor which may render retailers reluctant to take the initiative of bringing them to an end, whether with a view to replacement of an HB cabinet by a retailer-owned cabinet or by that of another supplier.

(ii) Likelihood of persuading a retailer to install an additional freezer cabinet

(172) Alternatively, a competing supplier wishing to gain access to an outlet in which the only freezer cabinet/s for the storage of impulse ice cream in place in the outlet has/have been provided by HB, may attempt to persuade the retailer to install alongside the in situ HB cabinet/s, either an additional cabinet purchased or leased by the retailer himself from a source other than an ice-cream supplier or a cabinet supplied by the competing supplier (the new entrant to the outlet).

(173) In deciding whether to install an additional cabinet in his outlet, a retailer must take into account the constraints on space which are inherent in any outlet, and in outlets in the relevant market in particular (see paragraph 43). The retailer must be convinced of the economic viability of such a decision. In general, a retailer allocates sales space in accordance with his perception of the return potential of that space (76). In particular, he will take a variety of factors into account, some of which are detailed below. The Lansdowne survey found that 87 % of the outlets surveyed did not consider that it would be a 'viable option` for them to attribute additional space to the installation of a further ice-cream cabinet (see paragraph 96). The market research moreover demonstrates the overwhelming reluctance of most retailers to allow the installation of additional cabinets: (see Lansdowne survey, paragraphs 97 to 100 and 103; Rosslyn survey, paragraph 112; B& A survey, paragraph 119). The Lansdowne survey, for example, found that only 40 % of the 25 % of those retailers who are only in a position to offer HB ice cream, and who expressed an interest in stocking other brands, were prepared to accept the installation of an additional cabinet; the B& A survey found that only [. . .] % of retailers with only one cabinet in their outlet were prepared to contemplate the installation of an additional cabinet.

(174) For the sale of impulse ice cream, the accessibility and visibility of the freezer cabinet are essential. Not all sales space in an outlet is therefore suitable (77) (see paragraphs 10 and 43). At the same time, the space which is suitable must also be available for the display of other products, in particular impulse products. Sales of these other products are not in general seasonally variable and may provide higher annual returns on the space used. The allocation of retail space to cabinets for the sale of impulse ice cream will in general bear a direct relation to their contribution to sales.

(175) The installation of an additional retailer-owned freezer cabinet or of one supplied by a manufacturer other than HB has as its immediate result the availability of a wider range of products. This extension of product range, however, does not automatically generate a higher turnover in impulse ice cream in that outlet. In fact, while overall sales of impulse ice cream from that outlet may well increase, they will not necessarily increase to the level which a retailer would expect from such additional allocation of space. Not only will the newly- installed cabinet not be likely to generate the same level of sales as the incumbent cabinet/s did until the new installation, but the incumbent cabinet/s, will not be likely to continue generating the same level of sales as previous to the new installation. It is only if all sales resulting from the new cabinet are entirely attributable to specific demand for the newcomer's products, to the exclusion of demand for any other impulse ice-cream products available in the outlet, that sales from the incumbent cabinet/s will not suffer as a result of the new installation. From the retailer's point of view, the additional installation will only be viable where he forecasts that the additional demand created by the new cabinet will compensate for the lost opportunity of using the retail space in another way. The substantial level of de- stocking of brands recorded by the B& A survey (paragraph 118) is consistent with these space viability considerations, the retailers in question may have considered that the sales of a supplier's ice cream did not justify the use of space which that supplier's cabinet required.

(176) The strength of HB's position on the relevant market means that a retailer is therefore very unlikely to install an additional retailer-owned cabinet in an outlet which already contains one or more of HB's network of cabinets. Because of the range and popularity of the products offered by HB, the additional profits likely to result from the installation of a retailer-owned cabinet alongside the HB cabinet are reduced. The likely resulting inefficiencies described more generally above will therefore be all the more pronounced in this instance.

(177) Furthermore, the most efficient way of increasing retailer capacity so as to satisfy an increase in consumer demand, is by more frequent deliveries to a minimum number of freezer cabinets rather than by the installation of additional cabinets (78). Having to install additional cabinets in order to cater for each supplier's products leads to obvious inefficiencies for the retailer.

(178) A further factor rendering such additional placement all the more unlikely is the increasing floor space being taken up by HB's recently-installed cabinets (see paragraph 74). This naturally reduces the likelihood of a retailer allocating yet more space to the sale of impulse ice-cream products.

(179) The possibility of replacing an incumbent HB freezer cabinet by two (or more) smaller ones should also be considered. While this option may not have the effect of increasing the amount of retail space taken up by the storage of impulse ice cream, and appears to appeal to some retailers (see Lansdowne survey, paragraph 104), it is constrained by the fact that suppliers seem to provide cabinets designed to accommodate their entire range: these are tending to increase in area for the full range supplies, in particular HB (see paragraphs 74 and 75 regarding HB cabinets). It will in any event not normally be possible for a retailer to obtain cabinets from suppliers which cater for the exact range and amount of products he may wish to display. Installing a larger number of smaller cabinets may moreover complicate a retailer's attempt to offer to his customers a complete range of all types of impulse ice-cream products. From a supplier's point of view, very small freezer cabinets have the disadvantage of limiting opportunities for expansion in the relevant market. In addition, it should be noted that the purchase, installation, maintenance and running costs of two cabinets are considerably higher than for one cabinet of twice their size.

(180) HB has argued that the average number of cabinets per outlet might continue to increase, as appears to have occurred over the past five years: the Rosslyn survey showed an increase in the average number of cabinets from 1,22 in 1991 to 1,42 in 1996; the Lansdowne survey found an average figure of 1,50. It is clear, however, that there must be an upper limit on the number of cabinets which an outlet is prepared to install for the sale of impulse ice cream. This point is illustrated, not only by the very small proportion of outlets with more than two cabinets (79), but also by the reluctance expressed by retailers when questioned about their willingness to take further cabinets: the Lansdowne survey came up with detailed attitude data in this regard (see paragraphs 97 to 100 and 103). It should in particular be noted that retailers considered 1,57 (average) to be the optimal number of cabinets to have in place during the height of the season, indicating (on the basis of the figures for the average number of cabinets in place in outlets) that the market is more or less cabinet-saturated, and that new entry or expansion in the future is therefore that much less likely.

(181) In addition to the general considerations just outlined in relation to the likelihood of a retailer choosing to install a cabinet in addition to the HB cabinet/s already present in his outlet, the following points should be made in relation to the specific options mentioned at paragraph 172.

(a) Addition of retailer's own cabinet alongside an HB cabinet/s

(182) Most of the disincentives described at point (i)(a) in relation to the acquisition of their own cabinets by retailers apply also to this scenario. A retailer will be deterred by the risk inherent in making such an investment (probably 'sunk`). The inconvenience involved in acquiring and maintaining a cabinet also amounts to a substantial deterrent for retailers.

(b) Addition of a supplier-exclusive cabinet alongside an HB cabinet/s

(183) From the point of view of the supplier, the factors mentioned at paragraph 169 (and further discussed below) are at least equally applicable. Where a supplier provides an additional cabinet, as opposed to a replacement one, his profits will be constrained by competition from the remaining incumbent/s, whereas a replacement cabinet does not have to compete with the displaced incumbent in that outlet. When that incumbent is HB, the strength of HB's position on the relevant market means that a retailer is even more unlikely to install an additional supplier-exclusive cabinet in an outlet which already contains one or more of HB's network of cabinets. Because of the greater range and popularity of the products offered by HB than by any of its competitors, the additional profits likely to result from the installation of another supplier's cabinet alongside the HB cabinet are reduced (80).

(iii) Conclusion

(184) From the foregoing it can be concluded that HB freezer-cabinet agreements, concluded in relation to cabinets installed in outlets where the only freezer cabinets in place in the outlet has/have been provided by HB, fulfil the conditions set out in paragraph 143 and so have as their effect the restriction of competition in the relevant market. In view of this conclusion, the category of outlets in question can be said to be de facto exclusively tied to the sale of the impulse ice-cream products of HB. Competing suppliers are consequently foreclosed from access to these outlets.

5. Other relevant market conditions

(i) Freezer exclusivity as a substantial logistical and cost barrier to entry and expansion by competing suppliers

(185) The general provision of supplier-exclusive freezer cabinets to retailers by ice-cream suppliers renders access to the relevant market, and expansion therein, difficult (see paragraphs 147 to 156), a majority of outlets are in a position to offer for sale the impulse ice- cream products of only one supplier. In consequence, it amounts to a substantial impediment to competition between suppliers.

(186) It is clear that the provision of supplier-exclusive freezer cabinets amounts to a logistical barrier to market entrants and to expansion by existing ice-cream suppliers. In point 4, it was demonstrated that, in outlets with only HB cabinets, it is difficult to persuade retailers to replace incumbent HB freezer cabinets, and difficult to persuade retailers to install additional freezer cabinets for the display of impulse ice cream. It is moreover clear from point 4 that, in general, this is true not only for HB cabinets in outlets with only HB cabinets, but also for supplier-exclusive cabinets in retail outlets generally. While other suppliers do not have the position of market strength enjoyed by HB, they none the less provide cabinets to retailers on very similar terms and the cabinets are installed in outlets with the same space constraints. Inconvenience to retailers, their risk aversion, and unwillingness to rupture with established suppliers are all factors contributing to this logistical barrier to competing suppliers.

(187) The market research relied on by the Commission illustrates this general logistical barrier. The three market surveys demonstrate that retailers do not generally replace incumbent supplier-exclusive cabinets with cabinets other than from the same supplier, and that retailers are in general reluctant to attribute space to the installation of further cabinets (see paragraphs 164, 165 and 173).

(188) The provision of supplier-exclusive cabinets also amounts to a cost barrier to entry or expansion in the relevant market. Where a supplier of impulse ice cream in the relevant market, whether an established market participant or a new entrant to the market, wishes to have his products offered for sale from an outlet where a supplier-exclusive cabinet/s only is/are in place (that is to say, from most outlets in the relevant market, see paragraph 149), and where he cannot persuade the retailer to acquire his own non-exclusive cabinet, the supplier's only course of action is to make available to the retailer a cabinet for the storage of his (the newcomer to the outlet's) products. Apart from the difficulty involved in persuading the retailer to accept the offer (detailed above), the cost to the supplier of providing such a service is very substantial.

(189) The expense involved in acquiring a stock of freezer cabinets for installation in outlets which will ensure that a supplier's products can achieve viable distribution levels renders it very difficult for small and even many medium-sized companies to enter the relevant market and sustain their presence there. There are also ongoing cabinet maintenance costs, which costs are only sustainable where there is a minimum number of the supplier's cabinets present in a reasonably concentrated geographic area. The two small impulse ice-cream companies that have been operating in Ireland, Valley (now in liquidation) and Leadmore, have both experienced serious financial difficulties in recent times, and both have resorted to cooperation in distribution with newcomers to the market (Valley with Mars, and Leadmore with Nestlé). HB, Mars, Nestlé, Dale Farm and Häagen Dazs all belong to large multinational groups of companies. Apart from the greater risk involved for small companies in financing such an investment, large groups can achieve scale economies in both the acquisition and maintenance of cabinets which are not possible for smaller companies, again raising the cost of entry for such companies.

(190) The breadth of the range of products offered by the supplier is also a consideration. The smaller the range, the more difficult it is to justify the investment. If a supplier only offers one or two products, he will almost certainly not offer the retailer a cabinet, and so will be precluded from entering the outlet. This militates in particular against niche suppliers of impulse ice cream, limiting their opportunities for obtaining access to wide-scale distribution.

(191) Established suppliers, with cabinets in place in a significant number of outlets, are protected from the entry of competitors to outlets in which they are present by this cost barrier: the presence of a number of networks of incumbent supplier-exclusive cabinets, extending to almost all outlets in the relevant market, has the effect of raising the costs of the entry to the relevant market of non-incumbent competitors and expansion therein by all suppliers. Retailers will not be likely to accept cabinets from suppliers not offering terms which are at least equally favourable to those being offered by suppliers with cabinets already present in the outlet, or to those being offered by suppliers in the market generally. In the context of the relevant market, this means that the supplier must be prepared to offer a modern cabinet on 'free on loan` terms (hence entailing no direct charge), and with maintenance assured by that supplier. A new entrant will moreover not normally expect to get the same return from installing a cabinet as an incumbent would, at least not until his position in the market has been well-established. This incumbency advantage has ensured that the only new entrants to the relevant market in recent times have had 'deep pockets`: Mars, Häagen Dazs and Nestlé. The cost of the entry of Mars is illustrated by the investment which it made in cabinets in 1994, as a proportion of its ice-cream turnover in that year (see paragraph 80). This heavy investment in cabinet provision means that an ice-cream supplier's other investment requirements suffer accordingly, particularly during the period following market entry when the supplier has not yet recouped from the retailer the cost of its investment in freezer cabinets.

(192) HB is by far the most important incumbent supplier in the relevant market. Its longstanding position of strength in that market has been described in detail above. Its network of cabinets is the largest in the relevant market, about three times the size of that of its nearest competitor, and HB cabinets are present in most outlets. These cabinets also tend to be larger than those installed by its competitors, and the fleet has been extensively modernised in recent years.

(193) HB has expressed the view, in this regard, that a distribution system whereby freezer cabinets are made available to retailers subject to a condition of exclusivity does not, in all the circumstances of the relevant market, amount to a restriction of competition within the meaning of Article 85(1). HB has stated that 'Competition of the type considered under Article 85(1) in conjunction with Article 3g is broad enough to encompass a system where ice-cream suppliers compete, inter alia, by providing cabinets at point of sale`. This contention is essentially based on the fact that the system is established commercial practice in the relevant market and elsewhere, and that competition in the relevant market includes competition in the supply to retailers of freezer cabinets subject to exclusivity. This competition in the supply of freezer cabinets applies both to established operators and to new entrants.

(194) To this it can be responded that competition in the relevant market should not compel ice-cream manufacturers to engage in the supply of freezer cabinets as an unavoidable prerequisite to operating in the market for the supply of impulse ice cream to retailers. In most cases where suppliers wish to have access for the sale of their products to an outlet having no 'open` freezer cabinet, such compulsion appears to be a reality. Moreover, interbrand competition at the consumer level cannot be substituted by competition for access to the retail level. This is particularly the case where impulse products are concerned, in that competition tends to a large extent to be intra-outlet rather than between outlets (see paragraph 195 et seq.). In the present case, the necessity to compete in the supply of freezer cabinets has the effect of substantially restricting competition in the market for impulse ice cream. It goes without saying that competition between ice-cream suppliers in the provison of freezer cabinets to retailers is not in itself undesirable; it is only when such provision is made on exclusive terms by all ice-cream suppliers that any economic benefits brought about by the practice are outweighed by its harmful effects on competition in the relevant market.

(ii) Other anti-competitive consequences of the barrier to entry and expansion discussed above

(195) There are at least two additional factors rendering entry to the relevant market more difficult: first, the relatively undeveloped independent wholesale business in impulse ice cream in Ireland means that access to distribution via such independent intermediaries is rendered more difficult; secondly, the strength of existing brands in the relevant market, and customer loyalty towards them, amounts to a formidable obstacle to new entrants. It is acknowledged that brand recognition in the ice-cream business, as in that of most fast-moving consumer goods, is of paramount importance. In this regard, it should be pointed out that Mars and Nestlé have both benefited substantially from the strength of their respective brands in the confectionery sector, which brands are in many cases identical to those of their ice-cream products. The other new entrant in recent times, Häagen Dazs, benefits from a very high international brand profile and has invested considerable resources in advertising; its market share remains in any event very small. The particular strength of the HB brand in the relevant market, and of the wide product range which it offers under this brand, is an obstacle rendering competition with HB all the more difficult.

(196) New entrants are, as a result of the barriers to entry described above, precluded in practice from many retail outlets for the sale of their products. When entry is possible, it can only be achieved at great cost, thereby presenting a particular obstacle to the entry of small and medium-sized companies. Existing suppliers are similarly restricted in their scope for expansion in the market. The relative numeric and weighted distribution figures achieved by the various ice-cream suppliers provides clear evidence of the difficulty in achieving wide distribution (see paragraphs 46 to 52): in particular, no supplier approaches the distribution levels being obtained by HB. Mars, for example, had achieved a higher degree of market penetration in the year after its entry (when retailers had been placing its products in HB cabinets) than it has managed to achieve in the more than five years which have elapsed since then (see paragraphs 47 and 48). Nestlé, who entered the market in 1994, has seen its numeric distribution level remain unchanged since then, and only a slight increase in the weighted level (see paragraph 50). The effectiveness of these barriers is understood by the suppliers operating the practice of cabinet provision subject to exclusivity. This is most clearly illustrated by the various reactions of HB to the market entry of Mars in 1989 (see paragraphs 64 to 68): it is evident from company documentation that HB's insistence on maintaining freezer exclusivity was not motivated by concern about 'piracy` of its property alone, but rather saw the exclusivity as a very useful means of checking the effective entry and expansion of the new rival.

(197) As mentioned above, the heavy investment in cabinet provision means that other investment requirements suffer as a result, and that the consequent allocation of resources by suppliers is likely to lead to a distortion of competitive forces in the relevant market. Competitors who are obliged to invest so heavily in achieving physical distribution, particularly when entering a market, may be obliged to neglect other aspects of the marketing of their products. Such marketing (advertising, promotion and so on) is particularly crucial at the market entry stage, the point at which investment in cabinets is also most important. New entrants or suppliers seeking to grow in the market are similarly constrained in their ability to innovate and develop new products, to the detriment not just of those suppliers but also of consumers, as a result of this cost of obtaining or increasing distribution levels.

(198) As demonstrated above, supplier-exclusive cabinet networks have the effect of severely restricting the number of competitors' products which an outlet can offer. As competition in impulse products is to a large extent intra-outlet on account of the spontaneous nature of the demand created by the impulse to purchase (see paragraph 12), this results in a serious restriction of interbrand competition (competition between suppliers) in the relevant market. Where an outlet is only in a position to offer the products of a single supplier, such competition at the consumer level is excluded altogether: competition is only between the various products offered by the same supplier. This is the case in over half of all retail outlets in the relevant market, where there is a supplier-exclusive cabinet/s alone in the outlet (81). In these circumstances, price competition between suppliers is likely to be reduced (82). Consumers consequently suffer both from a poorer choice of competing products and from the effects of weaker price competition.

(199) The harmful effects to retailers produced by supplier-exclusive cabinet networks consist in a restriction in their ability to offer the products of competing suppliers, limiting their ability to select from the product ranges of such suppliers in accordance with demand or perceived demand. This restriction can result in lost opportunities for the retailer to become more efficient, and hence more profitable, on the basis of his choice of product offering: a retailer may often stock products in supplier-exclusive cabinets which, in the absence of exclusivity, he would not necessarily choose to offer for sale. The exclusive networks are also likely to often result in inefficient space allocation inside an outlet (as discussed above); the more that a retailer seeks to compensate for his inability to stock all the products which he wishes to offer for sale to consumers by accepting the installation of more cabinets, the more these space inefficiencies will become apparent.

(200) HB's cabinet agreements, as the largest and most extensive single network of supplier- exclusive cabinet agreements, contribute substantially to the anticompetitive effects just outlined. Indeed, in the case of HB's network, these effects are accentuated by the strength of HB's position on the market. In those outlets where the only cabinets in place in the outlet are HB cabinets (some 40 % of all outlets in the relevant market), HB is exposed to no interbrand competition at the consumer level whatsoever.

6. Likelihood of affecting trade between Member States

(201) Where, for the reasons described above, the effect of the HB freezer-cabinet agreements in question is to eliminate the freedom of retailers to stock and offer for sale to the consumer the impulse ice-cream products of competing suppliers, those suppliers are impeded, irrespective of their geographical location and the origin of the goods, from gaining access to the retail outlets concerned. This restriction of retailers and suppliers has the effect of rendering penetration of the Irish market more difficult for foreign competitors, with the consequence that the level of trade in ice-cream products may be at a lower level than would otherwise be the case. The opportunities for foreign suppliers to establish themselves in the Irish impulse ice cream market are adversely affected by the cabinet agreements, thereby contributing to the consolidation of markets along national lines (83). There is, moreover, significant trade between Ireland and other Member States in ice-cream products, as exhibited, inter alia, by the import of all Mars' ice-cream products from France, and by the import of a portion of HB's products into Ireland (paragraphs 25 and 29). Accordingly, the HB freezer-cabinet agreements may affect trade between Member States.

7. Appreciability

(202) The identified part of the network of HB cabinet agreements referred to at paragraph 184 only infringes Article 85(1), however, if it affects competition and trade between Member States to an appreciable extent. It has been established that the identified part of HB's network of cabinet agreements relates to cabinets installed in some 40 % of all outlets in the relevant market, representing some 40 % of total impulse ice-cream sales in the relevant market. This alone, but particularly in combination with the far-reaching anti-competitive effects brought about by the various networks of supplier-exclusive cabinet agreements operated by suppliers in the relevant market generally, and against the background of the other relevant market circumstances discussed above, amply demonstrates an appreciable restriction of competition within the meaning of Article 85(1) and an appreciable effect on trade between Member States.

(203) It can thus be concluded that the part of HB's network of cabinet agreements relating to cabinets installed in outlets where the only freezer cabinets in place in the outlet has/have been provided by HB, fulfils all of the conditions for the application of Article 85(1).

(204) HB has argued that the Commission has not observed the principle of equal treatment in that it has made a finding in relation to HB's network of supplier-exclusive cabinets alone, and not in relation to those of HB's competitors in the relevant market, thereby 'tilt(ing) the competitive playing field` against HB. To this, it should be responded that the Commission has taken into account the overall anticompetitive effects produced by the other networks of supplier-exclusive cabinets in place in the market. The Commission's examination has not, however, identified a significant contribution to the foreclosure of the relevant market by any single supplier's network of supplier-exclusive cabinets (or identifiable part thereof) other than that produced by the relevant part of HB's own network (see paragraph 95).

(205) HB has submitted that Article 85(1) does not require that all impediments to the availability of different manufacturers' impulse ice-cream products in any outlet in the relevant market must be removed. In this context it has made reference to the Court's reasoning in Delimitis (84), and concluded that there need only be access for competitors to the minimum number of outlets necessary for the economic operation of a distribution system (paragraph 21 of the judgment). In HB's view there need only exist some possibility for expansion by competitors in the market. HB considers that these conditions are met if a supplier is able, at whatever cost, to achieve a minimum degree of market penetration and to operate profitability.

(206) In relation to this point, it should be emphasised that exclusivity agreements of this kind should not be considered per se infringements of Article 85(1), without reference to the context in which they are found (see paragraph 129). Neither is it suggested that Article 85(1) requires that all impediments to the availability of different suppliers' impulse ice-cream products in any outlet in the relevant market must be removed. Where, however, a supplier's network of exclusive agreements (or an identifiable part thereof), in all the circumstances of the relevant market, produces the very substantial anticompetitive effects described above, thereby also affecting trade between Member States, the network of agreements in question is incompatible with the common market.

(207) HB has further argued that the condition of exclusivity contained in its freezer-cabinet agreements falls outside Article 85(1) as a restraint ancillary to the allegedly otherwise legitimate object of those agreements, namely the provision of installations to retailers to enable them to stock ice cream for sale (85). To this, it must be replied that such a restraint can only escape from the application of Article 85(1) where it is directly related and objectively necessary to that purpose. Given that there is no objectively indispensable link between the supply of freezer cabinets to retailers for the purpose of stocking impulse ice cream and the condition of exclusivity, that condition cannot be considered ancillary.

(208) HB considers that the Commission's reasoning is flawed in concluding that freezer-cabinet agreements which are 'effectively terminable at will` (in that HB apparently does not seek to enforce the two months' notice requirement) can lead to a de facto tie of the kind referred to at paragraph 184. A 'temporary relationship` of this kind is, according to HB, inconsistent with the existence of such a tie. The Commission does not agree. The cabinet agreements are of indefinite duration, requiring the initiative of either party to bring them to an end. Far from indicating that the contractual relationship is a temporary one, the evidence demonstrates that the opposite is true. The economic reality, as confirmed by the market research, is that retailers who take HB cabinets only very rarely replace them with their own or competitors' cabinets (see paragraph 158 et seq.). HB cabinet agreements are almost invariably replaced by identical cabinet agreements (see paragraph 111).

(209) Nor does HB accept that there is evidence that the category of outlets with only HB cabinets identified at paragraph 184 is foreclosed to competitors, or that there is any causal connection between that restrictive effect and the exclusivity provision in the HB cabinet agreements relating to those outlets. In this regard, it principally relies on the degree to which these retailers appear to be satisfied with their current supply arrangements. HB contends that those outlets where there is 'no interest on the part of the retailer to stock another brand of ice cream` should be excluded in any proper calculation of foreclosure. HB also points to the apparent low level of consumer demand for brands other than HB in the outlets in question as being the probable reason for this lack of interest.

(210) Again, the Commission cannot accept this reasoning. An agreement between any two parties is capable of producing an effect restrictive of competition within the meaning of Article 85(1), irrespective of the motivation of the parties in concluding such a restrictive agreement. More specifically, an exclusive purchasing arrangement of whatever kind is, in principle, liable to fall foul of Article 85(1) where it has the effect of foreclosing third party suppliers to a significant extent, notwithstanding the fact that the party subject to the exclusivity may express himself, at a given moment in time, to be disinterested in purchasing from any such third parties. As regards the apparently low level of consumer demand for brands other than HB, it should be pointed out that competitors' inability to gain access to wide-scale distribution for these impulse products is one of the principal reasons for absence of demand, particularly in the case of products which are being newly-launched (see paragraphs 12 and 44). In the absence of the exclusivity provision, retailers could more easily be persuaded to stock additional products, given that this would not in such circumstances require the replacement of an incumbent, or the installation of an additional, freezer cabinet. It can be expected that such additional brand stocking might in turn stimulate higher levels of consumer demand for brands other than HB.

8. Property rights

(211) HB has argued that the application of Article 85(1) (86) of the EC Treaty to the exclusivity provision in its cabinet agreements would be tantamount to interference with its property rights, contrary to Article 222 of the EC Treaty, in that it would permit other manufacturers' products to be stored in its freezer cabinets.

(212) The Commission does not accept this argument. In the Community legal order the right to property, as laid down in Article 222 of the EC Treaty, is guaranteed in accordance with the principles common to the constitutions of the Member States. In the constitutions of all Member States, however, it is recognised that the exercise, as distinct from the essence, of property rights may be restricted in the public interest, and to the extent necessary in that regard (87).

(213) As one of the fundamental provisions of Community law, Article 85 of the EC Treaty serves the public interest. Moreover, the application of Article 85(1) in the present proceedings does not concern HB's use of its own property, but the restrictions which it imposes on others to whom it has granted the use of its property. Furthermore, HB's exercise of its property rights would, in the present case, only be restricted to the extent necessary to ensure that competition in the common market is not distorted (Article 3g of the EC Treaty).

(214) As a general principle, this approach is reflected in the Commission's block exemptions with regard, inter alia, to bans on dealing in competing products contained in exclusive purchasing agreements (see Commission Regulation (EEC) No 1984-83 (88), as amended by the Act of Accession of Austria, Finland and Sweden, Article 8(2)(b)), and exemptible obligations contained in technology transfer agreements (Commission Regulation (EC) No 240-96 (89), Article 1). In these Regulations, restrictions imposed by the owners of property rights on the users of those rights are, under certain circumstances, exempted and thus, in the first instance, fall foul of Article 85(1) (90).

(215) HB has argued that the exclusivity provision in its freezer-cabinet agreements is analogous to the types of obligations permitted, for example, according to Article 2 of Regulation (EC) No 240-96, and in particular those listed at 1(8) and 1(2) of that Article; these are the clauses permitting restrictions to specific 'technical fields of application`, and restrictions forbidding the granting of sub-licences. This analogy is flawed for the following reasons.

(216) First, a 'field-of-use` type restriction would relate to the use of the freezer cabinet for the stocking of ice cream as opposed to other categories of products; the exclusivity condition in HB's freezer-cabinet agreements, however, restricts the use of the freezer cabinet for stocking similar products of different manufacturers. Even were the 'field-of-use` type of restriction to be considered similar to the exclusivity condition in HB's freezer-cabinet agreements, Article 2 of Regulation (EC) No 240-96 does not exclude the possibility that such obligations may be restrictive of competition in certain circumstances, as is the case here.

(217) Secondly, the present proceeding does not interfere with the right of HB, as the owner of the property, to deny parties other than the retailer to whom the freezer cabinet has been supplied the use of that freezer cabinet. Rather, it is intended to permit the retailer to exercise his commercial freedom of choice by allowing him to decide what products he should offer in his outlet. The present proceeding is accordingly aimed at the contractual provisions concerning the conditions on which the retailer may use the property which the owner has, for a rental fee, made available to him.

(218) HB has further contended that it would suffer considerable competitive detriment if third party competitors could, by inducing retailers to also use the freezer cabinets which HB supplies for the storage of those competitors' products, avoid the costs that the competitors themselves might incur if they had to provide their own freezer cabinets at point of sale. HB argues that other manufacturers thus benefit from the investment which it has made.

(219) In relation to this argument, it should not be forgotten that it is ultimately the retailers who are paying for the cabinet provision, not HB. It should also be pointed out that there need be no objective connection between the exclusivity provision in HB's freezer-cabinet agreements and arrangements for the repayment of its investment in those cabinets. The inclusion of this cost element in the price of the ice cream charged to retailers is a choice which HB has made for its own commercial reasons. Other methods of recouping the investment in freezer cabinets could be devised in order to avoid the consequence that the retailer passes on this cost to consumers only via the price which he charges for HB's products. The application of Article 85(1) to the exclusivity provision in HB's freezer-cabinet agreements would not, for example, prevent HB from charging the retailer a separate rental fee for the supply of a freezer cabinet sufficient to cover the entire cost of that investment. Although HB asserts that the charging of such a rent would involve the additional transaction costs involved in administering and collecting the rental payments, this would hardly be over- burdensome on HB, given that the company already arranges for the billing of its customers in relation to the ice cream itself, and for the collection of payment from those customers.

(220) HB further argues in this context, as described at paragraph 230, that if it introduces a separate rental system, this will involve considerable costs, may not prove attractive to retailers, and will motivate retailers to turn to HB's competitors for 'free-on-loan` cabinet provision. The Commission can only respond to this point in the same manner as it has done at paragraph 231.

B. ARTICLE 85(3)

(221) The provisions of Article 85(1) cannot, under Article 85(3), be declared inapplicable to the freezer-cabinet agreements concluded between HB and retailers, and relating to cabinets installed in outlets where the only freezer cabinets in place in the outlet have been provided by HB, for the reasons set out below.

1. Improving the distribution of goods

(222) Any improvements in distribution which the part of HB's network of freezer-cabinet agreements which infringes Article 85(1) (see paragraph 203) may bring, have to be balanced against the restrictive effect of those agreements, and in particular how they limit access by competing ice-cream suppliers in the relevant market to the retail level and ultimately to consumers.

(223) The restrictive effect brought about by the relevant part of HB's network of cabinet agreements is broadly analogous to that produced by exclusive purchasing obligations. The fifth recital in the preamble to Regulation (EEC) No 1984-83 on the application of Article 85(3) to categories of exclusive purchasing agreements states that exclusive purchasing obligations '. . . lead in general to an improvement in distribution`. They '. . . enable the supplier to plan the sales of his goods with greater precision and for a longer period, and ensure that the reseller's requirements will be met on a regular basis for the duration of the agreement . . . this allows the parties to limit the risk to them of variations in market conditions and to lower distribution costs`.

(224) The HB cabinet agreements may secure some or all of the benefits described in the fifth recital to Regulation (EEC) No 1984-83 for HB itself and for the retailers who are the other parties to the agreements. The fact that these agreements may be advantageous to the parties to those agreements is, however, not sufficient to constitute an improvement in distribution within the meaning of Article 85(3). Rather, they must produce appreciable objective advantages in the public interest of such a character as to compensate for the disadvantages which they cause in the field of competition (91).

(225) An objective benefit of this kind which exclusive purchasing obligations may conceivably bring about is an intensification of interbrand competition (sixth recital to Regulation (EEC) No 1984-83). It is clear that the cabinet agreements in question considerably strengthen HB's position in the relevant market, particularly with respect to potential competitors. However, a strengthening of an undertaking which is as important on the market as HB leads not to more but to less competition, because the network of that undertaking's agreements constitutes a major barrier to the entry of others into the market, as well as to expansion within the market by its existing competitors.

(226) HB has, for example, pointed to the importance of 'merchandising` in the sale of branded products; the condition of exclusivity enables HB to market its portfolio of products/brands together in a unified display or 'corporate block`, with the brands giving authority to each other, as well as benefiting from the advantages of visibility and prominent positioning in an outlet. According to HB, the comprehensiveness of its range means that an HB freezer cabinet will generally satisfy consumer demand for different types of impulse ice- cream products. HB points out that, in the absence of exclusivity, some of its product offering (particularly 'lower priced and lower margin impulse products`) might be displaced. The advantages to HB produced by the exclusivity provision clearly strengthen its position in the market and, in those outlets with only supplier-exclusive cabinets, and in particular those with only an HB cabinet/s, they contribute to the barriers to entry and to expansion within that market described above under Article 85(1).

(227) The wide availability of freezer cabinets in retail outlets for the sale of impulse ice- cream products, covering the totality of the geographic market, and brought about in large part by HB's cabinet network, can be considered an objective advantage (in the distribution of products) which is in the public interest. In this regard, HB has said that it cannot be assumed that it would continue to be commercially attractive for it to provide freezer cabinets to all of the outlets which it currently supplies with a cabinet/s, were the exclusivity provision in its freezer-cabinet agreements to be dropped. HB has argued that, were it to cease supplying a freezer cabinet/s to a number of these retailers, it is likely that some of them would no longer continue to sell impulse ice cream at all, as they would not be able to justify the investment required for the installation of their own freezer cabinets.

(228) In relation to this contention, it must be pointed out that it is unlikely that retailers would cease altogether to be offered freezer cabinets, on whatever terms, by HB in any but a small number of instances, were it to be restricted in its ability to impose an exclusivity obligation in relation to those cabinets. The commercial reality for a company like HB, wishing to maintain its position of strength on the market, is that it will seek to be present in as many outlets as possible, even in those with lower levels of impulse ice cream sales. As described above, widespread visibility and availability are important for all impulse products. The need to ensure sales and thus profit levels provides sufficient incentive for an ice-cream supplier, particularly one with a range of products and a share of the market as important as HB's, to continue ensuring that retail outlets are able to stock its products for sale to consumers. Moreover, even where the freezer cabinet/s in an outlet would no longer be subject to exclusivity, it is likely that the products stocked in the cabinet/s would more or less reflect the different market shares of the various different manufacturers in the relevant market. HB could therefore expect, on the strength of its product offering alone, to continue to achieve a very substantial share of ice-cream sales in such outlets. HB is accordingly unlikely to run the risk of losing these sales altogether by ceasing to make freezer cabinets available to outlets which would otherwise not choose to continue stocking ice cream at all. HB has, in any event, a policy of only providing freezer cabinets to retailers on the basis of certain established financial and trading criteria which ensure a minimum return on HB's investment (see paragraph 62), with the consequence that the least important outlets in the relevant market are often not currently provided with HB freezers. It should also be pointed out, in this regard, that most of the recent cabinet disposals by HB took place in low-turnover outlets (see paragraph 73).

(229) HB argues that, in the absence of its ability to enforce cabinet exclusivity, the competitive playing field would be tilted against it as a supplier which provides freezer cabinets in favour of those who do not. It says that other ice-cream suppliers would be able to offer preferential terms to retailers in that they would not be carrying the cost of freezer cabinet provision. HB has said that its natural reaction in such circumstances would be to discontinue making freezer cabinets available to retailers. In relation to this argument it should be remembered that there is no question of the application of Article 85(1) (where Article 85(3) is found to be inapplicable) to the exclusivity provision in HB's freezer-cabinet agreements not permitting HB to use whatever methods it chooses of recouping the investment it has made in freezer cabinets. Such methods need not have the consequence that other ice-cream suppliers benefit from HB's investment (paragraphs 218 and 219).

(230) HB further argues that, if it does introduce a separate rental system, it has no certainty that retailers will opt to continue taking HB cabinets. In support of this contention, HB points to the failure of the hire-purchase scheme which it introduced in 1995 to attract the interest of retailers, the generally low number of retailer-owned or leased cabinets in the market, and the low level of reported retailer preparedness to buy or rent their own cabinets found by the Rosslyn survey. In these circumstances, HB contends, retailers will be more inclined to turn to the other suppliers of ice cream in the market, who may continue to make available freezer cabinets to retailers on the basis of cabinet exclusivity, thereby effectively excluding HB from many of those outlets.

(231) In relation to this argument, it must be stated that the reasons for the failure of the hire- purchase scheme introduced by HB in 1995 are not solely explicable by an unwillingness in principle on the part of retailers to own or lease their own cabinets. Rather, the Lansdowne survey found (see paragraph 100) that the principal reason was satisfaction with current arrangements, experience has shown that retailers, faced with the option of taking from HB a 'free on loan` cabinet or one on hire purchase, have invariably opted for the former. That it not to say that, in the absence of cabinet exclusivity, retailers would not continue to stock ice cream: the B& A survey found that over [. . .] % of retailers said that they would continue to do so in such hypothetical circumstances, and that many would choose to buy/lease their own cabinet to that end (see paragraph 120). In relation to HB's assertion that retailers will instead turn to HB's competitors for the provision of 'free on loan` supplier-exclusive freezer cabinets, it can only be said that if retailers do so to such an extent that that network, or those networks, of supplier-exclusive freezer cabinets (and where these are the only cabinets in place in those outlets) then contribute significantly to the overall foreclosure of the relevant market, then the relevant part of such network would equally fall foul of Article 85(1).

(232) It cannot however be ruled out that HB may choose to cease providing freezer cabinets to a certain small number of outlets, for example those no longer meeting certain criteria relating to HB ice cream sales or distance from an HB depot. These outlets will not all necessarily cease stocking impulse ice cream as a result. Such a retailer may choose to buy/lease, his own cabinet or another ice-cream supplier may decide to make a freezer cabinet available to him. Other competing manufacturers may well have a policy of making freezer cabinets available to outlets which generate a lower turnover in impulse ice cream than would make the outlet eligible for an HB cabinet, and on terms more attractive than the retailer could expect to obtain himself.

(233) Even where such competitors are not forthcoming to make available freezer cabinets in outlets from which HB has withdrawn its cabinets, or to which it does not choose to provide them, there still remains the possibility that cabinets could be installed by independent dealers who would obtain supplies from a variety of sources and supply all of those outlets' requirements. The fact that there are so few independent dealers of this kind at present is at least partly due to the exclusivity obligations arising from the freezer-cabinet agreements customary in the trade, in combination with the leading supplier's (HB's) policy of direct delivery to almost all outlets. In the absence of cabinet exclusivity, the supply of ice cream to consumers would not therefore be likely to be affected to any significant extent.

(234) The possible slight reduction in the number of outlets stocking impulse ice cream in the relevant geographic market which might result from the dropping of the exclusivity provision in HB's freezer-cabinet agreements concluded in relation to cabinets installed in outlets with only HB cabinets must be balanced against the restrictive effect of those agreements. In this assessment, the advantage produced by the fact that a small number of outlets would continue offering impulse ice cream only if the exclusivity provision in the network of HB freezer- cabinet agreements in question were to be upheld, cannot outweigh the disadvantages arising from the restriction of competition which those agreements cause in their totality.

(235) In this regard, reference should be made to the hire-purchase scheme introduced by HB in 1995. It was envisaged that this scheme would motivate retailers to purchase their own cabinets, thereby bringing about a far-reaching structural change in the market which might have enabled the restrictive effect of HB's network to be reduced to an extent where its benefits to distribution could outweigh its restrictive effect. This expectation was however misconceived (see paragraph 71).

(236) HB has also argued that its distribution arrangements lead to planning, logistic and distributive efficiencies; it contends that they facilitate the planning of deliveries and the rhythm of supply; the arrangements lead to efficiencies in distribution in that 'drop sizes` (the quantity taken by the retailer on one delivery) to freezer cabinets which only stock its products are greater than they would be in the absence of exclusivity, and that a high concentration of HB freezers also contributes to such efficiencies. As pointed out above (paragraph 224), the fact that the cabinet agreements may be advantageous to HB and even to the retailers is not sufficient to constitute an improvement in the distribution of goods within the meaning of Article 85(3); the agreements can only be considered as leading to such an improvement where the advantages can be demonstrated to be of an objective kind. Whilst it is clear that the current method of distribution operated by HB may produce certain advantages for itself and retailers in terms of efficiency, it should be pointed out that HB's exclusivity arrangements have the effect of undermining the efficiency of other impulse ice- cream suppliers by effectively requiring them to compete not only in the supply of ice cream but also in the provision of cabinets to retailers on attractive terms. The result is a strengthening of HB's position in the relevant market, reinforcing a barrier to entry into or expansion within it. As such, any advantages resulting from the agreements are clearly outweighed by the disadvantages which they bring in the form of restricted competition.

(237) An assessment of the advantages produced by HB's cabinet network should take into account the scale economies which it achieves in the purchase and maintenance of freezer cabinets (see paragraph 63). HB has argued that these scale economies contribute to improving the distribution of goods within the meaning of Article 85(3). It must be accepted that exclusivity arrangements offer an attractive incentive to a manufacturer to install a large number of freezer cabinets, thereby reducing procurement and maintenance costs. It must be emphasised, however, that exclusivity is not a necessary condition for the achieving of such economies of scale. There is no apparent reason why these scale economies could not continue to be achieved by HB in relation to a fleet of freezer cabinets in the absence of exclusivity.

(238) Although it is accepted that the provision of freezer cabinets to retailers by ice-cream suppliers leads to an improvement in the distribution of impulse ice cream by, inter alia, dispensing with the need for the retailers to concern themselves with capital requirements or the inconvenience of purchasing and maintaining cabinets, it none the less follows from what has been said in the previous paragraph that these advantages can equally be achieved in the absence of the exclusivity requirement. Neither can it be said that HB's network produces only advantages in distribution for retailers. As is discussed in paragraph 199, the network also produces disadvantages by contributing substantially to a reduction in retailers' ability to choose the products which they wish to offer for sale, as well as creating space inefficiencies in those outlets.

2. Allowing consumers a fair share of the benefit

(239) In outlets where only an HB freezer cabinet/s is/are, in place, the freedom of the retailers in stocking and offering for sale to the consumer ice cream of competing suppliers is restricted. The part of HB's cabinet network relating to the cabinets installed in these outlets thereby restricts the choice of impulse ice-cream items available to consumers: only HB impulse ice-cream products are offered for sale in those outlets. Even if there is another outlet selling other manufacturers' products in the neighbourhood, the alternative this represents is not equivalent to a choice available from an individual outlet, as consumers of impulse ice cream will not generally travel to buy or defer a purchasing decision. In any event, such an option will often not be available. Moreover, a consumer who wishes to buy products from different ranges will find it inconvenient to have to visit separate outlets in order to do so. He will not generally take trouble of that kind for an impulse purchase. Due to this restriction of choice, consumers cannot be considered as sharing in whatever alleged benefits those HB cabinet agreements may bring. In view of the restriction of interbrand competition brought about by HB's network, price competition between suppliers is also likely to be weakened, to the detriment of consumers (see paragraph 198).

(240) In so far as the economies of scale in the purchase and maintenance of freezer cabinets and distributive efficiencies (see paragraph 222 et seq.) allegedly facilitated by the exclusivity provision in HB's cabinet agreements are concerned, it should be noted that, given HB's economic strength which enables it to prevent the maintenance of effective competition (see paragraph 255 et seq.), there is no guarantee that these benefits will in fact be passed on to the consumer.

3. Indispensability

(241) The exclusivity condition contained in the relevant part of HB's network of cabinet agreements could only be described as indispensable to the attainment of any of the alleged benefits referred to above if it was the least restrictive means of bringing them about. No compelling reason has been put forward by HB which would indicate that any such allegedly objective advantages, leading to a general improvement of production and distribution for the benefit, inter alia, of consumers, could not be secured equally effectively by removing the exclusivity in favour of HB's products, thereby 'unbundling` the provision of freezer cabinets and the supply of ice cream in the manner described in paragraph 219.

4. Possibility of eliminating competition in respect of a substantial part of the products in question

(242) In this respect, consideration must be given to the barriers to entry to the relevant market, as well as to their effect on competitive relationships existing within that market. These are discussed in detail under paragraph 130 et seq. From that analysis, it can be concluded that HB's network of freezer-cabinet agreements in general, and that part of the network relating to cabinets installed in outlets with only HB cabinets in particular, constitute an important barrier to entering the relevant market or to expansion within it. These barriers lead in turn to a deterioration in competition between competing suppliers in the market. As such, and particularly in combination with the strength of HB's position on the relevant market, they contribute substantially to an elimination of competition in that market.

(243) This elimination of competition is clearly illustrated by the very large proportion of retail outlets in the relevant market which are only currently in a position to offer for sale the impulse ice-cream products of HB as a result of the exclusivity provision in the relevant part of HB's cabinet agreement network: some 40 % of all outlets in the relevant market fall into the category, being outlets which, moreover, account for some 40 % of total impulse ice-cream sales in the relevant markets (see paragraphs 156 and 184).

(244) HB's dominant position in the relevant market, discussed below, enables it to prevent the maintenance of effective competition on that market. As such, it must be considered a major barrier to competition in the market. The brand reputation of HB's impulse ice-cream products, which has been built up over many years, also represents a major impediment to any changes in the competitive structure of the relevant market.

(245) The economic considerations which have just been examined make it clear that competition in the relevant market is seriously restricted. This assessment is confirmed by the fact that it is a long time since any substantial change in the competitive structure of the relevant market has occurred. The market continues to be dominated by one supplier who has consistently secured a share in the region of four fifths of the market, with a number of smaller suppliers (some of whom may change) making up the remainder. HB's freezer-cabinet agreements have helped to ensure the perpetuation of this structural inertia. These agreements, and in particular those concluded in relation to outlets with only HB cabinets, thus afford HB the possibility of eliminating competition in respect of a substantial part of the products in question in such a way as to preclude exemption under Article 85(3).

(246) The cabinet hire-purchase scheme introduced in 1995 by HB as one of the elements in the revision of its distribution arrangements (see paragraph 69 et seq.) appeared prima facie to the Commission to provide, in combination with the other changes introduced at the same time, an opportunity for a far-reaching structural change in the market. HB expressed the belief that the scheme would motivate retailers to acquire their own cabinets on account of the attractiveness of its terms. The scheme was thereby expected to result in a significant reduction in the restriction of competition to an extent which would mean that the residual provision by HB of cabinets subject to exclusivity could no longer be regarded as having the effect of substantially eliminating competition in the relevant market. This result has, however, not materialised.

5. Conclusion

(247) HB's freezer-cabinet agreements, therefore, do not qualify for exemption under Article 85(3) for the reasons outlined above. HB's application for such an exemption must accordingly be refused. In particular, it should be pointed out that the various changes made by HB to its distribution arrangements (see paragraphs 69 to 73) with a view to making those arrangements qualify for exemption under Article 85(3) have, contrary to the preliminary view taken by the Commission (see paragraph 6), not proved capable of bringing about the structural changes which might have made exemption possible. Specific reference to the various elements has in particular been made at paragraphs 228, 230, 231, 235 and 246. It is therefore no longer possible for the Commission to maintain the 'favourable position` which it proposed to take in the publication on 15 August 1995 pursuant to Article 19(3) of Regulation No 17 (see footnote 6).

(248) HB contends that the Commission's refusal to proceed towards an exemption decision amounts to a breach of its legitimate expectations. In this regard, HB says that it 'struck a deal with the Commission`, and that it 'implemented its side of the bargain in full, at its own commercial cost`. HB continues that the Commission has 'renege(d) on its favourable view` in the light of the 'failure of one element of the settlement to attract retailer interest (namely the failure of the hire-purchase element)`. HB points out that the reason for this failure is 'entirely outside HB's control`.

(249) In claiming breach of its legitimate expectations, HB seeks to rely on an indication by the Commission that the proposals which HB had put forward appeared likely, on the basis of a preliminary assessment, to produce effects which would render HB's distribution arrangements eligible for an exemption under Article 85(3). This preliminary view was expressed in a press release issued following HB's notification and, subsequently, in a publication in the Official Journal of the European Communities pursuant to Article 19(3) of Regulation No 17. To describe the contacts between the Commission and HB as resulting in a 'bargain` or 'deal` by virtue of which the Commission renounced the possibility of proceeding towards a prohibition decision is therefore completely misleading. It goes without saying that the Commission could not, in any event, agree to desist from the enforcement of rules which are of objective application. None of the preliminary indications given by the Commission, as just described, are capable of creating legitimate expectations of the kind which could render invalid a subsequent Commission measure inconsistent with those preliminary indications. The press release made explicit reference to the preliminary nature of the Commission's assessment. An Article 19(3) notice can, by its very nature, contain no more than an announcement of the Commission's preliminary intentions, the purpose of its publication being to solicit the views of third parties before the adoption of a final position.

(250) As is mentioned above (paragraphs 235 and 246), the Commission's preliminary assessment was based on an expectation that the revised distribution arrangements would, via an evolution towards wider freezer-cabinet ownership by retailers, lead to a substantial reduction in the foreclosure effect produced by HB's cabinet agreements. The fact that this expectation was disappointed is not attributable to the fact that HB did not implement the proposed changes, but rather to the fact that those changes did not produce the effects in terms of opened outlets which HB had expected.

(251) HB is particularly critical of the fact that the Commission's assessment is based on the failure of only one element in those revised arrangements, the hire-purchase scheme. To this it should be responded that the hire-purchase scheme was not just one, but the central, element of the revised arrangements, and the key to the expected evolution towards wider retailer cabinet ownership. Its failure to attract retailers therefore inevitably entailed the failure of the desired effects to materialise. HB is also critical of the fact that the Commission appeared to have reached a premature conclusion about a measure which was only intended to produce effects over the longer term. In relation to this it can only be said that, by the time the Commission issued its Statement of Objections in January 1997, the scheme had already been running for two ice-cream seasons. Given that not a single retailer had by that stage opted to take a cabinet under the scheme, it was reasonable to assume that it was unlikely to have any long-term impact.

(252) Subsequent to the Commission's indication to HB that the conditions for the granting of an exemption under Article 85(3) no longer appeared to be fulfilled, HB devised new proposals, apparently designed to guarantee the result intended by the introduction of the hire- purchase scheme, and put these to the Commission in June 1996. The proposals can be summarised as follows: (i) the exclusivity provision in two types of smaller cabinets which HB provided to retailers would be limited to five years; (ii) at the end of the five-year period, the retailer would have 'the opportunity to take the cabinet, free of all ties, to use as he wishes`; (iii) following a 12-month interval, an outlet taking such a cabinet would be eligible for the differential pricing lump sum.

(253) The Commission considers that these proposals would not render HB's distribution arrangements eligible for exemption, principally for the following reasons: (i) the offer would be confined to two types of small freezer cabinet and, more fundamentally, the choice as to which outlets could take up this offer would rest with HB, rather than with the retailer, as is the case under the current hire-purchase scheme; (ii) there would be no possibility for retrospective application of the scheme, and so the first effects would be felt, at the earliest, five years following its introduction; (iii) there would be no guarantee that the cabinet would not be replaced by an HB-exclusive cabinet during the five-year period, or shortly thereafter, in which case the exclusivity in favour of HB would continue.

(254) HB is of the view that the Commission has not given serious consideration to these new proposals, and that an assessment of them should have been contained in the Statement of Objections. The Commission does not agree. The reasons why the new proposals would not alter the Commission's view as to the eligibility for exemption of HB's distribution arrangements, as just described, were detailed in a letter sent to Unilever several weeks after receipt of the proposals. This was an appropriate administrative response to proposals made by a company with a view to obtaining an exemption under Article 85(3). Given that the proposals were not implemented, it was not necessary to take them into consideration in the Statement of Objections. The Commission's position in relation to the proposals was moreover reiterated at the oral hearing in June 1997.

C. ARTICLE 86

1. Dominant position

(i) Relevant market

(255) The relevant market in which the position of HB falls to be assessed is that of single-wrapped items of impulse ice cream in Ireland. There is no reason, nor has HB suggested one, to take the view that the geographic and product markets differ from the Commission's findings in the context of the assessment on the basis of Article 85 (paragraphs 120 et seq.). Ireland is a substantial part of the common market within the meaning of Article 86.

(ii) Economic strength

(256) The Court of Justice has held (92) that the dominant position referred to in Article 86 relates to a position of economic strength enjoyed by an undertaking which enables it to prevent effective competition being maintained on the relevant market by giving it the power to behave to an appreciable extent independently of its competitors, customers and ultimately of its consumers.

(257) The existence of a dominant position may be inferred from a combination of several factors which, taken separately, might not necessarily be conclusive but which, together with others, constitute such a position; among these, an important one is the existence of a very large market share.

(258) Very large market shares are in themselves, save in exceptional circumstances, evidence of a dominant position (93). The Court of Justice in Akzo (94) held that under normal circumstances a 50 % market share is sufficient to infer dominance. In Hilti (95) the Court of First Instance stated: 'In this case it is established that Hilti holds a share of between 70 % and 80 % in the relevant market. Such a share is, in itself, a clear indication of the existence of a dominant position in the relevant market`.

(259) HB has for a long time had a share in volume and value of over 75 % of the relevant market (see paragraph 27). By virtue of its volume of production and scale of supply, this undertaking enjoys a position of great economic strength. For many retailers stocking HB ice cream, HB's competitors do not constitute reasonable alternative sources of supply, in view of the range and popularity of HB's products. HB's competitors would be unable effectively to meet the expectations and demand requirements of such retailers. In practice, therefore, HB is an unavoidable trading partner for many retailers in the relevant market, as illustrated by the fact that the large majority of retailers all choose to sell HB ice cream, and many of these exclusively so. HB's position of strength is further confirmed by the relative market shares of its rivals, as well as by their more limited product ranges (96). This secures for HB, at least for the foreseeable future, that freedom of action (97) which is the special feature of a dominant position (98). The freedom of action enjoyed by HB is further enhanced by the fact that, in some 40 % of retail outlets in the relevant market, which outlets account for some 40 % of impulse ice cream sales in the relevant market, it is the sole impulse ice-cream supplier (see paragraph 159).

(260) In addition to HB's very large market share, consideration has to be given to other factors which contribute to its economic strength. HB is part of a multinational group of companies which has been producing and distributing ice cream for many years in all Member States and in many other countries, in many of which the Unilever group companies are the market leaders. Consequently, it has access to long established know-how concerning production and distribution of these products, acquired from its experience in these areas. No other operator in the relevant market has this degree of relevant knowledge and experience. Furthermore, HB has a strong position on the neighbouring frozen food market and other undertakings of the Unilever group of companies enjoy similar positions in the food market as a whole, giving it an added advantage in its dealings with the grocery trade, the most important distribution channel for impulse ice cream in Ireland (paragraph 39 et seq.). In addition, HB has an extensive product range including most of the leading brands of impulse ice cream in the relevant market, and a distribution network which provides nationwide coverage. It is also clear that HB, by virtue of its huge share of the market, achieves scale economies in distribution which cannot be enjoyed to the same extent by its competitors.

(261) Having regard to all these factors, it can be concluded that HB has a dominant position in the market for impulse ice cream in Ireland.

2. Abuse of dominant position

(i) The concept

(262) The Court of Justice has defined an abuse of a dominant position as 'an objective concept relating to the behaviour of an undertaking in a dominant position which is such as to influence the structure of a market where, as a result of the very presence of the undertaking in question, the degree of competition is weakened and which, through recourse to methods different from those which condition normal competition in products or services on the basis of the transactions of commercial operators, has the effect of hindering the maintenance of the degree of competition still existing in the market or the growth of that competition` (99). In particular, small competitors should not be the victims of behaviour by a dominant firm, facilitated by that firm's market power, which is designed to exclude those competitors from the market or which has such an exclusionary effect.

(ii) Infringement

(263) HB abuses its dominant position in the relevant market, contrary to Article 86, in that it induces retailers (as described at paragraph 266) who do not have a freezer cabinet for the storage of impulse ice cream either procured by themselves or provided by another ice-cream supplier than HB to enter into freezer-cabinet agreements subject to a condition of exclusivity. The inducement takes the form of an offer to supply the freezer cabinets to retailers, and to maintain them, at no direct charge to the retailer.

(264) Given the difficulties inherent in persuading retailers either to replace such cabinets or to install additional cabinets in those outlets (described in detail under Article 85(1)), HB's inducement of retailers to enter into freezer-cabinet agreements subject to a condition of exclusivity, as described in the previous paragraph, has the effect of rendering those outlets de facto exclusive sellers of HB impulse ice-cream products. This strengthens HB's dominant position, and as such infringes Article 86. Any inducement by a dominant supplier of a customer to grant it exclusivity, so as to prevent competing suppliers over significant periods from dealing with the customer, is prohibited by Article 86, as the Court of Justice and the Court of First Instance have in a number of cases confirmed (100).

(265) Where, as in the present case, an economic operator holds a dominant position in the market, exclusive supply (whether this is the object or the effect (101) of agreements concluded by this operator) constitutes an unacceptable obstacle to entry into the market and impairs the effective competitive structure envisaged by Article 3g of the Treaty (102). HB's freezer-cabinet agreements in outlets with only HB cabinets have the effect of eliminating the freedom of a very substantial number (103) (some 40 % of all outlets in the relevant market, which account for some 40 % of impulse ice cream sales in the relevant market, see paragraphs 156 and 184) of retailers to stock and offer for sale to the consumer ice cream of any competing suppliers. These retailers, as a result of the agreements, obtain all their requirements exclusively from HB. Consequently, access to the relevant market for other suppliers is made more difficult. The effect of a freezer-cabinet agreement subject to the condition of exclusivity is, in these circumstances, the same as that of any other measure taken by a dominant supplier which excludes its competitors from dealing with that retailer (104). Although the concept of abuse is an objective one, it may be noted that HB is not only aware of this exclusionary effect produced by its freezer-cabinet agreements but has indeed targeted it (see paragraph 64 et seq.).

(266) By inducing retailers to grant exclusivity in the circumstances mentioned at paragraph 263, HB is behaving in a manner different from the conditions governing normal competition in consumer goods. This practice is attractive to retailers, thereby amounting to an incentive, because of its general convenience, in particular the fact that the cabinet is also maintained by HB, and because it relieves retailers of concerns about capital requirements, including the risk inherent in such an investment. Moreover, it provides retailers with a cabinet, the cost of which reflects HB's scale economies in providing the service, a cost which no individual retailer could match if he chose to purchase his own cabinet. However, it interferes with the retailers' freedom to choose suppliers on the basis of the merits of the products which they offer. The exclusivity resulting from the inducement not only harms HB's competitors in the relevant market by making market penetration and expansion more difficult, but is also contrary to the interests of the retailers and ultimately the consumers, in that the former are prevented from exercising their freedom of choice in the products they may stock as well as in how they maximise outlet space efficiency, and the latter as to which products they may purchase. HB therefore abuses its dominant position within the meaning of Article 86 (105).

(267) HB has argued that freezer-cabinet agreements are normal commercial usage in the specific market under analysis in these proceedings and therefore do not constitute an abuse within the terms of Article 86. It may be true that the use of such agreements is standard practice in the relevant market and that these contracts may be advantageous to both contracting parties in that the supplier secures a minimum level of sales and the retailer can rely on security of supply and the benefit of the facility provided. Those considerations, however, which may apply in a normal competitive market situation, cannot be unreservedly accepted in the case of a market where, because of the dominant position of a firm, the degree of competition is already weakened. An undertaking in a dominant position has a special responsibility not to allow its conduct to impair genuine undistorted competition in the common market (106).

(268) Furthermore, HB has contended that a prohibition of its cabinet agreements on the basis of Article 86 would amount to condemning it to harm its own interests and penalising it merely for having a dominant position. It points to the Court of First Instance's view that the mere fact that an undertaking in a dominant position enters into agreements restricting competition does not per se constitute an abuse within the meaning of Article 86 (107). In relation to this point it should be stressed that, for the purpose of applying Article 86, the circumstance surrounding the agreements and particularly their effect on the structure of competition in the relevant market must be taken into account in establishing the existence of an abuse, as has been done at paragraphs 264, 265 and 266.

(269) HB has put forward the argument that the exclusivity provision in its freezer-cabinet agreements is justified in the interest of protecting its property from being used by other ice-cream suppliers. In this context, HB considers that Community law may, exceptionally, treat aspects of property rights, such as contractual provisions, as abusive in circumstances where they could be described as 'unfair` or 'unreasonable` (108). It is, however, HB's contention that the usual terms of its cabinet agreement, including the cabinet exclusivity provision, are not liable to be considered as 'unfair`.

(270) In addition to the earlier arguments advanced concerning the exercise of the right to property in the context of Article 85 (see paragraphs 211 to 220) it must be noted that the retailers pay a rent, included in the ice-cream price, as consideration for the leasing of the equipment which HB has made available to them. In such circumstances, an additional exclusivity provision concerning the use of the property by the lessee cannot be justified by reference to property rights. The exclusivity granted is for the benefit of the supplier alone, at the expense of the retailer's freedom of choice, and can in that sense be regarded as unfair. HB uses its property as an inducement to grant exclusivity in relation to the purchase of goods, with the result that this use constitutes in itself an abuse within the meaning of Article 86 (109).

3. Effect on trade between Member States

(271) The abuse by HB of its dominant position is liable to affect trade between Member States in the same way as was discussed in relation to Article 85 (paragraph 201).

IV. ARTICLE 3 OF REGULATION No 17

(272) In response to the applications by Mars and Valley, and in response to the application for negative clearance or, in the alternative, exemption by HB for its distribution arrangements, the Commission, for the reasons outlined above, finds pursuant to Article 3(1) of Regulation No 17 that, in Ireland, and as regards single-wrapped items of impulse ice cream sold through retail outlets:

(a) the exclusivity provision in HB's freezer-cabinet agreements relating to cabinets installed in outlets where the only freezer cabinets in place in those outlets have been provided by HB, constitutes an infringement of Article 85(1),

and

(b) HB's inducement to retailers who do not have a freezer cabinet/s either procured by themselves or provided by an ice-cream supplier other than HB, to enter into freezer-cabinet agreements subject to a condition of exclusivity by offering to supply the freezer cabinets to retailers, and to maintain them, at no direct charge to the retailer, constitutes an infringement of Article 86.

(273) The freezer-cabinet agreements referred to in point (a), containing the exclusivity provision, are still in force. HB also continues to offer cabinets on the terms referred to in point (b). The Commission therefore requires HB, pursuant to Article 3(1) of Regulation No 17, to bring those infringements to an end. The Commission further requires HB to inform the parties with whom it has concluded the agreements referred to in point (a) that each of the said agreements is void, and that HB's inducement of them to enter into the agreements in question constitutes, in the circumstances, an abuse of its dominant position contrary to Article 86.

(274) HB has expressed the opinion that the Commission, by continuing with the present proceeding, is not acting in accordance with the principle of subsidiarity as tangibly expressed, in the field of competition law, by the notice of 13 February 1993 on cooperation between national courts and the Commission (110). HB has argued that the present case, which concerns the Irish market, has already been fully analysed and is the subject of a decision by the Irish High Court (paragraph 3), which held that neither Article 85 nor Article 86 had been infringed.

(275) Freezer exclusivity is a contractual practice applied by the majority of ice-cream manufacturers in all parts of the Community. The legality of this practice is challenged not only by the two complainants in this case, but also by Mars in a supplementary complaint concerning the German market. Freezer exclusivity in the United Kingdom impulse ice-cream market has also been the subject of a recent report by the Monopolies and Mergers Commission (111). Finally, a German ice-cream manufacturer has notified standard freezer- cabinet agreements to the Commission. The Commission is obliged to state its position in relation to such notifications. In these circumstances, a Commission decision is appropriate in order to ensure that the Community competition rules are applied consistently to various forms of exclusivity practised by ice-cream manufacturers throughout the Community.

(276) In its notice of 13 February 1993 on cooperation between itself and national courts in applying Articles 85 and 86 of the Treaty, the Commission stated that there will not normally be a sufficient Community interest in examining a case when the plaintiff is able to secure adequate protection of his rights before the national courts (at least when those courts are concurrently seised of the question, judgment of the Court of First Instance in Case T-24-90 Automec II (112). The Commission intends to concentrate on proceedings having particular political, economic or legal significance for the Community.

(277) Where a case does have that kind of significance for the Community, the Commission may take action itself even if proceedings have already been brought before a national court. In deciding whether or not it will do so, the consideration that the case is significant for the Community prevails over the consideration that the particular complainant may be able to obtain a remedy in national proceedings or that it has already taken proceedings.

(278) This case does have particular political, economic and legal significance for the Community, since it addresses fundamental questions about business practices which are found throughout the Community, not only in the ice-cream market, but also (as HB itself has pointed out) in many other markets where suppliers provide equipment to resellers in connection with the sale of goods. It therefore belongs to the category of cases which the Commission will continue to investigate itself, even if national courts have already ruled on them.

(279) In its judgment of 28 May 1992 (paragraph 3), the Irish High Court referred to the Commission decisions of 25 March 1992 granting interim measures (113). The High Court also considered the option of contacting the Commission, as well as the possibility of referring a question to the Court of Justice under Article 177 of the EC Treaty. Exercising its discretion in this matter, the High Court did not find either of these options necessary for it to give judgment. It is not inconsistent with the principles governing the concurrent powers of the national courts and the Commission in the application of Article 85(1) and Article 86 of the Treaty, for the Commission to take a decision which differs from a judgment delivered by a national court, provided that there exists a sufficient Community interest in doing so. As has already been said, in the present case this interest consists in settling fundamental questions about business practices which are found throughout the Community.

(280) As to HB's concern with the existence of conflicting decisions, and with the legal uncertainty allegedly resulting from this, it should be pointed out that it is for the Court of Justice to give a final interpretation of the Treaty Articles involved. Any Commission decision is susceptible to judicial review by the Court of First Instance and ultimately, on points of law, by the Court of Justice. Under Article 177 of the Treaty, the Irish Supreme Court, before which an appeal against the decision of the High Court is pending, may also ask the Court of Justice for a preliminary ruling,

HAS ADOPTED THIS DECISION:

Article 1

The exclusivity provision in the freezer-cabinet agreements concluded between Van den Bergh Foods Limited and retailers in Ireland, for the placement of cabinets in retail outlets which have only one or more freezer cabinets supplied by Van den Bergh Foods Limited for the stocking of single-wrapped items of impulse ice cream, and not having a freezer cabinet either procured by themselves or provided by an ice-cream manufacturer other than by Van den Bergh Foods Limited constitutes an infringement of Article 85(1) of the EC Treaty.

Article 2

The request by Van den Bergh Foods Limited for an exemption of the exclusivity provision described in Article 1 pursuant to Article 85(3) of the EC Treaty is hereby rejected.

Article 3

Van den Bergh Foods Limited's inducement to retailers in Ireland not having a freezer cabinet either procured by themselves or provided by an ice-cream manufacturer other than by Van den Bergh Foods Limited, to enter into freezer-cabinet agreements subject to a condition of exclusivity by offering to supply to them one or more freezer cabinets for the stocking of single-wrapped items of impulse ice cream, and to maintain the cabinets, free of any direct charge, constitutes an infringement of Article 86 of the EC Treaty.

Article 4

Van den Bergh Foods Limited is hereby required immediately to cease the infringements set out in Articles 1 and 3, and to refrain from taking any measure having the same object or effect.

Article 5

Van den Bergh Foods Limited is hereby required, within three months of notification of this Decision, to inform retailers with whom it currently has freezer-cabinet agreements constituting infringements of Article 85(1) of the EC Treaty as described in Article 1 of the full wording of Articles 1 and 3, and to notify them that the exclusivity provisions in question are void.

Article 6

This Decision is addressed to:

Van den Bergh Foods Limited, Whitehall Road, Rathfarnham, Dublin 14, Ireland.

Done at Brussels, 11 March 1998.

For the Commission

Karel VAN MIERT

Member of the Commission

(*) Business secret

(1) OJ 13, 21.2.1962, p. 204-62.

(2) OJ 127, 20.8.1963, p. 2268-63.

(3) Case No IV-34.073.

(4) Masterfoods v. HB Ice Cream, [1992] 3 CMLR 830.

(5) Case No IV-34.395.

(6) Case No IV-35.436.

(7) Commission press release of 8 March 1995 IP-95-229.

(8) OJ C 211, 15.8.1995, p. 4; publication pursuant to Article 19(3) of Regulation No 17, which contained a summary of HB's distribution arrangements as revised, inviting the comments of interested third parties. Comments were received from Mars.

(9) Noel McEneaney, HB's Marketing Manager is quoted in the March 1996 issue of the Irish retail magazine 'Shelflife` as saying: 'There is a rule of thumb that for every three people who purchase an impulse product, only one entered the shop with the intention of buying an ice cream. We try to maximise this factor by ensuring that the first thing the customer sees when he or she enters the shop is an ice-cream cabinet, either inside the entrance or next to the cash register`.

(10) Source: AC Nielsen, a leading market-research company. The value figures are for retail sales, and are not the same as the GSV figures referred to elsewhere; GSV figures include neither the retailers' margin nor the VAT rate. The figures shown above include 'water ices`.

(11) According to HB, market information on this sector is not publicly available.

(12) Overnight, HB's share of the ice-cream market increased from 42 % to 80 %. In return, Premier Dairies acquired HB's milk business.

(13) These market-share figures are derived from AC Nielsen which, for the purpose of calculating those figures, uses the same market definition as in the present Decision, with the exception of the inclusion of 'water ices`, this inclusion does not, however, affect the proportional shares of the various market participants to any significant extent.

(14) AC Nielsen's impulse ice cream audit is the only regular reliable source of market data concerning the distribution of impulse ice cream in Ireland. The outlets included in Nielsen's 'Universe` are the same as those surveyed for its ambient confectionery audit and excludes some marginal ice cream outlets such as cinemas, theatres and ice-cream stands.

(15) The market survey carried out by Lansdowne in the summer of 1996, described in detail below, found that there was only minimal freezer movement to satisfy seasonal demand variation (see paragraph 97).

(16) The real figures for unweighted/numeric distribution are somewhat higher, because not all the outlets surveyed stock impulse ice cream. For example, only 93 % of all the outlets surveyed in this time period were selling impulse ice cream. In this instance, therefore, it is more accurate to adjust the figure of 79 % to 79-93 = 85 %. A similar adjustment can be made for all the time periods, but has not been for the purposes of this Decision. The same distortion does not exist for the weighted figure.

(17) The storage of HB's take-home ice-cream products is also permitted.

(18) HB has, however, indicated that it does not in practice, seek to enforce the requirement of two months' notice for retailers wishing to terminate on shorter notice or without notice.

(19) Some cabinets are kept in the rear of the outlet for storage purposes.

(20) HB has indicated, however, that, in practice, it does not always insist on strict compliance with these criteria as a necessary precondition for receipt of an HB freezer cabinet.

(21) Fozen products coordination.

(22) It should be determined in accordance with objective criteria, for example, the Bank of Ireland lending rate to small businesses.

(23) The cost of maintenance was calculated as amounting to IEP [. . .] in 1995, and so the sum paid to hire-purchasers was IEP [. . .] p.a. in that year. HB was to be flexible in relation to payment of the bonus to those who chose to hire-purchase during the course of 1995.

(24) Only [. . .] were sold during 1995, and most of these in the later part of the year.

(25) The average lifetime of a cabinet is estimated by HB to be 10 years; cabinets are amortised by HB after eight years.

(26) HB states that it 'undertook a programme designed to maintain the image of the HB brand and the visibility of the HB product range by, inter alia, replacing ageing cabinets with more modern units`; Mars states that 'Unilever ran an exceptional freezer placement programme in 1994 and 1995; a comparison between the two Rosslyn surveys shows that the 1994-95 replacement rate was over 230 % of the 1990-91 replacement rate`. Mars goes on to say that 'the new Unilever freezers are all larger than the ones that they replace (save one type which has not been widely placed)`.

(27) 68 % of the sample were grocery/convenience stores, 21 % were TSNs, 11 % were garages. Multiples were not included. 92 % of the sample were independently owned; a further 6 % were part of a chain (symbol group).

(28) Includes cabinets owned by HB agents.

(29) Includes cabinets owned by Dale Farm agents.

(30) Includes 2 % (of all outlets) where cabinets are owned by HB agents.

(31) Includes 1 % (of all outlets) where cabinets are owned by Dale Farm agents.

(32) 15 % of the outlets refused to disclose their turnover.

(33) In the Statement of Objections, this figure was mistakenly quoted in the text as 65 % (a figure from the preliminary survey results); the survey was annexed in full to the Statement of Objections.

(34) Ibid. mistakenly quoted in the Statement of Objections as 37 %.

(35) Ibid. mistakenly quoted in the Statement of Objections as 28 %.

(36) Ibid. mistakenly quoted in the Statement of Objections as 22 %.

(37) Ibid. mistakenly quoted in the Statement of Objections as 16 %.

(38) Ibid. mistakenly quoted in the Statement of Objections as 6 %.

(39) An almost identical survey was carried out for Mars by the same company in December 1991/January 1992, the results of which were cited in the Commission's Statement of Objections of 29 July 1993. All but 40 were the same outlets as interviewed in 1991.

(40) 79 % in 1991.

(41) 20 % in 1991.

(42) 1 % in 1991.

(43) 1,22 in 1991.

(44) 89 % in 1991.

(45) 11 % in 1991.

(46) 71 % in 1991.

(47) 68 % of 'back of shop` freezers.

(48) 1 % in 1991.

(49) As no suppliers appear to supply cabinets on non-exclusive terms, it is most likely that the remaining 12 % were also subject to an exclusivity condition of which the retailer was unaware.

(50) 52 % in 1991.

(51) 44 % in 1991.

(52) The calculation is based on the fact that, in the 408 outlets sampled, no more than 45 cabinets were not supplier-provided (they were retailer-owned, 'other owner` or 'don't know`), if any outlet contained more than one such cabinet, the foreclosure level would be even higher.

(53) 234 outlets had only a supplier-provided cabinet, 87 had one HB cabinet and one other supplier-provided cabinet, 23 outlets had two HB cabinets: these 344 outlets represent 84 % of the total sample.

(54) This was the method of calculation used by William Bishop, an economist who analysed foreclosure levels on the basis of the 1991 Rosslyn survey: the method involves dividing the total number of outlets sampled by the mean number of freezers present in all these outlets, and then multiplying by the number of retailer- and 'other`-owned cabinets. This method assumes that retailer-owned cabinets are likely to be spread across outlets in the same proportion as any other cabinets.

(55) 20 % were TSNs; 63 % were grocery outlets, of which 10 % were 'symbol` stores; 10 % were garages; others 7 %. Multiples were excluded.

(56) Dale Farm acts as distributor for Mars in Northern Ireland.

(57) See Case 23-67 Haecht I [1967] ECR 407, at p. 415.

(58) Case 27-76 United Brands [1978] ECR 207, at paragraph 12.

(59) It is generally accepted that non-ice-cream products do not fall into the same product market/s as ice cream. See judgment of Irish High Court; see footnote 4.

(60) Owing to the insignificant quantities of 'artisan` ice cream sold in Ireland, the distinction between it and industrial ice cream does not need to be made for the purposes of this assessment. All ice cream referred to is industrial ice cream.

(61) Case C-234-89 Delimitis [1991] ECR I-935 paragraph 17; Cases T-7 Langnese, paragraph 63 and T-9-93 Schöller, paragraph 42, [1995] ECR II-1532.

(62) Single-wrapped items can also be used for catering purposes.

(63) Cases T-7 and T-9-93 Langnese & Schöller, paragraphs 64 and 65 and 43 and 44.

(64) Case 322-81 Michelin [1983] ECR 3461, paragraph 37.

(65) See Delimitis, paragraph 18 in relation to the German beer market; Case T-9-93 Schöller, paragraph 54 where the Court of First Instance (CFI) accepted that Germany was the correct geographic market for impulse ice cream.

(66) Case C-234-89 Delimitis at paragraphs 19 to 24, Cases T-7-93 and T-9-93 Langnese and Schöller, at paragraphs 129 to 131 and 95 to 99.

(67) The CFI in Langnese and Schöller stated at paragraphs 129 and 95 that '. . . where there is a network of similar agreements concluded by the same producer, the assessment of the effects of that network on competition applies to all the individual agreements making up the network`.

(68) In practice, given that all suppliers in the relevant market operate cabinet exclusivity, this means a cabinet owned by the retailer himself or on loan from a source other than an ice-cream supplier.

(69) That is following the disposal by HB of a portion of its fleet of cabinets (see paragraphs 72 and 73).

(70) That is before the disposal by HB of a portion of its fleet of cabinets (see paragraphs 72 and 73).

(71) The disposal of in situ cabinets by HB during the first half of 1996 will not have affected this assessment to any significant degree; apart from the fact that the cabinets disposed of have an average age of nine years (i.e. they are at least approaching the end of a cabinet's normal lifespan), the disposals took place primarily in low-turnover outlets (see paragraph 73).

(72) It can be seen from the foregoing that this is a conservative figure.

(73) Where an incumbent supplier-exclusive cabinet is replaced by a cabinet exclusive to a competitor of the incumbent supplier, the outlet becomes foreclosed to the (former) incumbent, provided it was his only cabinet in the outlet, and remains foreclosed to any third party suppliers not already present in the outlet.

(74) The return of the HB freezer cabinet may also have the consequence that a retailer will have to give up stocking HB's take-home ice-cream products, as these are sometimes kept in the same freezer cabinet as impulse products (paragraph 42); this additional consequence renders the option of replacing an HB freezer cabinet by that of another manufacturer even more unattractive.

(75) HB's apparent disregard of this requirement (see above) can only be considered a partially mitigating factor in this regard, given possible retailer ignorance or uncertainty about the practice.

(76) The B& A 'Qualitative study` finding that retailers generally operate on the basis of instinctive, rather than more precise, methods of calculating profit return potential (see paragraphs 122 and 123) does not undermine the validity of this assertion: their instinctive calculations will take the same basic considerations into account.

(77) See Commission Decision 78-172-EEC in Liebig (OJ L 53, 24.2.1978, p. 20, at paragraph 9).

(78) This is confirmed by the B& A 'Qualitative study`, which found that retailers favour frequent restocking of a freezer over the installation of an additional one (see paragraph 123).

(79) 7 % according to Lansdowne; 4 % according to Rosslyn; [. . .] % according to B& A, a survey weighted significantly toward larger outlets (see paragraph 86).

(80) Where an additional supplier-exclusive cabinet is installed, the outlet remains foreclosed to any third party suppliers not already present in the outlet.

(81) 56 % according to the Lansdowne survey at paragraph 92; at least 60 % according to the Rosslyn survey at paragraph 108. HB faces direct intra-outlet competition from other ice cream brands in [. . .] of the outlets surveyed by B& A in which HB products are sold (see paragraph 116).

(82) In June/July of 1995, Mars ran a special price promotion whereby it reduced the recommended retail prices (and hence also trade prices) of its bar products by over 20 %: this promotion was not responded to by HB - see Mars letter of 9 August 1996. Mars ran the same promotion again in the summer of 1996; again, it has not been responded to; nor did it appear to affect HB's market share, which increased during the period.

(83) See Langnese judgment at paragraphs 119 to 124.

(84) Case 234-89.

(85) HB has stated that 'the cabinet exclusivity provision is necessary to secure the full benefits of the system`, and goes on to say, inter alia, that it is not 'unreasonably restrictive` and so falls outside Article 85(1) as an ancillary restraint.

(86) This issue is further discussed in relation to Article 86.

(87) Case 44-79 Hauer v. Land Rheinland Pfalz [1979] ECR 3727, at paragraph 18.

(88) OJ L 173, 30.6.1983, p. 5.

(89) OJ L 31, 9.2.1996, p. 2.

(90) See also Commission Decision 88-491-EEC, Aalsmeer Bloemenveilingen (OJ L 262, 22.9.1988, p. 27).

(91) See Joined Cases 56 and 58-64 Consten/Grundig [1966] ECR 299, at p. 348; see also the Langnese judgment at paragraph 180.

(92) Case 27-76 United Brands and Case 85-76 Hoffmann-La Roche [1979] ECR 461.

(93) Case 85-76 Hoffmann-La Roche [1979] ECR 461, at paragraph 41.

(94) Case C-62-86 Akzo Chemie [1991] ECR I-3439, at paragraph 60.

(95) Case T-30-89 [1991] ECR II-1439, paragraph 92 (confirmed on appeal by the Court of Justice, Case C-53-92 P, judgment of 2 March 1994 [1994] ECR I-667).

(96) Case 27-76 United Brands.

(97) The price promotion introduced by Mars in the summer of 1995 (see footnote 83) was not responded to by any of its competitors.

(98) Case 85-76 Hoffmann-La Roche, at paragraph 41.

(99) Cases 85-76 Hoffmann-La Roche at paragraph 91.

(100) Cases 85-76 Hoffmann-La Roche; C-62-86 Azko; T-65-89 BPB and British Gypsum [1993] ECR II-389.

(101) Cases 85-76 Hoffmann-La Roche, at paragraph 91; 322-81 Michelin, at paragraph 70.

(102) Hoffmann-La Roche, at paragraph 89; BPB and BG, at paragraph 68; cited above.

(103) In T-65-89 BPB and British Gypsum, at paragraph 68, the Court said that '. . . the conclusion of exclusive supply contracts in respect of a substantial proportion of purchases constitutes an unacceptable obstacle to entry to that market . . .`.

(104) Cases 85-76 Hoffmann-La Roche, at paragraph 89; 322-81 Michelin, at paragraph 70; C-62-86 Akzo, at paragraph 149; T-65-89 BPB and British Gypsum cited above; C-393-92 NV Energiebedrijf Ijsselmij [1994] ECR I-1477, at paragraph 44.

(105) Cases 85-76 Hoffmann-La Roche, at paragraph 89; C-62-86 Akzo, at paragraph 149.

(106) Case 322-81 Michelin, at paragraph 57; T-65-89 BPB and British Gypsum, at paragraph 67.

(107) Case T-51-89 Tetra Pak, at paragraph 24.

(108) Case 247-86 Alcatel [1988] ECR 5987 at paragraph 10.

(109) See in this respect Commission Decision 88-589-EEC London European/Sabena (OJ L 317, 24.11.1988, p. 47); and the Commission's decision of 11 June 1992, B& I/Sealink, (not published), Competition Report 1992 No 219 and [1992] 5 C.M.L.R. 255.

(110) OJ C 39, 13.2.1993, p. 6.

(111) Report on the supply in the UK of ice cream for immediate consumption, April 1994.

(112) [1992] ECR II-2223.

(113) Decisions in Cases IV-31.533 and IV-34.072 Schöller and Langnese-Iglo.