England and Wales Court of Appeal (Civil Division) Decisions

IN THE COURT OF APPEAL (CIVIL DIVISION)
ON APPEAL FROM THE COMPETITION APPEAL TRIBUNAL
Mr Andrew Lenon QC, Mr Michael Cutting, Ms Jane Burgess
[2019] CAT 20

		Royal Courts of Justice Strand, London, WC2A 2LL
		05/03/2020

Lord Justice Henderson and Lady Justice Asplin

Introduction

1. This appeal raises important issues about the scope of competition law where the conduct complained of affects those who are not in a direct contractual relationship with the alleged perpetrator of the anti-competitive conduct. It is concerned with the schemes operated by or on behalf of the Appellant, Network Rail Infrastructure Limited ("Network Rail"), which impose rules upon individuals or companies which wish to work upon Network Rail's managed mainline railway infrastructure.
2. The appeal is from a decision of the Competition Appeal Tribunal (the "CAT") dated 19 July 2019, the citation of which is [2019] CAT 20. The CAT decided that Network Rail had breached the prohibitions in Chapter I and (assuming dominance in the market) in Chapter II of the Competition Act 1998 ("the 1998 Act") by the inclusion of a rule within the Sentinel Scheme and the On-Track Plant Operations Scheme ("OTPO") (referred to together as "the Schemes") requiring suppliers providing works and/or services on Network Rail's managed infrastructure to be vetted by the Rail Industry Supplier Qualification Scheme ("RISQS") which is run by the Rail Safety and Standards Board ("RSSB").
3. Network Rail is a public sector company registered in England and Wales. It is the owner and operator of most of the mainline railway infrastructure in Great Britain. It sells services to train operators in return for charges and operates the railway network pursuant to a licence granted by the Secretary of State for Transport. It is regulated by the Office of Rail and Road ("ORR"), an independent statutory body.
4. The Respondent, Achilles Information Limited, ("Achilles") is a provider of supplier assurance services in competition with RISQS. It challenged the requirement that vetting be conducted only by RISQS once its concession to run RISQS had come to an end in May 2019 and the core IT and audit services had been transferred to competing providers following a competitive tender process organised by the RSSB, from which Achilles withdrew in May 2017. Achilles provides supplier assurance services in a number of industries in the United Kingdom and overseas. From 1997 until 2014 it was the sole provider of a rail industry qualification scheme known as Achilles Link-Up which was the precursor to RISQS. From 2014 until 2018 it was the sole operator of RISQS. It now wishes to continue to provide a supplier assurance scheme to the rail industry in Great Britain but is, in part, prevented from doing so because of the exclusive status conferred on RISQS under the Schemes and Network Rail's refusal to recognise supplier assurance schemes provided by any other undertaking.
5. The RSSB is a not-for-profit company limited by guarantee. It oversees rail safety and standards on the rail network. It is governed by a Board which includes representatives from industry bodies including infrastructure owners, contractors, train operating companies and rail trade associations. It provides a range of services and products to its members in return for a membership levy and charges for some of its services.
6. RISQS is the main supplier assurance service used by buyers of products and services throughout the rail industry in the United Kingdom. It is used by Network Rail itself, TfL, passenger, light rail and freight operators, other train operating companies, rolling stock organisations, main infrastructure contractors and other buyers and suppliers of rail products and services. They wish to establish that suppliers with whom they contract are suitably competent and adequately resourced and can deliver products and services to the appropriate specification. There is a rail industry standard published by the RSSB which defines standards for providers of supplier assurance to the rail industry in Great Britain.
7. Governance of RISQS is provided through the RSSB by the RISQS committee which is comprised of representatives from across the rail industry, including Network Rail. RISQS is not governed by Network Rail, therefore, nor is it owned by Network Rail. Furthermore, Network Rail derives no commercial benefit or competitive advantage from its use.
8. Pursuant to its powers as the operator of the rail network, Network Rail has developed schemes including those under consideration in this case which impose terms on persons wishing to supply Network Rail itself or to have access to its infrastructure. As we have already mentioned, the two schemes which are relevant for the purposes of this appeal are the Sentinel Scheme and OTPO.
9. The Sentinel Scheme governs access by individuals to Network Rail's infrastructure. Data relating to an individual's competence and fitness to work on the infrastructure are recorded on a database. Individuals must be sponsored by an organisation approved by Network Rail. Sponsors are required to have in place a process for undertaking pre-sponsorship checks for all individuals. The rules of the Sentinel Scheme provide that in order to be approved by Network Rail as a "sponsor" an organisation must initially register with RISQS. It is said that the Sentinel Scheme applies to around 1,500 potential suppliers and approximately 175,000 sponsored individuals.
10. The OTPO Scheme sets out processes to support safe planning, control and use of on-track plant. A supplier who needs to operate on-track plant on Network Rail's infrastructure is required to be approved under the OTPO scheme which requires on-track plant suppliers to be approved by Network Rail through RISQS. It applies to around 150 potential suppliers.
11. The effect of the Schemes, therefore, is that any sponsor of individuals carrying out trackside work or any supplier operating on-track plant on Network Rail's infrastructure is required to be accredited by RISQS whether or not the works are carried out pursuant to a contract with Network Rail (the "RISQS-only rule"). Network Rail's licensing and authorisation teams ensure that undertakings wishing to obtain a licence under one of the Schemes satisfy the requirements set out in the relevant level of assurance.
12. Network Rail also uses RISQS as the supplier assurance provider in relation to suppliers with which it contracts directly. It uses it as a pre-qualification system for would-be suppliers of infrastructure, maintenance or construction services. The scheme which applies in such circumstances is the Principal Contractor Licensing Scheme ("PCLS"). The PCLS also includes terms which require any organisation wishing to contract to be assessed and verified by RISQS in order to confirm compliance with various RISQS standards. The PCLS was included within the definition of "Key Schemes" in the CAT judgment and, initially, it was alleged that it too fell foul of Chapters I and II of the Competition Act 1998. It became clear, however, towards the end of the hearing before the CAT that Achilles accepts that Network Rail may choose to use RISQS exclusively for supplier assurance in relation to the contracts which it enters into itself. Therefore, by implication, the RISQS-only rule in the PCLS does not constitute a breach of the Chapter I and Chapter II prohibitions: see paragraphs [6], [28] and [44] of the CAT judgment.
13. This appeal, therefore, is only concerned with the CAT's decision in relation to the effects of the Sentinel and OTPO schemes and Achilles' objection to the requirement which Network Rail imposes through those Schemes upon others who require access to the rail infrastructure to use RISQS for supplier assurance, whether or not they have any direct contractual relationship with Network Rail itself.
14. After a trial lasting ten days, during which the CAT, a specialist tribunal (comprising Mr Andrew Lenon QC (Chairman), Mr Michael Cutting and Ms Jane Burgess) heard complex expert and factual evidence, they delivered a very detailed and meticulous judgment of 105 pages and 316 paragraphs, dated 19 July 2019. The CAT held that the terms of the Schemes containing the RISQS-only rule infringe section 2 of Chapter I of the 1998 Act and also constitute an abuse of a dominant position for the purposes of Chapter II of the 1998 Act (dominance having been assumed.)
15. By an order drawn on 20 September 2019 ("the Order") it was ordered that: (i) Network Rail cease to impose on suppliers or persons seeking access to its infrastructure under the Sentinel and OTPO schemes a requirement to obtain supplier assurance only from RISQS and not through alternative schemes, save that it may do so in relation to suppliers contracting directly with it, but may not require such contractors to impose such a requirement on any sub-contractors (the proviso); (ii) that Network Rail accept any supplier assurance through alternative schemes that are equivalent to the assurance obtained through RISQS in relation to granting access to its infrastructure or granting permission to operate on-track plant, save where the proviso applies; (iii) that Network Rail shall publish such reasonable and proportionate conditions for recognition as it wishes to impose in accordance with (ii) as soon as reasonably practicable; and (iv) that on publication of the conditions in (iii) Network Rail shall on request confirm and publish its recognition as equivalent of assurance provided by an alternative supplier assurance scheme, such confirmation to be published promptly and not unreasonably withheld. For the purposes of (ii), the Order further provided that:

"Recognition of such equivalence shall be subject only to such reasonable and proportionate conditions as are objectively justified by the need to ensure safety on the railway network."

16. The effect of the Order has been stayed pending the outcome of this appeal.

The Competition Act 1998

17. Before turning to the decision of the CAT in more detail and the grounds of appeal, it is helpful to set out the parts of Chapters I and II of the 1998 Act which are relevant to this appeal. Section 2 of the 1998 Act provides as follows:

"(1) Subject to section 3, agreements between undertakings, decisions by associations of undertakings or concerted practices which –

(a) may affect trade within the United Kingdom, and

(b) have as their object or effect the prevention, restriction or distortion of competition within the United Kingdom,

are prohibited unless they are exempt in accordance with the provisions of this Part.

(2) Subsection (1) applies, in particular, to agreements, decisions or practices which –

(a) directly or indirectly fix purchase or selling prices or any other trading conditions;

(b) limit or control production, markets, technical development or investment;

(c) share markets or sources of supply;

(d) apply dissimilar conditions to equivalent transactions with other trading parties, thereby placing them at a competitive disadvantage;

(e) make the conclusion of contracts subject to acceptance by the other parties of supplementary obligations which, by their nature or according to commercial usage, have no connection with the subject of such contracts.

(3) Subsection (1) applies only if the agreement, decision or practice is, or is intended to be, implemented in the United Kingdom.

(4) Any agreement or decision which is prohibited by subsection (1) is void.

…

(8) The prohibition imposed by subsection (1) is referred to in this Act as "the Chapter I prohibition"

18. Section 9 of the 1998 Act provides as follows:

"(1) An agreement is exempt from the Chapter I prohibition if it –

(a) contributes to –

(i) improving production or distribution, or

(ii) promoting technical or economic progress, while allowing consumers a fair share of the resulting benefit; and

(b) does not –

(i) impose on the undertakings concerned restrictions which are not indispensable to the attainment of those objectives; or

(ii) afford the undertakings concerned the possibility of eliminating competition in respect of a substantial part of the products in question.

(2) In any proceedings in which it is alleged that the Chapter I prohibition is being or has been infringed by an agreement, any undertaking or association of undertakings claiming the benefit of subsection (1) shall bear the burden of proving that the conditions of that subsection are satisfied."

19. Further, section 10(1) of the 1998 Act provides that an agreement is exempt from the Chapter I prohibition if it is exempt from the Community prohibition, amongst other things, by virtue of a "Regulation". "Community prohibition" relevantly means the prohibition contained in Article 101(1) of the Treaty on the Functioning of the European Union ("the TFEU") and "Regulation" means a Regulation adopted by the Commission or by the Council: section 10(10). An exemption under section 10 is termed a "parallel exemption": section 10(3).
20. Section 18, which is contained in Chapter II of the 1998 Act, is concerned with abuse of a dominant position. It provides as follows:

"(1) ... any conduct on the part of one or more undertakings which amounts to the abuse of a dominant position in a market is prohibited if it may affect trade within the United Kingdom.

(2) Conduct may, in particular, constitute such an abuse if it consists in –

(a) directly, or indirectly imposing unfair purchase or selling prices or other unfair trading conditions;

(b) limiting production, markets or technical development to the prejudice of consumers;

. . .

(d) making the conclusion of contracts subject to acceptance by the other parties of supplementary obligations which, by their nature or according to commercial usage, have no connection with the subject of the contracts.

(3) In this section –

"dominant position" means a dominant position within the United Kingdom; and

"the United Kingdom" means the United Kingdom or any part of it.

(4) The prohibition imposed by subsection (1) is referred to in this Act as "the Chapter II prohibition"."

21. Section 60 of the 1998 Act, which is headed "Principles to be applied in determining questions", provides that:

"(1) The purpose of this section is to ensure that so far as is possible (having regard to any relevant differences between the provisions concerned), questions arising under this Part in relation to competition within the United Kingdom are dealt with in a manner which is consistent with the treatment of corresponding questions arising in EU law in relation to competition within the European Union.

(2) At any time when the court determines a question arising under this Part, it must act (so far as is compatible with the provisions of this Part and whether or not it would otherwise be required to do so) with a view to securing that there is no inconsistency between—

(a) the principles applied, and decision reached, by the court in determining that question; and

(b) the principles laid down by the Treaty and the European Court, and any relevant decision of that Court, as applicable at that time in determining any corresponding question arising in EU law.

(3) The court must, in addition, have regard to any relevant decision or statement of the Commission.

(4) Subsections (2) and (3) also apply to—

(a) the [CMA]; and

(b) any person acting on behalf of the [CMA], in connection with any matter arising under this Part.

(5) In subsections (2) and (3), "court" means any court or tribunal.

(6) In subsections (2)(b) and (3), "decision" includes a decision as to—

(a) the interpretation of any provision of EU law;

(b) the civil liability of an undertaking for harm caused by its infringement of EU law."

22. The trial before the CAT proceeded on the assumed basis that Network Rail is in a dominant position in relation to a market related to the relevant market, namely the market for the operation and provision of access to national rail infrastructure. In relation to the allegations under section 18 of the 1998 Act, therefore, the CAT was concerned solely with whether the conduct complained of in relation to the RISQS-only rule in the Schemes was an abuse for the purposes of that section.

The Judgment of the CAT

23. As we have already mentioned, the CAT judgment is detailed and lengthy and reference should be made to it for the full background to this appeal. However, in summary, the CAT approached the matters relevant to this appeal in the following way. First, for the purposes of the claim under section 2(1) of the 1998 Act, it considered whether the Schemes constitute agreements between undertakings, decisions by associations of undertakings or concerted practices. It concluded at paragraphs [99] – [102] of the judgment that the Schemes embody an agreement or concerted practice between Network Rail and the undertakings who wish to have access to Network Rail's managed infrastructure. In particular, in this regard it held that:

"99. . . . Each of the Key Schemes in terms provides for responsibilities and obligations that are undertaken by the undertakings which sign up to them as a condition of being authorised. It does not matter that the agreements are imposed on suppliers rather than freely negotiated. By participating in the Key Schemes, undertakings acquiesce in their provisions. That is sufficient to constitute an agreement, as established by the car dealership cases relied on by Achilles.

100. As to whether the Key Schemes are subject to competition rules, we consider that the distinction sought to be drawn by Network Rail between its economic activity and the regulation of its managed infrastructure is not well founded and is not supported by the judgment of the Court of Justice in FENIN.

101. In FENIN, the issue was whether state-funded management bodies which purchased medical goods and equipment to be used to provide services to patients free of charge in the Spanish public health service were undertakings engaged in economic activity. The Court of Justice held that the management bodies' purchasing activity should not be dissociated from the downstream use to which those purchases were put (which was not an economic activity) and that the management bodies were therefore not undertakings engaged in economic activity.

102. In the present case, allowing access to its managed infrastructure is an essential part of, and not dissociable from, Network Rail's operation of the railway infrastructure which is an economic activity. The fact that in operating the Key Schemes Network Rail is setting out rules with a view to ensuring safety and to comply with regulatory obligations does not take it outside the scope of that activity."

24. The CAT also concluded that the jurisdictional requirement that the agreement between undertakings affects trade within the United Kingdom was clearly met because the conduct in question is that of an undertaking, Network Rail, whose rules affect the procurement of supplier assurance services in the rail industry in Great Britain. See paragraph [104] of the judgment.
25. Having decided that the harm to competition arising from the RISQS-only rule is not so obvious that an examination of its effects could be dispensed with and that, therefore, the RISQS-only rule was not a restriction by object (see paragraphs [105] – [120] of the judgment, and [120] in particular) the CAT went on to consider whether the RISQS-only rule has as its effect the prevention, restriction or distortion of competition for the purposes of section 2(1)(b) of the 1998 Act.
26. The CAT applied the following principles: (i) that in cases where it is not plain and obvious that the object of the agreement in question is to restrict competition, it is necessary to consider the effect of the agreement; (ii) there is no presumption of anti-competitive effect and it is necessary to demonstrate with a reasonable degree of probability that there is an appreciable effect on competition in the sense of an effect which is more than de minimis or insignificant; (iii) that in order to gauge the restrictive effects of an agreement it is necessary to conduct an analysis of its effect on the relevant market or markets; and (iv) that the effect on competition must be demonstrated by reference to the situation which would pertain on the market in the absence of the agreement or restriction in question which requires consideration of the appropriate counterfactual situation. See sub-paragraphs [121(1) – (4)] of the judgment.
27. The CAT went on to conclude at paragraphs [121(5)] and [123] that: an anti-competitive effect can consist in the segmenting of a market and a distortion or restriction of the way that competition operates in that segment of the market, even if competition may continue in other segments; and if the effect of the RISQS-only rule was to reserve to RISQS/RSSB a significant part of demand for supplier assurance, thereby impairing actual or potential competition from other suppliers of such services or schemes, there may be an appreciable restriction of competition. The CAT came to this conclusion on the basis of the decision in Socrates Training Limited v The Law Society of England and Wales [2017] CAT 10 (see paragraph [122] of the judgment) and having rejected Network Rail's attempt to distinguish that authority on the basis that the Law Society, like the OTOC in the decision of the Court of Justice in Case C-1/12 Ordem dos Técnicos Oficiais de Contas v Autoridade da Concorrência EU:C:2013:127 ("the OTOC case"), had reserved a part of the market to itself.
28. The CAT concluded, therefore, that in order to determine whether there was an appreciable effect upon competition, it was necessary first to define the relevant market and thereafter to identify the relevant counterfactual, in order to be able to assess the effect of the conduct complained of. The CAT turned to the definition of the relevant market at paragraphs [124] – [128] of the judgment followed by consideration of the counterfactual at paragraphs [129] – [154].
29. In relation to the relevant market, the CAT set out both the common ground between the economic experts and, at paragraph [127], the differences between the way in which they defined the scope of the relevant market for supplier assurance services. As this forms the basis of the third ground of appeal it will be necessary to consider these paragraphs in more detail below. Suffice it to say at this stage that the CAT stated at paragraph [128] that it preferred the definition of the market advanced by Mr Holt, Network Rail's economic expert, for the reasons he gave, and that this definition corresponded with the way in which the relevant market was defined in Achilles' claim form. The definition adopted was "the market for the provision of supplier assurance services in the GB rail industry".
30. Having considered the evidence of the economic experts and the witnesses at some length, the CAT concluded at [150] that the correct counterfactual "is one in which Achilles would compete with RISQS at least for a time and that its competition would lead to some benefits in terms of lower prices and product differentiation. . ." Its reasoning was as follows:

"151. In this context, the Tribunal attaches significant weight to the fact that Achilles, with its experience and detailed knowledge of the market, wishes to compete with RISQS and believes that it can do so. It has already incurred a substantial proportion of any costs of entry and has a recent presence in the market. The Tribunal accepts that there may well be real benefits to allowing that choice and allowing competition in the market to evolve. These benefits would include not just the general benefits which can be expected to flow from making the quality and pricing of the services currently provided subject to competitive constraints, but also the possibility of solutions better tailored to the needs of buyers and suppliers, whose activities are not limited to the GB rail sector. Whilst it might be argued that the impact of competition from Achilles (and other suppliers of assurance schemes) in other sectors would continue to be felt by members of the RSSB and the community of suppliers in the rail industry who are also active in other sectors even if Achilles was not active in the railway sector, that impact would be less direct.

152. Given that the RISQS-only rule in the Sentinel Scheme and OTPO Scheme has a prima facie restrictive effect on competition, in that it reserves to a single scheme provider a significant segment of the market for supplier assurance to the rail industry, the Tribunal would be reluctant to find that the restriction has no actual effect on the basis of Network Rail's case that only the market would tip in favour of RISQS. It is fundamentally not for Network Rail to make the decision for other buyers and suppliers that they would prefer RISQS to other supplier assurance services.

153. Furthermore, the difficulties facing Achilles in establishing itself as a viable competitor to RISQS are partly attributable to the effects of the RISQS-only rule itself. If Achilles had been allowed to compete with RISQS for business from buyers and suppliers using the Sentinel Scheme and OTPO Scheme from May 2018 onwards, it might be at less of a competitive disadvantage now. The Tribunal considers that the issue whether the RISQS-only rule has an appreciable effect on the market should be assessed by reference to the state of the market as it would have been had the rule never existed rather than by the reference to the state of the market now."

31. Having defined the relevant market and the appropriate counterfactual, the CAT concluded at paragraph [154] of the judgment that:

"Although the scope for price competition (in the absence of loss-leading or cross-subsidisation on a long-term basis) and product differentiation would be limited, we conclude that the RISQS-only rule does cause significant foreclosure of demand in a significant segment of the market for supplier assurances schemes in the GB railway sector and that the RISQS-only rule has an appreciable effect on competition in that market."

32. The CAT then went on to consider whether the RISQS-only rule could be objectively justified. If an agreement is not itself anti-competitive, restraints which can be regarded as necessary for the agreement to be workable or to achieve its purpose will also fall outside the Chapter I prohibition as an ancillary restraint: see paragraph [155] of the judgment. The burden was on Network Rail to establish an objective justification for the RISQS-only rule. It was submitted that it was necessary and proportionate to the goal of health and safety which it pursued through the Schemes.
33. In a section of the judgment which covers 30 pages, the CAT considered all of the detailed factual evidence in relation to safety, the relevant provisions and the rival submissions. It concluded at paragraph [254] that it was not persuaded the RISQS-only rule is objectively justified as being ancillary to the safety purposes of the Schemes. It held that:

". . . Network Rail has not established that those purposes would be impossible to achieve without the RISQS-only rule. The Tribunal considers that those safety purposes could be achieved by alternative providers of supplier assurance services working to the same standards as RISQS and subject to effective monitoring, with their IT platforms linked to RISQS's and/or their data freely accessible to Network Rail and with the RISQS forum open to participation by other providers of supplier assurance services."

This part of the judgment is not the subject of an appeal.

34. The CAT went on to conclude that the RISQS-only rule is not exempted from the Chapter I prohibition pursuant to section 9 of the 1998 Act. It held that the first condition, namely that the requirement to use RISQS must contribute to improving the production or distribution of goods or contribute to promoting technical or economic progress, was not met (see paragraphs [257] – [275], and [275] in particular). Despite not being strictly necessary, the CAT went on, nevertheless, to consider the third and fourth conditions in section 9, having decided that, in the interests of brevity, it was not necessary to consider the second condition (fair share for consumers): see paragraph [276] of the judgment. The CAT held that Network Rail had also failed to satisfy the third and fourth conditions. It dealt with those conditions in the following way:

"277. If we had found that the RISQS-only rule gave rise to safety benefits, it would have been necessary to consider whether Condition 3 (indispensability) was satisfied in relation to those benefits. Paragraph 75 of the Article 101(3) Guidelines states that Condition 3 "requires that the efficiencies be specific to the agreement in question in the sense that there are no other economically practicable and less restrictive means of achieving the efficiencies".

278. In the OTOC case considered above the Court of Justice's finding that OTOC's requirement for institutional training was not objectively necessary (because its purpose could be achieved by less restrictive means) was determinative of the issue of indispensability. Similarly in the present case, it necessarily follows from our finding that the RISQS-only rule is not objectively justified on safety grounds because its safety purposes can be achieved by other less restrictive means, such that the RISQS-only rule is not indispensable.

279. If it had been necessary to consider Condition 4 (no elimination of competition), we would have found that the RISQS-only rule in the Sentinel Scheme and OTPO Scheme affords Network Rail the possibility of eliminating competition in respect of a substantial part of the market for supplier assurance services. Namely, that part of the market consisting of buyers and suppliers needing access to Network Rail's managed infrastructure pursuant to those Key Schemes. The effect of the tender for the RISQS services is considered later in this judgment in the context of Achilles' Chapter II claim."

35. The CAT's conclusion in relation to the claim under Chapter I of the 1998 Act, therefore, was that by including the RISQS-only rule in the Schemes, Network Rail is entering into agreements between undertakings which have as their effect the prevention, restriction or distortion of competition within the United Kingdom: see paragraph [281] of the judgment.
36. Lastly, the CAT decided that the insistence upon the RISQS-only rule is an abuse of a dominant position for the purposes of Chapter II of the 1998 Act, (assuming that Network Rail is dominant in the relevant market): see paragraph [314] of the judgment. Before turning to the way in which the CAT dealt with that issue, it is useful to note that the definition of "abuse" for these purposes is not in dispute. It is contained in Case 85/76 Hoffmann-La Roche & Co AG v Commission EU:C:1979:36 at [91] and is as follows:

"The concept of abuse is 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."

See paragraph [288] of the judgment.

37. In this regard, the CAT relied upon its conclusion for the purposes of its analysis under Chapter I that the RISQS-only rule causes significant foreclosure of demand in a significant segment of the market for supplier assurance services in the railway sector in Great Britain, and that the rule has an appreciable (in the sense of more than de minimis) effect on competition in the market for supplier assurance services in that sector. It concluded that: (a) there were anti-competitive effects on competition in the market for supplier assurance services in the railway sector in Great Britain (paragraphs [290] – [292] ); (b) the tender exercise that was carried out and the possibility of a further tender exercise for the market in the future did not affect, justify or compensate for the elimination of competition in the meantime (paragraph [297] ); (c) the RISQS-only rule was not consistent with the normal conditions of competition in a market of this kind (paragraphs [298] – [301]); and (d) Network Rail's argument that there must be a commercial benefit to be gained by the dominant undertaking from its conduct before that conduct can amount to abuse of a dominant position is inconsistent with the law explained by the General Court in Case T-128/98 Aéroports de Paris v Commission EU:T:2000:290 and the High Court in Arriva the Shires Limited v London Luton Airport Operations Limited [2014] EWHC 64 (Ch) (paragraphs [302] – [308], and [306] in particular.)
38. Lastly, the CAT rejected the argument that even if the RISQS-only rule was considered to be a potential abuse of dominance, it could be objectively justified on safety, efficiency and costs grounds for the same reasons as applied in relation to the Chapter I justification arguments: see paragraph [313].
39. Accordingly, the CAT went on to hold that, on the assumption that Network Rail has a dominant position in the market for the operation and provision of access to national rail infrastructure in Great Britain, its conduct in mandating the RISQS-only rule is an abuse of its dominant position: see paragraph [314].

Grounds of Appeal

40. There are six grounds of appeal. First, in relation to the Chapter I claim, it is said that the CAT erred in law in finding that setting out rules with a view to ensuring safety and to comply with Network Rail's regulatory operations was not dissociable from Network Rail's operation of the rail infrastructure, which is admittedly an economic activity, and that therefore the Schemes themselves constitute economic activities subject to UK competition rules. Secondly, it is said that the CAT misapplied the law in finding that each of the Schemes embodies an agreement or concerted practice within the meaning of section 2 of the 1998 Act.
41. It is convenient to take the third and fourth grounds together as they both relate to whether the CAT erred in finding that there was an appreciable effect on competition for the purposes of section 2 of the 1998 Act. The third ground is that the CAT relied upon evidence which manifestly did not support the conclusion which it reached as to the scope of the relevant market, which was a necessary measure by which to determine whether there had been an appreciable effect on competition. The fourth ground, which follows from the third, is that the CAT erred in law in finding that the RISQS-only rule in each of the Schemes has an appreciable effect on competition in the relevant market. In particular, it is said that the CAT: (a) failed to consider evidence of the impact that removing the RISQS-only rule would have had on the tender carried out by the RSSB to appoint new service providers; (b) misapplied the Socrates and OTOC cases and did not carry out any market share analysis or properly assess the degree of foreclosure arising from the RISQS-only rule; (c) contrary to section 60 of the 1998 Act, failed to apply the European Commission's Notice on agreements of minor importance which do not appreciably restrict competition under Article 101(1) TFEU (the "De Minimis Notice") 2014/C 291/01; and (d) was wrong in law in concluding that it is "fundamentally not for Network Rail to make the decision for other buyers and suppliers that they would prefer RISQS to other supplier assurance services." See paragraph [152] of the judgment.
42. The fifth ground relates to the exemption from the Chapter I prohibition afforded by section 9 of the 1998 Act. In summary, it is said that the CAT failed to apply Commission Regulation (EU) No 330/2010, erred in law in finding that the relevant benefits must be causally linked to the relevant restriction, erred in weighing the costs associated with removing the RISQS-only rule against the benefits of competition under the first criterion for the exemption rather than the third criterion, and erred in applying the fourth criterion.
43. Lastly, the sixth ground of appeal is that the CAT erred in law in finding that there was an abuse of a dominant position, in particular, because: (a) the CAT's anti-competitive effects analysis was flawed; (b) the evidence was contrary to its conclusion in relation to the effects of the periodical tender process; (c) it erred in law in finding that a dominant company need not benefit commercially from the conduct complained of for it to be found to be abusive; and (d) it erred in law in finding that the conduct was an abuse without finding that Network Rail's taking supplier assurance from Achilles was indispensable or essential to enable Achilles to be active in supplier assurance in the GB rail sector or safety critical industries in the United Kingdom. We consider each of the grounds of appeal in turn.

(1) Do the terms of the Schemes constitute economic activities subject to UK Competition Rules?

44. In essence, this ground of appeal is concerned with whether Network Rail is acting as an 'undertaking" when insisting upon the RISQS-only rule in the Schemes. The issue arises because section 2 of the 1998 Act is concerned, amongst other things, with "agreements between undertakings". The concept of an "undertaking" for the purposes of Community Competition law (which is relevant for the purposes of section 2 as a result of section 60 of the 1998 Act) is described succinctly in Case C205/03P Federación Española de Empresas de Tecnología Sanitaria v Commission ("FENIN"), as covering:

"25. . . . any entity engaged in an economic activity, regardless of the legal status of that entity and the way in which it is financed (Höfner and Elser (C-41/90): [1991] ECR I-1979; [1993] 4 CMLR 306 at [21]; and AOK-Bundesverband (C 264, 306, 354 & 355/01): [2004] ECR I-2493; [2004] 4 CMLR 22 at [46])."

It is also accepted that for these purposes, "it is the activity consisting in offering goods and services on a given market that is the characteristic feature of an economic activity": see Commission v Italy (C-35/96): [1998] ECR I-3851; [1998] 5 C.M.L.R. 889 at [36], also quoted and affirmed in FENIN at [25].

45. In summary, Mr Went, on behalf of Network Rail, submits that the RISQS-only rule is concerned with the safety of the rail infrastructure for which Network Rail is responsible and that when insisting upon the application of the rule, Network Rail is not conducting any economic activity or offering any goods or services on a market. He says that it is fulfilling a regulatory function. Accordingly, he says that it was wrong in law for the CAT to hold, as it did at paragraph [102] of the judgment, that allowing access to the railway infrastructure is not dissociable from Network Rail's operation of the railway infrastructure, which is an economic activity. It follows, therefore, that Network Rail was not acting in its capacity as an undertaking when imposing the RISQS-only rule and, accordingly, cannot fall within the prohibition in section 2.
46. Mr Went also criticised the CAT's reliance upon the case of FENIN in relation to its conclusion. The FENIN case was concerned with the purchase of goods and services by the public bodies which run the Spanish national health service which provides its services free of charge. The relevant issue in that case was whether in conducting the purchasing activity the public bodies were acting as "undertakings" for the purposes of Article 82 (now Article 102) of the TFEU, even though they did not act as undertakings when providing the health services in which the goods and services were used. In this regard, the Court of Justice (Grand Chamber) held as follows:

"26. The Court of First Instance rightly deduced, in para.[36] of the judgment under appeal, that there is no need to dissociate the activity of purchasing goods from the subsequent use to which they are put in order to determine the nature of that purchasing activity, and that the nature of the purchasing activity must be determined according to whether or not the subsequent use of the purchased goods amounts to an economic activity."

In that case, therefore, the purchasing by the public bodies was not dissociated from the subsequent use of the goods and services which was not an economic activity and, accordingly, the public bodies were not "undertakings" for the purposes of Article 102 in relation to the purchasing. The appeal was dismissed.

47. Mr Went submits that it does not follow from the FENIN case that where a particular activity carried out by an undertaking is, on the face of it, non-economic in nature, the activity must be deemed economic merely because the undertaking is otherwise engaged in economic activity. He says that the RISQS-only rule is not an input to any economic activity carried out by Network Rail and there is no connection between the two.
48. He also says that the CAT ignored the functional approach to undertakings pursuant to which a body may be an undertaking for one purpose and not for another: see MOTOE v Eliniko Dimosio [2008] ECR I-4863. In that case, a body vested with public powers to grant applications to organise motorcycling events was found not to be acting as an undertaking when making such authorisation decisions but was considered to act as an undertaking when carrying out economic advertising and sponsorship activities relating to the events. In that case, the Grand Chamber of the Court of Justice made clear that:

"25. As regards the possible effect of the exercise of public powers on the classification of a legal person such as ELPA as an undertaking for the purposes of Community competition law, it must be noted, as the Advocate General did at point AG49 of her Opinion, that the fact that, for the exercise of part of its activities, an entity is vested with public powers does not, in itself, prevent it from being classified as an undertaking for the purposes of Community competition law in respect of the remainder of its economic activities (Aéroports de Paris v Commission of the European Communities (C-82/01 P) [2002] E.C.R. I-9297; [2003] 4 C.M.L.R. 12 at [74]. The classification as an activity falling within the exercise of public powers or as an economic activity must be carried out separately for each activity exercised by a given entity.

26. In the present case, it is necessary to distinguish the participation of a legal person such as ELPA in the decision-making process of the public authorities from the economic activities engaged in by that same legal person, such as the organisation and commercial exploitation of motorcycling events. It follows that the power of such a legal person to give its consent to applications for authorisation to organise those events does not prevent its being considered an undertaking for the purposes of Community competition law so far as concerns its economic activities referred to above.

27. As regards the effect that the fact that ELPA does not seek to make a profit may have on that classification, it should be noted that, in Ministero dell'Economia e delle Finanze v Cassa di Risparmio di Firenze SpA (C-222/04) [2006] ECR I-289; [2008] 1 CMLR 28 at [122] & [123]), the Court stated that the fact that the offer of goods or services is made without profit motive does not prevent the entity which carries out those operations on the market from being considered an undertaking, since that offer exists in competition with that of other operators which do not seek to make a profit."

49. Mr Went submits that Network Rail's activity when insisting upon the RISQS-only rule is that of setting safety rules which is essentially a regulatory function which can be dissociated from its economic activities, all the more so because it makes no profit out of doing so, it has no competitors and is not offering goods and services on the market for supplier assurance in the rail network in the United Kingdom. Accordingly, he says that this case can be distinguished from the circumstances under consideration in Wouters [2002] ECR I-1577 and in the OTOC case. In both those cases the body imposing the regulations (the Dutch Bar Association in the Wouters case and the Portuguese Order of Chartered Accountants in OTOC) was also active in the market in question.
50. In the Wouters case the question for the Court of Justice was whether a regulation concerning partnerships between members of the Bar and other professionals, adopted by a body such as the Dutch Bar (the 1993 Regulation), was to be regarded as a decision taken by an association of undertakings within the meaning of Article 85(1) (now Article 101) of the TFEU. The national court sought, in particular, to ascertain whether the fact that power was conferred by statute on the Dutch Bar to adopt rules universally binding both on registered members of the Dutch Bar and lawyers who are authorised to practise in other Member States and come to the Netherlands in order to provide services there, had any bearing on the application of competition law.
51. Having held that registered members of the Dutch Bar carry on economic activities and, therefore, are undertakings for the purposes of the Treaty/TFEU (paragraph [49]), the Court of Justice went on to consider "whether when it adopts a regulation such as the 1993 Regulation, a professional body is to be treated as an association of undertakings or, on the contrary, as a public authority" (paragraph [56]). It concluded at paragraph [58] that when adopting the 1993 Regulation the Dutch Bar was neither fulfilling a social function nor exercising powers which are typically those of a public authority, but was acting as the regulatory body of a profession, the practice of which constitutes an economic activity, and must be regarded as an association of undertakings when doing so (paragraphs [58] and [64]). The Court went on at paragraph [64] as follows:

". . . Such a regulation constitutes the expression of the intention of the delegates of the members of the profession that they should act in a particular manner in carrying on their economic activity."

52. In the OTOC case, the national court asked, amongst other things, whether an institution such as OTOC must be regarded as an association of undertakings, and whether Article 101(2) must be interpreted as rendering subject to those rules an entity which, like OTOC, is required by law to lay down binding rules of general application in compliance with legal requirements concerning compulsory training of chartered accountants with a view to providing citizens with a quality service. It also asked whether, having regard to the Wouters case and similar judgments, concerned with rules having an impact on the economic activity of the professional members of a professional association, Articles 101 and 102 preclude rules on the training of chartered accountants which have no direct influence on their economic activity.
53. The Court addressed the matter in the following way:

"40. In accordance with the case law of the Court, the FEU Treaty rules on competition do not apply to an activity which, by its nature, its aim and the rules to which it is subject, does not belong to the sphere of economic activity, or which is connected with the exercise of the powers of a public authority (see, inter alia, Wouters at [57], and the case law cited).

41. First, rules such as those at issue in the main proceedings cannot be regarded as not belonging to the sphere of economic activity.

42. It is common ground in that regard, on the one hand, that the OTOC itself provides training for chartered accountants and, on the other, that the access of other providers wishing to offer such training is subject to the standards set out in the contested regulation. Consequently, such a regulation has a direct impact on economic activity on the market of compulsory training for chartered accountants.

43. In addition, the obligation on chartered accountants to undertake training in accordance with the rules laid down by that regulation is closely linked to the practice of their profession, as the Polish government and the European Commission point out. Failure to comply with that obligation can therefore lead to disciplinary sanctions under arts 57(1)(a), 59(2), 63 and 64 of the Statute of the OTOC, such as suspension for a maximum period of three years or expulsion from that professional association.

44. Even if that regulation did not directly affect the economic activity of the chartered accountants themselves, as the referring court appears to suggest in its third question, that fact cannot, of itself, remove a decision of an association of undertakings from the scope of art.101 TFEU.

45. Such a decision can be such as to prevent, restrict or distort competition within the meaning of art.101(1) TFEU, not only on the market on which the members of a professional association practice their profession, but also on another market on which that professional association itself has an economic activity."

Conclusion on Ground 1:

54. It seems to us that in the light of all the authorities, the conclusion reached by the CAT at paragraph [102] of the judgment is an unimpeachable finding of fact. It concluded that allowing access to the railway infrastructure is an essential part of Network Rail's operation of the railway infrastructure which is an economic activity and is not dissociable from it. The CAT was entitled to consider Network Rail's activities as a whole and to determine that the control of access to railway infrastructure was essential to its economic activity. It seems to us that there is nothing illogical or erroneous in concluding that there is an inherent link between Network Rail's economic activity of operating the railway infrastructure and seeking to ensure the safety of that infrastructure. It is not contended that the CAT's conclusion was contrary to the evidence before it.
55. The CAT's conclusion is also consistent with the approach taken by the European Court of First Instance in Case T-229/94 Deutsche Bahn v Commission [1997] ECR II-1689 at [55], to which Mr Woolfe, on behalf of Achilles, referred in his written submissions. It held that the provision of a range of services, in particular the provision of locomotives, traction and access to railway infrastructure, amounted to a rail services market with a specific character and a demand and supply of a very specific nature.
56. Furthermore, we cannot see any error arising from the reference made to the FENIN case at paragraphs [100] and [101] of the CAT's judgment. The exercise undertaken was analogous to that which was conducted in FENIN in the sense that the CAT considered whether access to the railway infrastructure was dissociable from the operation of that infrastructure, on the basis of the evidence before it. In the same way as the Court of Justice in FENIN decided that the purchase of goods and services for use in the Spanish health service could not be dissociated from the use of those goods and services in the public health service, which was not an economic activity, in this case the CAT decided, on the facts, that control of the access to the infrastructure and the operation of that infrastructure were not dissociable. There is nothing wrong in that.
57. Mr Went's argument in this regard begins from the premise that when insisting upon the RISQS-only rule in relation to access to the railway infrastructure, Network Rail is fulfilling a separate function from its economic activities in relation to which it makes no profit. He submits that it is merely seeking to impose safety rules and regulate health and safety and, in that respect, is not engaged in any economic activity.
58. We agree with Mr Woolfe that this is to start from a false premise. Network Rail is not, itself, a regulator in relation to the railway industry and it has no public function in that regard. It is not carrying out a public function when insisting on the RISQS-only rule. It is, itself, subject to regulations in the way in which it carries out its economic activities in running the railway infrastructure and, no doubt, seeks to impose the RISQS-only rule in an endeavour to comply with the regulations imposed on it in the conduct of its economic activities. It is different, therefore, from the circumstances in the MOTOE case in which the body in question was vested with public powers to grant applications and its activities when doing so were considered to be separate from its economic and sponsorship activities in relation to which it was considered to act as an undertaking. The MOTOE case also makes clear that it is irrelevant that no profit is made from the imposition of the RISQS-only rule: see paragraph [27].
59. It also seems to us that the fact that the bodies imposing the regulations in the Wouters and OTOC cases were also active in the market in question does not assist Network Rail, given the nature of the decisions in those cases to which we have referred.
60. To conclude, therefore, it seems to us that the CAT was entitled to decide, on the evidence before it, that because, in operating the Schemes, Network Rail is setting out rules with a view to ensuring safety and to comply with regulatory obligations, that fact does not take it outside the scope of its economic activity; and that access to the infrastructure is an essential part of Network Rail's economic activity of operating the railway infrastructure.

(2) Do the Schemes amount to an "agreement" falling within section 2 of the 1998 Act?

61. As we have already mentioned, section 2 of the 1998 Act is concerned, amongst other things, with "agreements between undertakings". Although the point was not taken before the CAT, Mr Went submits that the Schemes which include the RISQS-only rule involve unilateral rules imposed upon third party contractors and there is no "concurrence of wills" between them and Network Rail which might justify a conclusion that there is an agreement for the purposes of section 2. He says that suppliers did not sign up to the Sentinel Scheme, that the Schemes do not form part of agreements between suppliers and Network Rail, that the suppliers cannot choose whether or not to apply the Schemes and, therefore, there was neither an agreement with Network Rail nor acquiescence by the suppliers in relation to the RISQS-only rule. He says, therefore, that the CAT was wrong to hold as it did at paragraph [99] of the judgment and to rely upon the reasoning in the car dealership cases in that regard.
62. Mr Went submits that the car dealership cases (Case C-74/04 P Volkswagen v Commission EU: C:2006:460 and BMW v Commission EU:C:1979:191) involve a conceptually different situation and, therefore, the CAT misapplied the relevant law. In those cases, an apparently unilateral measure was imposed by car manufacturers on their distributors with whom they already had a distributorship agreement. Mr Went says that the question in those cases was whether the new requirement could form part of the ongoing contractual arrangements. He says that the situation here is more akin to that in Joined Cases C-2/01 P and C-3/01 P Bundesverband der Arzneimittel-Importeure eV and Commission v Bayer AG EU:C:2004:2 [2004] ECR I-23, because in those cases it was necessary to examine whether there was an anti-competitive agreement at all as opposed to whether a unilateral measure could form part of an ongoing contractual arrangement. Here, he says, there is no agreement at all and the RISQS-only rule in the Schemes is a genuinely unilateral measure. Network Rail merely lays down rules for suppliers of works and services on its rail infrastructure.
63. The issue in the BAI v Bayer AG cases was different from and more complex than the way in which Mr Went described it. The issue was whether it could be inferred that the wholesalers of drugs had tacitly acquiesced in an agreement not to make parallel exports, or whether Bayer was simply acting unilaterally to reduce the volumes of drugs that wholesalers had available to re-export. It is helpful to set out the paragraphs in which the Court of First Instance considered the nature of an agreement for the purposes of Article 85 (Article 101). They are as follows (see Case T-41/96 Bayer v Commission [2001] 4 C.M.L.R. 4; EU:T:2000:242):

"66. The case law shows that, where a decision on the part of a manufacturer constitutes unilateral conduct of the undertaking, that decision escapes the prohibition in Article 85(1) of the Treaty (Case 107/82. AEG v. E.C. COMMISSION; Joined Cases 25 & 26/84, FORD AND FORD EUROPE v. E.C. COMMISSION; Case T-43/92, DUNLOP SLAZENGER v. E.C. COMMISSION).

67. It is also clear from the case law in that in order for there to be an agreement within the meaning of Article 85(1) of the Treaty it is sufficient that the undertakings in question should have expressed their joint intention to conduct themselves on the market in a specific way (Case 41/69.,ACF CHEMIEFARMA v. E.C. COMMISSION; Joined Cases 209-215 & 218/78, VAN LANDEWYCK AND OTHERS v E.C. COMMISSION; Case T-7/89, HERCULES CHEMICALS v. E.C. COMMISSION).

68. As regards the form in which that common intention is expressed, it is sufficient for a stipulation to be the expression of the parties' intention to behave on the market in accordance with its terms (see, in particular, ACF CHEMIEFARMA, and VAN LANDEWYCK), without its having to constitute a valid and binding contract under national law (SANDOZ).

69. It follows that the concept of an agreement within the meaning of Article 85 (1) of the Treaty, as interpreted by the case law, centres around the existence of a concurrence of wills between at least two parties, the form in which it is manifested being unimportant so long as it constitutes the faithful expression of the parties' intention.

70. In certain circumstances, measures adopted or imposed in an apparently unilateral manner by a manufacturer in the context of his continuing relations with his distributors have been regarded as constituting an agreement within the meaning of Article 85(1) of the Treaty (Joined Cases 32, 36-82/78, BMW BELGIUM AND OTHERS v. E.C. COMMISSION; FORD AND FORD EUROPE; Case 75/84, METRO v. E.C. COMMISSION ("METRO II"); SANDOZ; Case C-70/93, BMW v. ALD).

71. That case law shows that a distinction should be drawn between cases in which an undertaking has adopted a genuinely unilateral measure, and thus without the express or implied participation of another undertaking, and those in which the unilateral character of the measure is merely apparent. Whilst the former do not fall within Article 85(1) of the Treaty, the latter must be regarded as revealing an agreement between undertakings and may therefore fall within the scope of that article. This is the case, in particular, with practices and measures in restraint of competition which, though apparently adopted unilaterally by the manufacturer in the context of its contractual relations with its dealers, nevertheless receive at least the tacit acquiescence of those dealers.

72. It is also clear from that case law that the Commission cannot hold that apparently unilateral conduct on the part of a manufacturer, adopted in the context of the contractual relations which he maintains with his dealers, in reality forms the basis of an agreement between undertakings within the meaning of Article 85(1) of the Treaty if it does not establish the existence of an acquiescence by the other partners, express or implied, in the attitude adopted by the manufacturer (BMW BELGIUM; AEG; FORD AND FORD EUROPE; METRO II; SANDOZ; BMW v. ALD)."

Conclusion on Ground 2:

64. It seems to us that there is no error of law in the CAT's decision that the imposition of the Schemes amounts to an "agreement" for the purposes of section 2 of the 1998 Act. First, in our judgment, the insistence on the RISQS-only rule does not amount to the imposition of a unilateral requirement by Network Rail. Mr Woolfe gave an example of what he said would be a unilateral requirement: that no red-headed person be allowed to have access to the rail infrastructure. Perhaps a more realistic example might be that Network Rail might refuse to allow access to any contractors who fail to meet its environmental criteria. It seems to us that that would be a truly unilateral requirement. On the contrary, here, as the CAT observed, at paragraph [99], it is clear from the Schemes that they impose responsibilities and obligations upon the bodies which sign up to them and that those responsibilities are ongoing. For example, they are required to report breaches of the requirements of the Schemes and to undertake investigations.
65. Secondly, it is also clear from paragraph [72] of the Court of First Instance Bayer decision that acquiescence is sufficient for the purposes of Article 85 (Article 101) and therefore it is sufficient for an agreement under section 2. On the evidence before it, the CAT was entitled to find such acquiescence. As Mr Woolfe pointed out, the fact that the suppliers had no choice but to participate in the Schemes if they wanted access to the rail infrastructure is irrelevant. Acquiescence under pressure is acquiescence nevertheless: see, for example, the BMW case at [36].
66. Thirdly, even if the Schemes do not form part of any contract or agreement for the supply of goods or services, that does not mean that they are not agreements within the scope of section 2: see the Wouters case at [56] – [64] and AC-Treuhand EU:C:2015:717 at paragraphs [27] and [33] – [35].
67. Accordingly, in our judgment, the imposition of the Schemes containing the RISQS-only rule amounts to an "agreement" for the purposes of section 2 of the 1998 Act and the CAT was correct so to find.

(3) Was there evidence to support the CAT's conclusion in relation to the scope of the relevant market?

68. As we have already mentioned, Network Rail's appeal in relation to the CAT's definition of the scope of the relevant market is closely linked to and relevant for the purposes of its challenge in relation to whether the RISQS-only rule had an appreciable effect on competition in that market. The paragraphs of the judgment upon which Network Rail focuses are as follows:

"127. There were differences in the way that the two economic experts defined the scope of the relevant market for supplier assurance services, applying the SSNIP or "small but significant non-transitory increase in price" test. This test seeks to identify the smallest relevant market within which a hypothetical monopolist could impose a profitable significant increase in price. Mr Parker, Achilles' economic expert, considered that the relevant market was the market for Key Scheme compliant supplier assurance schemes on the footing that it would not be possible for a provider operating a different supplier assurance scheme to start providing Key Scheme compliant supplier assurance services because only Network Rail can give the necessary recognition. Mr Holt, Network Rail's economic expert, disagreed and considered that the relevant market should be more broadly defined as the market for supplier assurance schemes in the GB rail industry on the basis that there would be significant competitive restraints on a hypothetical monopolist from collective intervention from buyers and suppliers. If a hypothetical monopolist sought to increase prices or decrease qualities, buyers and suppliers could intervene collectively as happened in 2016 when the previous RISQS scheme was put out to tender by the RSSB.

128. We prefer the definition of the market advanced by Mr Holt, for the reasons he gave. This definition of the market corresponds to the way the relevant market is defined in Achilles' claim form, i.e. the market for the provision of supplier assurance services in the GB rail industry."

69. Mr Went submits that the CAT's conclusion as to the relevant market is not based on the evidence cited or furnished to it. He also says that the CAT was wrong in stating that Mr Holt, who was Network Rail's expert, considered that the relevant market should be defined as the market for supplier assurance schemes in the GB rail industry on the basis that there would be significant competitive restraints on a hypothetical monopolist from collective intervention from buyers and suppliers. He says that Mr Holt's point was directed at showing that regardless of the precise market definitions, there are several constraints on the RISQS-only rule in terms of its ability to exercise market power.
70. He also says that at paragraph 28.4 of its Claim Form Achilles contended, in the alternative, that the relevant market is supplier assurance schemes in safety critical industries, recognising that the market may well be broader than merely supplier assurance in the GB rail industry. Accordingly, he says that the Claim Form cannot provide evidence that the market was as the CAT found it to be.
71. Although submissions were made in relation to the matters which should be taken into consideration when determining the relevant market and the different aspects of this case upon which that definition might bear, it is not necessary to rehearse them here. The real question for these purposes is whether the CAT's conclusion in this regard was open to it upon the evidence before it. When determining that question it is also important to bear in mind that the CAT is a specialist tribunal which is entitled to engage with the expert evidence itself.
72. In any event, we were taken to Mr Holt's treatment of this issue in his expert's report and the way in which the matter is dealt with in the joint experts' report. Although the extracts from the evidence upon which Network Rail relies are lengthy and detailed, it is necessary to consider them in detail in order to understand Mr Went's submissions in relation to this ground of appeal.
73. In his expert's report, Mr Holt stated that Achilles had correctly described the market for supplier assurance as "two-sided" (see paragraph 106 of the report). In a section headed "Market definition based on a SSNIP test" Mr Holt went on to consider whether it is appropriate to define an economic market for the provision of supplier assurance services in the rail industry in Great Britain or whether a wider or narrower market is more appropriate. Mr Holt then set out the order in which he intended to assess the appropriate market definitions at paragraph 122 of his expert's report. The definitions which he used were as follows:

"(a) Market C (very narrow, vn): provision of supplier assurance services for commodity codes (services) that are currently covered by RISQS. This market definition also excludes other supplier assurance schemes (e.g. RISAS) which specialise in other service or product markets related to the rail industry. I note that this definition is narrower than the definition proposed in Achilles' Claim Form.

(b) Market C (narrow, n): provision of supplier assurance services for the rail industry in the (sic) Great Britain. This market definition includes supplier assurance schemes other than RISQS, which specialise in other service markets related to the rail industry (e.g. RISAS).

(c) Market C (wide, w): the provision of supplier assurance services in safety critical industries in Great Britain. This market definition includes supplier assurance schemes that operate in non-rail safety critical industries.

(d) Market C (very wide, vw): the provision of supplier assurance services across all industries – safety critical or otherwise."

Mr Holt explained how to make the appropriate assessment at paragraph 124 of his expert's report, as follows:

"Consequently, to assess whether there is a market for the provision of supplier assurance services to the rail industry in Great Britain, one needs to assess whether a hypothetical monopolist provider of supplier assurance services could profitably increase total fees charged to buyers and suppliers of services related to the rail industry in Great Britain (while re-optimising the price structure as required) or otherwise worsen quality. The answer to this question depends on the extent of demand-side and supply-side substitutability for the supplier assurance services concerned."

74. He considered the "demand-side" substitution at paragraphs 127 and 128, concluding at paragraph 127 that "a narrow market definition is warranted from a demand substitutability point of view". This was consistent with the approach adopted by Achilles' economic expert, Mr Parker. Under the heading "Supply-side substitution", Mr Holt stated at paragraph 129 that "there may be significant competitive constraints on a hypothetical monopolist through supply-side substitution. Accordingly, the narrow market definition for supplier assurance may not be appropriate". He went on to consider that possibility at paragraphs 129 – 135 of the report and concluded this passage at paragraph 136 by stating that: ". . . as a starting point I begin with a narrowly defined market based solely on demand-side considerations (i.e. market C(vn)) and take account of relevant constraints on the operation of RISQS as an industry-led scheme in my assessment of competitive effects. . .".
75. Mr Holt returned to these issues at paragraphs 138 – 142 under the heading "Is there a market for the provision of supplier assurance services for the rail industry or safety critical industries or a wider market in … Great Britain?" Although this is a lengthy passage in the report, it is necessary to set it out in full so that the thrust of Mr Holt's evidence can be readily understood. It is as follows:

"138. The narrowest market put forward in Achilles' Claim Form is the market for the provision of supplier assurance services in the rail industry in Great Britain (Market C(n)) and the widest is for the provision of supplier assurance services in safety critical industries in Great Britain (market C(w)), but the market might be wider still (such as the provision of all closed related services to supplier assurance services, whether safety critical or not, C(vw)).

139. If one is purely considering demand-side factors, then neither of these alternative markets would be sustainable and the narrower definition associated with specific commodity codes would apply.

140. In this regard, as noted in paragraph 120 above, Achilles, Altius and Capita compete for the provision of supplier assurance in other industries. The IT and audit solutions required for these sectors may be easily adopted to meet the requirements of buyers and suppliers in the rail industry. This observation is in line with the fact that Achilles' own proposed TransQ Global ("TransQ") scheme aims to operate across the whole transport sector.

141. Consequently, a hypothetical monopolist in this market who decided to increase prices by a small but significant amount may be swiftly replaced by a new entrant and thus find it unprofitable. Accordingly, a wider definition of the market – to include credible prospective entrants – may be more appropriate.

142. However, I do not have sufficient information to identify the breadth of this wider supply-side market since this will depend on a wide range of facts, including the precise level of diversification costs and so on. However, this is not determinative as the relevant market power issue is whether an incumbent provider of supplier assurance services in the rail industry could increase prices above the competitive level (or worsen quality), and this does not require high supply-side substitutability from all providers of related services. As noted in the previous subsection, I proceed on the basis that the supplier assurance market is narrowly defined, but consider that there are important constraints on RISQS."

76. In the joint experts' report, however, Mr Holt adopted a slightly different position. At paragraph 17(a) he is recorded as having agreed "that there is limited demand-side substitution and that other schemes cannot provide these services in relation to the Key Schemes without recognition, but [he] considers that a broader market definition should be adopted to take account of supply-side constraints since NRIL [Network Rail] (and the industry more generally) could sponsor entry from a wide range of other providers in the event that RISQS would seek to exercise market power". Reference is made to statement 3.3.
77. At statement 3.3a Mr Holt disagreed with the statement that "there is a market for Key Scheme compliant supplier assurance services in the rail industry". He said: "This definition is narrower than the one proposed in Achilles' Claim Form (paragraph 28.3) and the market in which RISQS operates. This approach does not take account of supply-side substitution (the ability and incentive for the industry to sponsor entry)". Mr Holt went on to agree with statement 3.3b that there is a market for supplier assurance services in the rail industry. He referred back to paragraphs 129 and 133 of his expert's report and noted that buyers and suppliers could intervene collectively. He also stated that: "[s]ources of supply-side substitution could be other rail schemes, schemes operating in other sectors (if their diversification costs are low) and potentially a wider set of IT and process management services providers". He also agreed with statement 3.3c that there is a market for supplier assurance services in safety critical industries but made clear that his conclusions did not rest on the answer.
78. Mr Went submits, therefore, that Mr Holt's evidence was that the relevant market was broader than supplier assurance to the rail industry in Great Britain and that the CAT's approach at paragraphs [127] and [128] of its judgment was not based upon the evidence before it.

Conclusion on Ground 3:

79. It seems to us that this ground of appeal is hopeless. Having heard all of the evidence, the CAT, in its role as a specialist tribunal, formed the view that for the purposes of assessing the anti-competitive effect alleged in this case, it preferred a market definition that took account of the threat of collective intervention by buyers and suppliers to sponsor entry of a new scheme. It recorded Mr Holt's opinion accurately and, in effect, it chose his definition (b) Market C (narrow, n) which he had set out at paragraph 122 of his expert's report. It was entitled to come to the conclusion it did, having itself engaged with the entirety of the evidence before it.
80. At best, Mr Holt had floated the possibility of a wider definition as a result of supply-side substitution at paragraph 133 of his expert's report when he mentioned that sources of supply-side substitution could potentially include "providers of supplier assurance in other industries if their diversification costs [were] low . . .; and potentially a wider set of IT and process management service providers through their use of sub-contractors with specific expertise and training as regards these matters". He did not adopt that definition ultimately. Even if he had, the CAT would not have been obliged to accept it. As we have said, it seems to us that there is nothing in this point at all.

(4) Did the CAT err in its approach to appreciable effect of the Schemes on competition in the supplier assurance market in the UK?

81. Mr Went also submits that the CAT's approach to whether the Schemes have an appreciable effect on competition is flawed for a number of reasons. We will consider each in turn.
82. Although it was not pursued in oral argument, in written submissions it was argued that when considering the relevant counterfactual situation, the CAT failed to consider the evidence of the impact that removing the RISQS-only rule would have had on the tender process carried out by the RSSB.
83. The CAT addressed Network Rail's case in relation to tendering as a defence to the abuse of a dominant position claim at paragraphs [293] – [297] of the judgment and rejected it. It concluded on the facts that the competition in the tender process run by the RSSB to supply audit or IT services to the RSSB, which it would then use to provide supplier assurance, was not sufficient to take the place of continuing competition between the RSSB and Achilles to provide the supplier assurance services themselves: see [297]. Mr Went is, however, correct to point out that the CAT did not address the tender process when considering whether there was an appreciable effect on competition.
84. We agree with Mr Woolfe that the CAT was right not to do so, and that this does not provide a basis for undermining its conclusions on appreciable effect. The argument that by restricting competition between providers of supplier assurance through the RISQS-only rule (competition in the market) it is possible to promote competition to win the RSSB's tender process (competition for the market) is not relevant to whether the RISQS-only rule has an appreciable effect on competition. The process of weighing pro- and anti-competitive effects (in this case, the stifling of competition between providers of supplier assurance in the market versus competition for the market through the tendering process) only becomes relevant once the question of whether there are appreciable effects upon competition has been determined. This is made clear by the Commission's Horizontal Co-operation Guidelines (2011 C 11/01) at [20] which state as follows:

"20. The assessment under Article 101 consists of two steps. The first step, under Article 101(1), is to assess whether an agreement between undertakings, which is capable of affecting trade between Member States, has an anti-competitive object or actual or potential restrictive effects on competition. The second step, under Article 101(3), which only becomes relevant when an agreement is found to be restrictive of competition within the meaning of Article 101(1), is to determine the pro-competitive benefits produced by that agreement and to assess whether those pro-competitive effects outweigh the restrictive effects on competition. The balancing of restrictive and pro-competitive effects is conducted exclusively within the framework laid down by Article 101(3). If the pro-competitive effects do not outweigh a restriction of competition, Article 101(2) stipulates that the agreement shall be automatically void."

The same approach was adopted by the Court of First Instance in Case T-112/99 Metropole Television (M6) v Commission [2001] ECR II-2459 at [107].

85. Next, Mr Went took issue with the reliance placed by the CAT upon the decisions in the Socrates and OTOC cases for the proposition that an appreciable anti-competitive effect can consist in the segmenting of a market and a distortion or restriction of the way that competition operates in that segment, even if competition may continue elsewhere: see the judgment at [121(5)] and [154]. He says that those cases are not authority for that proposition and that Network Rail's position can be distinguished from that of the Law Society and OTOC in any event.
86. In this regard, Mr Went referred us to the factual background recorded in the OTOC case and, in particular, to the conclusion reached at [108], which is as follows:

"108. Having regard to all the foregoing considerations, the answer to the fourth question is that a regulation which puts into place a system of compulsory training for chartered accountants in order to guarantee the quality of the services offered by them, such as the contested regulation, adopted by a professional association such as the OTOC, constitutes a restriction on competition prohibited by art.101 TFEU to the extent, which it is for the referring court to ascertain, that it eliminates competition on a substantial part of the relevant market, to the benefit of that professional association, and that it imposes, on the other part of that market, discriminatory conditions to the detriment of competitors of that professional association."

He says, therefore, that the decision in the OTOC case concerned a situation in which the market was segmented and OTOC reserved one segment to itself and also imposed discriminatory conditions in the other. He says that this was also the situation in the Socrates case. In that case, the CAT stated as follows:

"160. . . the question of effect is not to be assessed simply on the basis of market share or complete foreclosure, but can result from a segmenting of the market and a distortion in the way competition operates affecting one segment."

Mr Went submits, therefore, that neither the OTOC nor the Socrates case was relevant to the circumstances in relation to the Schemes because there was no allegation of segmentation of the market for supply assurance or of entire elimination of competition in one segment and distortion of competition in another. Network Rail is not seeking to reserve part of the market to itself. Further, in both those cases, unlike Network Rail, the competition law infringer took active steps to exclude a competitor from the market. Network Rail was not and is not in competition with Achilles.

87. It seems to us that Mr Went is seeking to find a flaw in the CAT's reasoning by over-analysis of the authorities, and that he seeks to ignore the approach taken by the CAT overall, in the light of the factual situation before it. Paragraph [160] of the decision in Socrates, which was binding on the CAT, and is relied upon by it as the basis for the proposition at paragraph [121(5)] of the judgment, can be read as extending the principles in the OTOC case to circumstances in which the market is segmented and there is a distortion in competition affecting only one segment. This, however, is beside the point. The CAT did not find, on the basis of the Socrates case, that such a restriction was automatically appreciable. Rather, it accepted the submission made to it on behalf of Achilles that "if" the effect of the RISQS-only rule was to reserve to RISQS/RSSB a significant part of demand for supplier assurance, there "may" be an appreciable restriction of competition: see paragraph [123] of the judgment. It then went on to decide on the facts that there was a significant foreclosure of demand and that there is an appreciable anti-competitive effect. It did so on the basis of the detailed economic and factual evidence before it. That reasoning is at paragraphs [121] – [154] and [141] – [154] of the judgment, in particular.
88. Accordingly, we do not consider that the fact that the passage at paragraph [160] in the judgment in the Socrates case appears to extend the principles set out in OTOC is of assistance to Network Rail.
89. Mr Went criticises the CAT's reasoning and analysis in relation to whether there is an appreciable effect upon competition, as well. He says that adopting the proposition it did at paragraph [121(5)] of the judgment led the CAT into a mechanistic approach rather than to conduct the detailed analysis it ought to have done in order to determine whether there was an appreciable effect on competition. Mr Went says that the judgment contains no conclusions on market share or proper analysis of the actual extent to which demand for supplier assurance in the rail industry in the UK or any wider market is reserved to RISQS/RSSB through the RISQS-only rule. He submits that neither the OTOC nor the Socrates case obviates the need to consider market shares and the degree of foreclosure.
90. He also submits that despite being required to have regard to it pursuant to section 60 of the 1998 Act and having referred, at paragraph [121(2)] of the judgment, to a paragraph from the European Commission's Guidelines on Vertical Restraints which itself refers to the Commission's De Minimis Notice (Commission Notice 2014/C 291/01), the CAT nevertheless failed to apply that Notice when considering appreciable effects.
91. The De Minimis Notice, which is described in its title as a "Notice on agreements of minor importance which do not appreciably restrict competition under Article 101(1) of the Treaty on the Functioning of the European Union", states at paragraph 8(b) that the Commission holds the view that agreements between undertakings which may affect trade within the meaning of Article 101(1) TFEU do not appreciably restrict competition if the market share held by each of the parties to the agreement does not exceed 15% on any of the relevant markets affected by the agreement, where the agreement is made between undertakings which are not actual or potential competitors on any of those markets. Further, at paragraph 10 the Notice states that "a cumulative foreclosure effect is unlikely to exist if less than 30% of the relevant market is covered by parallel (networks of) agreements having similar effects". Mr Went points out that Network Rail is not in the relevant market at all, and the CAT had found that only 29% of supplier assurance requirements in the GB rail industry were met by the Sentinel scheme (paragraph [106(3)]). Nevertheless, the CAT did not follow the guidance in the De Minimis Notice.
92. In his written argument, Mr Went submitted that application of the De Minimis Notice would also have required an examination of the market shares held by each of the parties to each "agreement", looking at the Schemes separately and considering the cumulative effect of all the agreements. Instead of conducting such an analysis, however, the CAT conducted a high-level review on the basis of the counterfactual: see paragraphs [129] – [154] of the judgment. He submits, therefore, that the CAT failed to conduct the necessary exercise in order to be in a position to determine whether the Schemes and the RISQS-only rule, in particular, have an appreciable effect on competition in the relevant market.

Conclusion on Ground 4:

93. Addressing the last point first, the percentages in the De Minimis Notice are in our judgment merely guidelines and are not binding on the CJEU, national courts or competition authorities. As the Notice itself makes clear, it merely expresses the Commission's view: see Case C-226/11 Expedia Inc v Autorité de la Concurrence [2013] 4 CMLR 14; EU:C: 2012:795 at 27 -31 and section 60(3) of the 1998 Act. Accordingly, the reference to percentages in paragraph 10 of the De Minimis Notice and the CAT's finding that only 29% of supplier assurance requirements in the GB rail industry were met by the Sentinel scheme cannot be determinative.
94. Did the CAT, nevertheless, apply a mechanistic approach, based on the Socrates case, and fail to conduct the necessary analysis in order to determine whether the RISQS-only rule creates an appreciable effect on competition? In our judgment, it did not.
95. At paragraph [121] of the judgment, the CAT set out the legal principles by reference to which it intended to approach the question of whether the RISQS-only rule is a restriction of competition by effect. It described those legal principles as "largely common ground." Although they are summarised at paragraph [25] of this judgment, it is helpful to set out those principles in full:

(1) In cases where it is not plain and obvious that the object of the agreement in question is to restrict competition it is necessary to consider the effect of the agreement. See Case 56/65 Société Technique Minière v Maschinenbau Ulm GmbH EU:C:1966:38 ("LTM").

(2) There is no presumption of anti-competitive effect. It is necessary to show that the agreement has an appreciable effect on competition. Appreciable does not mean substantial; it means more than de minimis or insignificant. It must be demonstrated with a reasonable degree of probability that the agreement affects actual or potential competition to such an extent that negative effects on prices, output, innovation or the variety or quality of goods and services in the market can be expected. See Socrates Training Limited v The Law Society of England and Wales [2017] CAT 10 ("Socrates") at [154]; the Commission's Guidelines on Vertical Restraints at [8]; the Commission's Guidelines on the application of Article 81(3) of the Treaty ("the Article 101(3) Guidelines") at [24].

(3) In order to gauge the restrictive effects of an agreement, it is necessary to conduct a detailed analysis of its effect on the relevant market or markets. As stated in the Commission's Notice on the definition of relevant market for the purposes of Community competition law, the European Commission explained the purpose of market definition at [2] as follows: "Market definition is a tool to identify and define the boundaries of competition between firms. It serves to establish the framework within which competition policy is applied by the Commission. The main purpose of market definition is to identify in a systematic way the competitive constraints that the undertakings involve face. The objective of defining a market in both its product and geographic dimension is to identify those actual competitors of the undertakings involved that are capable of constraining those undertakings' behaviour and of preventing them from behaving independently of effective competitive pressure."

(4) An effect on competition must be demonstrated by reference to the situation which would pertain on the market in the absence of the agreement or restriction in question: "the competition in question must be understood within the actual context in which it would occur in the absence of the agreement in dispute" (see LTM). This requires a consideration of the appropriate counterfactual situation.

(5) An anti-competitive effect can consist in the segmenting of a market and a distortion or restriction of the way that competition operates in that segment of the market, even if competition may continue in other segments. See Socrates at [160]."

96. It is clear, therefore, that the CAT was fully aware of the exercise to be undertaken and that it was necessary to demonstrate, with a reasonable degree of probability, that the Schemes affect actual or potential competition to such an extent that negative effects can be expected: see [121(2)] above. It went on to accept in principle that "if the effect of the RISQS-only rule is to reserve to RISQS/RSSB a significant part of demand for supplier assurance, thereby impairing actual or potential competition from other suppliers of such services or schemes, there may be an appreciable restriction of competition": see [123]. Having accepted the principle, the CAT then considered market definition on the basis of the evidence before it (which did not include detailed market share data) and turned to the counterfactual at [129]. Both of those steps are consistent with the principles set out at [121(3)] and [121(4)].
97. After consideration of the evidence, the CAT stated its view at [150] that:

". . . the correct counterfactual is one in which Achilles would compete with RISQS at least for a time and … its competition would lead to some benefits in terms of lower prices and product differentiation, as contended by Achilles."

The reasoning which supports that view is then set out at [151] – [153], leading to the following conclusion at [154]:

"Although the scope for price competition (in the absence of loss-leading or cross-subsidisation on a long-term basis) and product differentiation would be limited, we conclude that the RISQS-only rule does cause significant foreclosure of demand in a significant segment of the market for supplier assurances schemes in the GB railway sector and that the RISQS-only rule has an appreciable effect on competition in that market."

98. In our judgment, the CAT was clearly entitled to approach the matter in this way and to come to the conclusion it did on the basis of the evidence. There is nothing to suggest that a detailed analysis of market share data is necessary in all cases. In fact, the Court of Justice adopted a broad-brush approach to market share in the OTOC case as did the CAT in the Socrates case. See, in particular, paragraphs [62], [78], [79] and [97] in the former and paragraphs [161], [163(b)] and [164] in the latter.
99. Finally, Mr Went submits that the CAT's statement at paragraph [152] of the judgment that it is "fundamentally not for Network Rail to make the decision for other buyers and suppliers that they would prefer RISQS to other supplier assurance services" appears to have been an important point underpinning its conclusion in relation to appreciable effects on competition, and is wrong in law.
100. Mr Went submits that a similar point is made at paragraph [240] when the CAT is considering objective justification. That paragraph is as follows:

"Sixthly, we note that any other buyer of supplier assurance services for works on Network Rail managed infrastructure or other railway services would also have safety-related obligations under applicable legislation which would extend to supplier assurance. Consequently, it does not follow that Network Rail is the only party concerned with safety or that Network Rail should be the sole arbiter of the relevant standards when other buyers (or at least buyers who are not in the Network Rail supply chain) are involved. Achilles pointed to dicta in Case T-30/89 Hilti AG v Commission EU:T:1991:70 at [118] to [119] that it is primarily the role of public authorities, not dominant undertakings, to set and enforce safety standards."

Mr Went says that the Hilti case can be distinguished from this case on the facts. In that case, he says, there were existing laws dealing with the aspect on which Hilti sought to impose restrictions in relation to the use of third parties' nails in Hilti's guns, on grounds of safety. However, in this case, the government requires Network Rail, as operator of the railway infrastructure, to comply with health and safety regulations and to make sure that risk is managed by the most effective means.

101. In the light of the fact that it does not appear that the CAT relied upon the decision in Hilti for its conclusion at paragraph [152] of the judgment or made any finding that it was not for Network Rail to seek to promote safety standards, it is not necessary for us to consider the Hilti case any further.
102. Further, and in any event, although the last sentence of paragraph [152] is expressed in trenchant terms, it did not in our judgment influence the exercise carried out by the CAT or its ultimate decision and, accordingly, is irrelevant.
103. For all of the reasons set out above, we conclude that the CAT's approach to appreciable effect of the Schemes on competition in the supplier assurance market in the UK was not erroneous in any of the ways that Network Rail suggests.

(5) Does the exemption under Chapter I apply?

104. By virtue of section 9 of the 1998 Act, an agreement is exempt from the Chapter I prohibition if it satisfies four cumulative conditions, two positive and two negative. The two positive conditions (Conditions 1 and 2) are:

(1) The agreement must contribute to improving production or distribution, or promoting technical or economic progress; and

(2) The agreement must allow consumers a fair share of the benefit resulting from Condition 1.

The two negative conditions (Conditions 3 and 4) are:

(3) The agreement must not impose on the relevant undertakings restrictions which are not indispensable to the attainment of the objects of Conditions 1 and 2; and

(4) The agreement must not afford the relevant undertakings the possibility of eliminating competition in respect of a substantial part of the products in question.

Although Condition 3 is framed in negative terms, it may be easier to think of it as essentially positive in nature: the restrictions imposed by the agreement must be indispensable to the attainment of Conditions 1 and 2. This is how the CAT stated Condition 3 at [256] of the judgment.

105. The burden of proof rests on the party claiming the benefit of the exemption, in this case Network Rail: see section 9(2). In its Defence, Network Rail pleaded (at paragraph 39) that "the criteria for exemption (namely those in section 9 of [the 1998 Act] are met" in relation to the requirement in the Schemes to use RISQS. In particular, Network Rail averred that:

(a) the requirement in the Schemes to use a single assurance scheme "gives rise to direct efficiencies (including through enhancing safety) and therefore contributes to improving production or distribution, and/or promoting technical or economic progress";

(b) there is "no other operationally and/or economically practical and less restrictive means of achieving the efficiencies", and the requirement is reasonably necessary to produce them;

(c) consumers (both suppliers active in the GB rail industry and railway users) receive a fair share of the resulting benefits; and

(d) the requirement "does not eliminate competition in respect of a substantial part of the products and/or services concerned."

106. As we have explained, the CAT held that Condition 1 was not satisfied, and although that conclusion alone would have sufficed to establish that Network Rail could not rely on the section 9 exemption, the CAT went on to hold, more briefly, that Network Rail also failed to satisfy Conditions 3 and 4. The CAT did not consider Condition 2, but Mr Woolfe made clear to us that, if Condition 1 were satisfied, Achilles accepted that Condition 2 would also be satisfied, or in other words that consumers would have a fair share of the resulting benefit.

Is it open to Network Rail to rely on the Vertical Block Exemption Regulation?

107. The first error of law alleged in Network Rail's fifth ground of appeal is that the CAT failed to apply Commission Regulation (EU) No 330/2010 of 20 April 2010 on the application of Article 101(3) TFEU to categories of vertical agreements and concerted practices ("the Vertical Block Exemption Regulation" or "VBER"). As Mr Went frankly conceded, this is a new argument which had not been pleaded or argued before the CAT. He submitted, however, that it raises a pure question of law, so Network Rail should be permitted to argue it on the appeal. This is opposed by Achilles, which says that reliance on the VBER at trial would have required certain facts to be determined which were not otherwise necessary, particularly in relation to market shares, and the leading of appropriate witness and expert evidence. In the absence of any pleaded defence based on the VBER, none of these steps were taken, and it is now far too late to raise the issue for the first time on appeal.
108. As a preliminary point, it should be noted that the pleaded claim to exemption under section 9 of the 1998 Act is quite separate from reliance on a block exemption under the VBER. Exemption on that basis would be a "parallel exemption" falling within the scope of section 10, not section 9.
109. Article 2 of the VBER states that, subject to the provisions of the Regulation, Article 101(1) of the Treaty (i.e. the EU law equivalent of the Chapter I prohibition) "shall not apply to vertical agreements." "Vertical agreement" is defined, in Article 1(1)(a), as meaning:

"An agreement or concerted practice entered into between two or more undertakings each of which operates, for the purposes of the agreement or the concerted practice, at a different level of the production or distribution chain, and relating to the conditions under which the parties may purchase, sell or resell certain goods or services."

110. Article 3(1) of the VBER further provides for a 30% market share threshold:

"The exemption provided for in Article 2 shall apply on condition that the market share held by the supplier does not exceed 30% of the relevant market on which it sells the contract goods or services and the market share held by the buyer does not exceed 30% of the relevant market on which it purchases the contract goods or services."

111. Detailed guidelines on the application of the VBER were published by the European Commission in 2010, under the heading "Guidelines on Vertical Restraints".
112. In our judgment, Mr Woolfe is clearly right that it is now too late for Network Rail to seek to rely on the block exemption contained in the VBER. The simplest basis for this conclusion is that, in order to rely on the exemption, it would have been necessary for Network Rail to plead, and prove by appropriate evidence, that the market share threshold conditions in Article 3(1) were satisfied. Since Network Rail is the provider of almost all mainline railway infrastructure in Great Britain, and for the purposes of the trial was assumed to be dominant on that market, it is very far from obvious that it would have had any reasonable prospect of establishing that its share of the relevant market for the purposes of the VBER was below 30%. Had it wished to do so, it would have needed to plead the contention with appropriate particularity, followed by disclosure and evidence which could have been tested at trial. It follows that the question cannot be treated as a pure point of law which may be raised on appeal for the first time with no unfairness to Achilles.
113. Apart from that fundamental procedural objection, we are also inclined to agree with Mr Woolfe that, in any event, it is hard to see how the RISQS-only rule could plausibly fit within the definition of "vertical agreement" in Article (1)(a) of the VBER. According to the definition, the agreement in question must be one which relates to "the conditions under which the parties may purchase, sell or resell certain goods or services"; but the rule instead relates to the terms on which undertakings are licensed to have access to Network Rail's infrastructure. We agree with Mr Woolfe that support for this interpretation of Article 1(1)(a) is found in the Commission's Guidelines on Vertical Restraints, at paragraphs 25 and 26. It is there stated that the purpose of the VBER is to cover "agreements which concern the conditions for the purchase, sale or resale of the goods or services supplied by the supplier and/or which concern the conditions for the sale by the buyer of the goods or services which incorporate these goods or services". The Regulation "does not cover restrictions or obligations that do not relate to the conditions of purchase, sale and resale such as an obligation preventing parties from carrying out independent research and development which the parties may have included in an otherwise vertical agreement." With the benefit of that guidance, it seems reasonably clear to us that the RISQS-only rule, concerned as it is with the terms of access to Network Rail's infrastructure, falls outside the scope of the VBER.
114. Mr Woolfe had a further argument, to the effect that the RISQS-only rule also falls within the definition of a "non-compete obligation" in Article 1(1)(d), and is therefore excluded from the exemption in Article 2 by Article 5(1)(a), which removes from the scope of the block exemption any term which imposes a direct or indirect non-compete obligation, the duration of which is indefinite or exceeds five years. Mr Woolfe may well be right about this, but the RISQS-only rule does not fit easily into the definition of a "non-compete obligation" and we therefore prefer to leave the point open.
115. The remaining alleged errors of law identified in Network Rail's fifth ground of appeal relate to Conditions 1, 3 and 4 in section 9 of the 1988 Act, to which we now turn.

The CAT's reasoning on Condition 1

116. The CAT's analysis of Condition 1 is to be found at [260] to [275] of the judgment. The analysis begins as follows:

"260. In order to satisfy the first condition, the relevant benefits must be causally linked to the relevant restriction, here the RISQS-only rule. It is not sufficient to identify benefits which result from supplier assurance or from the Key Schemes generally. Furthermore, the causal link between the relevant benefits must be established by facts and evidence supported by empirical analysis and data and not just economic theory. See the Article 101(3) Guidelines at [50] to [57].

261. With regard to safety, the Tribunal accepts the operation of the RISQS-only rule has ensured that supplier assurance has been carried out to a uniform standard and that it has permitted supplier assurance to develop by, for example, enabling the RISQS board to take account of the experience and feedback of a wide range of buyers and suppliers. The Tribunal does not accept, however, for the reasons set out in Issue 4 above, that the safety benefits attributed to the RISQS-only rule by Network Rail are causally linked to the restriction as opposed to being the consequence of an effective and efficient regime of supplier assurance.

262. With regard to economic efficiencies, the safety experts agreed that there would need to be significant oversight by Network Rail in a multiple scheme environment.

263. Network Rail's case was that, at the lower end of its estimates, being required to "use" one additional supplier assurance scheme would cost Network Rail several hundreds of thousands of pounds per year:

[details were then given, based on the evidence of Mr Kenneth Blackley, who has a senior procurement role with Network Rail]

264. Achilles challenged these estimates as excessive but recognised that there would be some need for additional costs for the monitoring of audit quality by alternative providers of supplier assurance services (consistent with the view of both parties' safety experts), for cooperation between schemes to achieve desired safety outcomes and consequential IT processes. Achilles accepted that the one-off costs of mapping any proprietary codes to the RICCL coding structure used by the RSSB and RISQS would be borne by the relevant scheme and not by Network Rail.

265. We agree with Achilles that Network Rail's estimates were excessive for the following reasons."

117. The CAT then reviewed the relevant evidence and submissions, before concluding as follows:

"274. We conclude that, if Network Rail is required to recognise alternative providers of supplier assurance services, some additional costs of oversight will be required and that these would not be limited to occasional meeting attendance as contended by Achilles. We accept Network Rail's argument that it would be appropriate for Network Rail to be involved in some monitoring rather than rely on a system of mutual recognition between supplier assurance audit providers as advocated by Achilles. Network Rail currently relies on internal RSSB oversight for monitoring of quality within the RISQS processes. Our assessment is that Network Rail will need to employ at least one full-time employee for the purposes of monitoring and coordinating safety processes. That suggests an upper limit for incremental HR costs of £65,000 to £85,000 (based on a pro rata adjustment to the estimates by Network Rail for a team of three employees), together with some additional technology costs which would not be significant.

275. In our view, these incremental costs are insufficient to outweigh the benefits of competition, either in terms of price or other benefits. Although the contestable market may only have a total estimated value of £1.8 million and the scope for price reductions is limited, the incremental costs are such that it would only take a slight reduction in margin, which would probably result from competition, to outweigh them. Moreover, any cost benefits are outweighed by the potential non-price benefits of competition such as the potential for offering supplier assurance crossing between different industry sectors."

118. No challenge is made to the CAT's findings of fact in relation to Condition 1, if the CAT did not misdirect itself in law. What, then, are the alleged errors of law which are said to vitiate the CAT's analysis?

Did the CAT err in law in its analysis of Condition 1?

119. Network Rail argues that the CAT erred in law in its analysis of Condition 1 in two respects: first, it was wrong to find that the relevant benefits must be causally linked to the relevant restriction as opposed to the relevant restrictive agreement; and secondly, it was wrong to weigh the costs associated with removing the RISQS-only rule against the benefits of competition in the context of Condition 1 instead of Condition 3. In our judgment, however, there is no real substance to either of these complaints. With regard to the first point, Mr Woolfe was able to satisfy us, by reference to the decision of the Court of Justice in the MasterCard case (Case C-382/12 P MasterCard v Commission [2014] 5 CMLR 23; EU:C:2014:2201) that, where it has been determined that a restrictive term is not objectively necessary to the impugned agreement as a whole, it is necessary to examine any benefits which flow from the specific restriction in question, and it is not possible for that purpose to rely upon benefits flowing from the wider agreement of which it forms part. In the MasterCard case, the question arose in relation to a specific part of the overall MasterCard scheme called the Multilateral Interchange Fee ("the MIF"). The argument advanced by the appellants, as recorded in the judgment of the Court of Justice at paragraph 220, was that the General Court had:

"wrongly ignored the significant advantages which the Mastercard system and the MIF bring for cardholders, and, moreover, the two-sided nature of the system and the optimisation of the system which the MIF helped to achieve."

120. This argument was rejected by the Court, for reasons which appear sufficiently from paragraphs 230 and 231:

"230. The Court must reject at the outset the argument that the General Court wrongly ignored the advantages to cardholders resulting in the MasterCard scheme. It will be recalled that any decision by an association of undertakings which proved to be contrary to the provisions of art. 81(1) EC may be exempted under art. 81(3) EC only if it satisfies the conditions in that provision, including the condition that it contribute to improving the production or distribution of goods or to promoting technical or economic progress… Furthermore, as is apparent from [89] and [90] of the present judgment, where it is not possible to dissociate a decision by an association of undertakings from the main operation or activity with which it is associated without jeopardising its existence and aims, it is appropriate to examine the compatibility of that decision with art. 81 EC in conjunction with compatibility of the main operation or activity to which it is ancillary.

231. By contrast, where it is established that such a decision is not objectively necessary to the implementation of a given operation or activity, only the objective advantages resulting specifically from that decision may be taken into account in the context of Article 81(3) EC…"

Thus, in the present case, having determined that the RISQS-only rule was not objectively necessary on safety grounds to the implementation of Network Rail's activities, the CAT was right to concentrate on whether any benefits flow from the RISQS-only rule viewed separately.

121. As to Network Rail's second point, this is at best a sterile argument that the CAT's cost benefit analysis of the RISQS-only rule should have been conducted in relation to Condition 3 rather that Condition 1. Since all four Conditions have to be satisfied by Network Rail, and since the cost benefit analysis is on any view relevant, it does not much matter under which heading the CAT considered it. In any event, however, Mr Woolfe was again able to satisfy us that the question may properly be considered in the context of Condition 1 as well as Condition 3. In a case brought by a German association of property insurers to quash a decision of the Commission which had found a recommendation for increases in fire risk premiums to breach EU competition law, and (materially for our purposes) had also refused an exemption under what is now Article 101(3), the Commission had, under the first exemption criterion (i.e. the equivalent of Condition 1), weighed the dangers for competition flowing from the measure against the improvements it brought. This approach was upheld by the Court of Justice, in Case 45/85 Verband der Sachversicherer e V v Commission EU:C:1987:34; [1988] 4 C.M.L.R 264.
122. As the Court explained, at paragraphs 58 to 61 of its judgment:

"58. With regard to those arguments it must be emphasised that the Commission's task under Article 85(3) is to determine whether the contested recommendation contributes to improving the provision of services on the insurance market. In that connection the Commission correctly took the view that its task was not merely to check whether the aim of the recommendation was to deal with the actual problems confronting the market as a result of the continuing fall in premiums for industrial fire and consequential loss insurance and to consider whether the recommendation was a proper means of dealing with that situation, but also to assess whether the measures put into effect by the recommendation went beyond what was necessary to that end.

59. … The question to be considered is whether the collective, fixed-rate and across-the-board increase in premiums was justified by the objective pursued.

60. By reason of its general and undifferentiated nature the increase involved a rise in premium rates which encompassed not only cover for the expenses resulting from insurance claims but also the operating costs of the insurance companies. It is apparent from the documents before the Court that there were considerable differences in the level of operating costs between different insurance companies. The global nature of the increase was therefore likely to result in restrictions on competition going beyond what was necessary to achieve the intended objective.

61. By taking the view that in those circumstances the disadvantages arising from the solution chosen were, from the point of view of competition law, greater than the advantages and that, consequently, there was no improvement in the provision of services in the insurance market, the Commission did not exceed the limits of the discretion vested in it in connection with the application of Article 85(3) EEC."

123. In our view, this decision shows that the question whether a particular restriction contributes to improving the provision of services on a relevant market may legitimately be answered by weighing up the consequential advantages and disadvantages for competition which flow from the restriction. That is, in essence, what the CAT did in the present case under the heading of Condition 1. There was no error of law in their approach.
124. Furthermore, it is common ground that the Commission's Guidelines on the application of Article 101(3) emphasise the importance of conducting a balancing operation of competition law costs and benefits in the context of Condition 3. See, for example, paragraph 101, which states:

"Finally, and very importantly, it is necessary to balance the two opposing forces resulting from the restriction of competition and the cost efficiencies. On the one hand, any increase in market power caused by the restrictive agreement gives the undertakings concerned the ability and incentive to raise price. On the other hand, the types of cost efficiencies that are taken into account may give the undertakings concerned an incentive to reduce price…"

This to our mind reinforces the point that, in substance, it is a matter of indifference whether the relevant analysis is conducted in the context of Condition 1 or Condition 3. What matters is that the CAT carefully considered the evidence, and came to the conclusion which it did at [275]. We agree with Mr Woolfe that this conclusion alone is enough to show that Network Rail's claim to exemption under section 9 of the 1998 Act must fail.

125. In these circumstances, it is unnecessary for us to consider the further arguments raised by Network Rail in relation to Conditions 3 and 4.

Conclusion on Ground 5:

126. For the reasons given above, we would dismiss Network Rail's appeal on Ground 5.

(6) Abuse of dominant position

127. We have already set out the relevant provisions of section 18 of the 1998 Act, which imposes the Chapter II prohibition. We have also summarised the relevant part of the CAT's judgment: see [36] to [39] above. The sixth ground of appeal contains four separate contentions, which are summarised at [43] above. We will now consider them in turn.
128. The first contention is that the anti-competitive effects analysis carried out by the CAT is vitiated by the same errors of law as have already been considered in the context of whether there was an appreciable effect on competition for the purposes of the Chapter I prohibition. The contention therefore stands or falls with our views on the issue of appreciable effect in the context of the Chapter I prohibition, and since we have upheld the approach and conclusion of the CAT on that issue, it follows that this aspect of Network Rail's appeal on abuse of a dominant position must likewise fail.
129. The second contention is a challenge to the CAT's acceptance, at [296] of the judgment, of Achilles' argument "that opening a market up to competition only periodically in the form of a tender limits the dynamic evolution of the market and risks locking in a sub-optimal outcome". It is said that this conclusion was manifestly contrary to the evidence, and should therefore be rejected. It is well established that, for a challenge of this nature to succeed, it is necessary to show either that there was no evidence capable of sustaining the conclusion reached, or that the conclusion is plainly wrong, in the sense that it is one that no reasonable judge or tribunal could have reached: see, for example, Henderson v Foxworth Investments Limited [2014] UKSC 41, [2014] 1 WLR 2600, at [61] and [67] per Lord Reed JSC. As we shall explain, we do not consider that Network Rail comes near to satisfying this exacting standard.
130. It is first necessary to set paragraph [296] of the judgment in its immediate context. Under the sub-heading of "The Tender Exercise", the CAT began by recording Network Rail's reliance on the tender process conducted by the RSSB as showing that competition is maintained in respect of supplier assurance IT and audit services in the rail industry in Great Britain, and that this was inconsistent with Achilles' case on abuse: see [293]. The CAT continued:

"294. In response, Achilles pointed out that the RSSB did not tender for the provision of an end-to-end pre-qualification service, but for the provision of certain inputs and that the RSSB Board retains control over pricing for the RISQS scheme. Whilst the RSSB as a whole acts as a not-for-profit body, its constitutional objective is to promote the interests of its members, a group largely distinct from RISQS users, and there is no formal requirement on the RSSB that Network Rail could point to which would require the RSSB to operate RISQS as a not-for-profit activity.

295. In response to Mr Holt's suggestion that tendering for the IT and audit inputs to RISQS was particularly effective because more firms can bid to provide IT and audit inputs than an end-to-end solution, Achilles noted that in fact only one bidder was left for the provision of audit services following Achilles' withdrawal from the bidding process and that two bidders submitted bids for both lots, which would suggest that there was no advantage in the separate tenders in terms of numbers of potential bidders.

296. The Tribunal agrees with Achilles' contention that opening a market up to competition only periodically in the form of a tender limits the dynamic evolution of the market and risks locking in a sub-optimal outcome as a result, for example, a bidder under-bidding and having to compromise on delivery quality or not being subject to sufficient competition in the tendering process. There was no evidence that Network Rail or the RSSB ever considered these in opting for the tender of the RISQS inputs.

297. In our view, whilst Achilles remains able at some future date to compete to replace the RISQS scheme or to replace the suppliers of the outsourced components of the RISQS scheme (should Network Rail consider moving away from the RISQS scheme supplied by the RSSB or when the RSSB re-tenders the IT and audit components of the RISQS scheme), the RISQS-only rule has the effect, in the interim, of weakening competition in the market. The tender exercise that was carried out, and the possibility of a further tender exercise for the market in the future, do not affect, justify or compensate for the elimination of competition in the meantime."

131. We have quoted this section of the judgment in full, because it can be seen, in context, that the passage at the beginning of [296] with which Network Rail takes issue does not purport to be a finding of fact, but rather records the CAT's agreement with a submission made by Achilles as a matter of general principle. It is a step in the reasoning of the CAT, leading to the conclusions stated in [297].
132. The real thrust of Network Rail's case on this issue, as Mr Went made clear in his written and oral submissions, is that the reasoning of the CAT in the whole of the above passage overlooks important aspects of the expert economic evidence provided by Mr Holt about the benefits of the RISQS tender process, and the benefits of competition for the relevant market, without any expert evidence from Achilles to counter it. The contention recorded in [296], it is said, does not originate from any of the expert evidence, but was merely contained in the closing written submissions of Achilles at trial.
133. Mr Woolfe's answer to this argument is that Mr Holt's expert evidence on the issue was tested in cross-examination, and the facts regarding the tendering process (which formed the basis of Mr Holt's evidence) were also explored in cross-examination of the relevant factual witnesses, who included Ms Pearson of the RSSB who had conduct of the tendering process. This material was then reflected in paragraphs 65 to 69 of Achilles' closing submissions, which gave detailed references to the relevant evidence, including transcripts of cross-examination.
134. Mr Woolfe goes on to submit that these detailed points are reflected in the CAT's judgment. For example, the fact that there had been only one qualifying bidder for audit services in the tender process after Achilles' withdrawal, and that only two bidders submitted bids for both lots, suggested to the CAT that "there was no advantage in the separate tenders in terms of numbers of potential bidders": see [295]. Similarly, the CAT noted, at the end of [296], that there was no evidence that Network Rail or the RSSB ever considered the risks that excluding competition in the market in favour of a tendering process risked locking in a sub-optimal outcome, which amounted to an implicit rejection of Mr Holt's evidence that Network Rail or the RSSB would have taken those factors into account when deciding to go down the tender route.
135. Furthermore, Achilles had itself adduced evidence of the benefits which could be expected to accrue through allowing competition to occur in the market, and specifically from realising synergies between supplier assurance needs in different sectors, which is what Achilles sought to achieve through its TransQ scheme: see the judgment at [273]. Mr Holt accepted in cross-examination that he had not given consideration to such factors in advancing his own view.
136. On any issue of this nature, it is essential for an appellate court to remember that it cannot replicate the trial process, and it has access to only a small proportion of the relevant material of all kinds which contributed to the conclusion of the court below. Furthermore, an expert tribunal like the CAT is fully entitled to bring its own expertise to bear on such issues, and it is not bound to accept the expert evidence adduced by either side, whether or not it is countered by expert evidence from the other side. With those considerations in mind, we hope we have said enough to explain why the challenge to this part of the CAT's conclusions must fail.
137. For completeness, we also agree with Mr Woolfe that Network Rail can derive no assistance from a decision of the Commission in 2003 in the case of ARA, ARGEV and ARO [2004] L 75/59. Apart from anything else, that was a case under Article 101 TFEU, and not a decision in relation to abuse of a dominant position under Article 102. Furthermore, the factual situation under consideration in that case (a joint venture created by the packaging industry in Austria for the purposes of fulfilling regulatory obligations in respect of recycling) was far removed from the facts of the present case, and although there was some consideration of tendering on a recurrent basis, it was in the different context of the fourth condition for exemption from Article 101 (no elimination of competition). The failure of the CAT to deal with this decision cannot therefore be characterised as an error of law. Nor, indeed, is the point included in the grounds of appeal. The point is therefore not open to Network Rail in any event.
138. Network Rail's next contention is formulated as follows in the sixth ground of appeal:

"the Tribunal erred in law in finding that a dominant company need not benefit commercially from the relevant conduct for such conduct to be found abusive without also finding that the dominant company is an essential trading partner of the party alleging abuse (and therefore misapplied Case T-128/98 Aéroports de Paris v Commission)"

139. The allegedly erroneous conclusion of the CAT is in paragraph [306] of the judgment:

"In the Tribunal's view, Network Rail's argument that it is necessary for there to be some commercial benefit to be gained by the dominant undertaking from its conduct before that conduct can be condemned is inconsistent with the law as explained by the General Court in Aéroports de Paris and the High Court in Arriva the Shires. The case law does not support the distinction, contended for by Network Rail, depending on whether the dominant company is, or is not, an essential trading partner of the party alleging abuse."

140. In Aéroports de Paris, the General Court held (at [165]) that:

"… where the undertaking in receipt of the service is on a separate market from that on which the person supplying the service is present, the conditions for the applicability of Article 86 are satisfied provided that, owing to the dominant position occupied by the supplier, the recipient is in a situation of economic dependence vis-à-vis the supplier, without their necessarily having to be present on the same market. It is sufficient if the service offered by the supplier is necessary to the exercise by the recipient of its own activity."

Relying on this passage, Network Rail submits that the CAT was wrong to say that the case law does not support the distinction contended for by Network Rail, and the CAT should therefore have gone on to analyse whether Achilles is economically dependent on Network Rail or whether Network Rail provides any service to companies such as Achilles.

141. Achilles' answer to this argument, in short, is that (a) the CAT rightly rejected at trial Network Rail's case that there can in general be no abuse where the dominant company derives no competitive advantage from the alleged abusive conduct, (b) in coming to that conclusion the CAT correctly understood and applied Aéroports de Paris and Arriva the Shires, and (c) the passage in Aéroports de Paris at [165] relied upon by Network Rail was dealing with a different point, which becomes clear when it is read in context. As will appear, we accept those submissions.
142. Aéroports de Paris ("ADP") was a publicly owned undertaking responsible for the management of airports in the Paris region. It was dominant in the market for airport management, and had let a concession contract for ground-handling services to a company, AFS, which later complained that it was subject to discriminatory pricing. As the CAT explained in the present case at [304], the General Court rejected the argument advanced by ADP, as the operator of Orly Airport, that its conduct in charging different fees to different ground-handling concessionaires was not abusive because it had presence in the market for ground-handling services and no interest in distorting competition on that market.
143. In that context, the General Court held at [173], in a passage quoted by the CAT:

"In that regard, it should be recalled that the concept of abuse is an objective concept and implies no intention to cause harm. Accordingly, the fact that ADP has no interest in distorting competition on a market on which it is not present, and indeed that it endeavoured to maintain competition, even if proved, is in any event irrelevant. It is not the arrival on the market in groundhandling services of another supplier that is in issue, but the fact that at the time of the adoption of the contested decision, the conditions applicable to the various suppliers of those services were considered by the Commission to be objectively discriminatory."

This passage clearly provides no support for Network Rail's argument, because it emphasises that the concept of abuse is an objective one, and it was irrelevant that ADP had no interest in distorting competition on a market on which it was not present, and from which it therefore derived no competitive advantage.

144. This reasoning was then followed by Rose J (as she then was) in Arriva the Shires (Arriva the Shires Limited v London Luton Airport Operations Limited [2014] EWHC 64 (Ch)), where at [99] she rejected the argument that it was necessary for there to be some commercial benefit to be gained by the dominant undertaking from its conduct before that conduct can be held abusive. As Rose J said:

"In my judgment, the ruling of the General Court in Aéroports de Paris shows that it is not necessary for there to be some commercial benefit to be gained by the dominant undertaking from its conduct before that conduct can be condemned as abusive... The complete absence of any commercial gain on the part of the dominant undertaking may well be highly relevant in a particular case, for example on the issue of objective justification… I do not accept, however, that as a matter of law, a foreclosure of the downstream market distorting competition among competitors on that market should be an abuse only if it generates an economic gain on the part of the dominant undertaking. That is inconsistent with the case law which emphasises the objective nature of abuses and which establishes that motivation and intention are generally not relevant to the question of infringement (otherwise than in some clearly established instances such as predatory pricing)."

145. It was this line of reasoning which supported the conclusion of the CAT at [306], quoted at [139] above. How, then, can it be said that the reasoning of the General Court in Aéroports de Paris at [165] leads to the introduction of a further requirement, namely that a person on a separate market from that of the dominant supplier can only complain of abuse of that dominant position if it is in a situation of economic dependence on the supplier? The answer to this question, as Mr Woolfe submitted, is that paragraph [165] of Aéroports de Paris needs to be read in context, and properly understood it provides no support for any such requirement.
146. In the section of its judgment headed "Findings of the Court", the General Court in Aéroports de Paris began its discussion at [162] as follows:

"162. The applicant relies essentially on four arguments in support of its complaint that its conduct does not fall within the scope of Article 86 of the Treaty.

163. First, it contends that Article 86 cannot be applied to it because it is not present on the markets in respect of which the Commission found… that competition was affected. It is said to follow from the judgment of the Court of Justice in TETRA PAK v E. C. COMMISSION that Article 86 of the Treaty cannot be applied in such circumstances.

164. That argument is entirely unfounded in law. The Court of Justice quite clearly stated in Case C-333/94 P TETRA PAK v E. C. COMMISSION that [two other cases] provide examples of abuses having effects on markets other than the dominated markets. There is no doubt, therefore, that an abuse of a dominant position on one market may be censured because of effects which it produces on another market. It is only in the different situation where the abuse is found on a market other than the dominated market that Article 86 of the Treaty is inapplicable except in special circumstances…

165. In the present case, although the conduct of ADP to which the contested decision objects, namely the application of discriminatory fees, has effects on the market in groundhandling services and, indirectly, on the market in air transport, the fact remains that it takes place on the market in the management of airports, where ADP occupies a dominant position. Furthermore, where the undertaking in receipt of the service is on a separate market from that on which the person supplying the service is present, the conditions for the applicability of Article 86 are satisfied provided that, owing to the dominant position occupied by the supplier, the recipient is in a situation of economic dependence vis-à-vis the supplier, without their necessarily having to be present on the same market. It is sufficient if the service offered by the supplier is necessary to the exercise by the recipient of its own activity."

147. Once paragraph [165] is placed in its context, it can be seen that the General Court held that ADP had committed the relevant abuse on the market for airport management, where it was present, and that it therefore did not matter that the anti-competitive effects arose on a related market. That was enough to dispose of ADP's argument, and as Mr Woolfe submits there is also a close analogy to the present case. Network Rail is (assumed to be) dominant in a market for the provision of rail infrastructure. As part of that activity, it allows access to its infrastructure and has imposed certain terms, including the RISQS-only rule. In so doing, it has created an anti-competitive effect on a market on which it is not present, namely the market for supplier assurance services.
148. The second part of paragraph [165], upon which Network Rail wishes to rely, evidently relates to the different situation where the abuse is found on a market other than the dominated market, and the existence of special circumstances then has to be shown if an abuse of dominant position is to be established: see the Tetra Pak case at paragraph 27 of the judgment of the Court of Justice.
149. It follows, in our view, that the CAT correctly understood and applied the principles to be derived from Aéroports de Paris and Arriva the Shires, and it was therefore not necessary for Achilles to establish that it was an essential trading partner of Network Rail.
150. Network Rail's fourth, and final, contention may be summarised by saying that the CAT erred in law by failing to recognise that the present case is one of refusal by Network Rail to supply a new customer, as opposed to a refusal to contract with an existing customer. Network Rail supports this argument by reference to the discussion in Bellamy & Child, European Union law of Competition, 8^th edition, at paragraph 10.155 and Arriva the Shires at [158] to [161]. We do not propose to review Network Rail's arguments on this issue at any length, because it seems to us completely unrealistic to treat Achilles as if it were a wholly new customer of Network Rail's. That would be to ignore the whole history of the case, including the provision of supplier assurance services by Achilles to Network Rail from 1997 onwards, culminating in the withdrawal of Achilles in May 2017 from the competitive tender process organised by the RSSB. What Achilles now wishes to do, in substance, is to resume its long-standing position as a substantial supplier of such services to companies requiring access to Network Rail's infrastructure, which it is currently prevented from doing by the existence of the RISQS-only rule. There is no meaningful parallel between such a situation and the position if Network Rail were to seek supplier assurance services from an entirely new provider. Nor, in our judgment, does the further passage in Arriva the Shires upon which Network Rail here relies throw any useful light on this aspect of the case.

Conclusion on Ground 6:

151. For these reasons, we would also dismiss the sixth ground of appeal.

Overall conclusion

152. We would therefore dismiss Network Rail's appeal on all grounds.

Lord Justice Peter Jackson:

153. I agree that the appeal should be dismissed for the reasons given in the judgment above. I add to it only to commend the thoroughness and clarity of the CAT's own judgment, which has survived all assaults upon it unscathed, and to add my own observations about two matters.
154. The first concerns the issue of safety. As appears from paragraphs 15 and 33 above, the CAT dealt exhaustively with that question in paragraphs [155] – [255]. It set out the eight safety-related arguments made by Network Rail at paragraph [173] and it considered the evidence it had received about them at [180] – [225] before coming to its analysis at [226] – [227], saying:

"226. There was no issue as to the vital importance of the Key Schemes in ensuring the safety on the railway network. Achilles accepted that the purpose of the Key Schemes is to ensure the safety of workers carrying out work and ultimately the safety of rail passengers. Nor was there any issue as to the need for suppliers working on the railway to be subject to supplier assurance. Achilles did not dispute that supplier assurance is part of Network Rail's safety management system, that it plays an important role in ensuring that suppliers adhere to high safety standards and that it contributes to a reduction in accident rates.

227. The issue for the Tribunal is a narrower and more specific one, namely whether the requirement in the Sentinel Scheme and OTPO Scheme that suppliers must use the supplier assurance provided by RISQS is objectively necessary to achieve the safety purposes of the Key Schemes. In other words, is it essential to the fulfilment of those purposes that supplier assurance be provided by the RISQS scheme to the exclusion of any other suitably qualified provider including Achilles? Would it be impossible to achieve those safety objectives if Network Rail were required to recognise other suitably qualified and competent providers of supplier assurance schemes?"

155. In answering this question, the CAT carefully examined each issue raised by Network Rail between paragraphs [241] and [253]. It started by acknowledging the need for a uniform and clear set of safety requirements. It was not, however, persuaded that the consistency or clarity of standards would be compromised by allowing more than one provider of supplier assurance audit services. Consistency could be achieved by requiring that any provider of supplier assurance is effectively monitored against the relevant auditing standards. Network Rail could impose a monitoring and supervisory regime to ensure quality, so that any differences between suppliers would then be at least above the relevant safety levels. It also considered that, contrary to Network Rail's case, it is far from clear that suppliers would be more incentivised to maintain high standards if there were only one provider of supplier assurance services in the market than if there were more than one. Competing providers of supplier assurance services might well be incentivised to improve their services with a beneficial effect on safety, as argued by Achilles.
156. As noted above, the CAT's ultimate conclusion on this issue appears at [254] – [255]:

"254. In summary, for the reasons set out above, the Tribunal is not persuaded that the RISQS-only rule is objectively justified as being ancillary to the safety purposes of the Key Schemes. Network Rail has not established that those purposes would be impossible to achieve without the RISQS-only rule. The Tribunal considers that those safety purposes could be achieved by alternative providers of supplier assurance services working to the same standards as RISQS and subject to effective monitoring, with their IT platforms linked to RISQS's and/or their data freely accessible to Network Rail and with the RISQS forum open to participation by other providers of supplier assurance services.

255. The Tribunal appreciates that if a change to Network Rail's HSMS so as to permit multiple assurance providers was considered to be a significant change it would have to be reviewed by the ORR [Office of Rail and Road – the railway regulator]. The Tribunal expects that this review process would be handled appropriately."

Furthermore, the CAT's order preserves the right of Network Rail to impose on those seeking recognition as an assurance provider "such conditions as are objectively justified by the need to ensure safety on the railway network".

157. I have referred in a little detail to this aspect of the proceedings before the CAT precisely because Network Rail has not sought to appeal from that part of its judgment. Where the railway is concerned, nothing is more important than safety, but it should be clearly understood that this appeal has been about competition, not about safety.
158. The second point I would acknowledge is that there may be some irony, perhaps felt most keenly by Network Rail and by those who succeeded in the tender process from which Achilles withdrew, in the fact that Achilles, after enjoying a monopoly position as assurance provider for the UK rail industry for over 20 years should now be able to compete in a market that it previously had to itself. But that is neither here nor there in the light of the CAT's expert findings.