NCCN 500 This is the non-confidential version. Confidential information and data have been redacted. Redactions are indicated by [] 1 August 2008 Contents Section 1 2 Page Glossary of terms 1 Executive summary 8 Market definition and dominance BT’s conduct 8 9 Background NTS 12 NTS hosting and revenue sharing The NTS regulatory regime NTS wholesale charging arrangements NTS call termination The NTS Calculator Transit 3 4 12 14 15 16 16 17 17 NTS regulatory policy NCCN 500 NCCN 651 The first complaint The NTS call termination market review 18 19 20 21 21 Ofcom’s investigation 23 The complaint The investigation Period of investigation Evidence gathered Evidence gathered from BT Evidence gathered from C&W Evidence gathered from third parties The draft decision Evidence gathered following the draft decision 23 25 25 25 25 28 30 30 31 The relevant market 34 Summary Identifying markets 34 34 Identifying the relevant product market Identifying the relevant geographic market Markets relevant to this investigation Introduction NTS markets Downstream markets: retail NTS calls Upstream wholesale markets The relevant market Introduction Analysis of NTS call termination/hosting as a two-sided market NTS call termination/hosting market: analysis Termination/hosting of unmetered internet traffic and calls to directory enquiries fall outside the relevant market Conclusion Relevant geographic market Conclusion Responses to draft decision: Ofcom’s approach to market definition 35 36 37 37 37 40 41 43 43 44 51 60 61 61 62 62 BT C&W CPW Thus Ofcom comments on responses Application of the SSNIP test Ofcom’s analysis of the market as a single market, rather than as two separate markets 5 66 67 67 BT’s position in the relevant market 74 Introduction Market shares 74 74 Volume market shares Conclusion on BT’s market share Barriers to entry and expansion Barriers to entry Barriers to expansion Constraints on BT from competing suppliers of NTS termination/hosting Special cases: BT-originated traffic Special cases: traffic transited by BT Initial conclusion: traffic originated and transited by BT Constraints that apply in the case of BT-terminated traffic Constraints on BT from customers OCP responses to a price increase by BT Absorbing increased costs Refusing to carry BT-terminated traffic Pass price increase on to own retail callers Effect if price increase were passed on Observed behaviour following termination price increases Conclusion on pass-through Evidence on profit effect of NCCN 500 Constraints imposed by NTS service providers Conclusion on the constraints faced by BT NCCN 651 6 62 64 65 65 75 77 77 77 79 80 82 83 85 85 85 85 87 88 91 92 94 96 96 97 98 99 Responses to draft decision: Ofcom’s provisional conclusions on dominance 100 The conduct 105 Introduction Alleged margin squeeze Alleged margin squeeze in retail markets 105 106 106 Alleged margin squeeze in retail markets: the allegation Alleged margin squeeze in retail markets: legal and economic framework Alleged margin squeeze in retail markets: analysis Alleged margin squeeze in retail markets: scope of margin squeeze test Alleged margin squeeze in retail markets: methodological issues Conclusion on retail margin squeeze Responses to draft decision: retail margin squeeze BT C&W Thus Ofcom comments on responses to the draft decision Possible margin squeeze in NTS hosting Introduction Hosting margin squeeze in a two-sided market Definition of the hosting margin Range of services Cost of capital Cost standard 106 107 111 112 116 129 129 129 129 130 131 134 134 134 135 135 136 136 1 Relevant time period Data and results Revenues Costs Summary of results Responses to draft decision: hosting margin squeeze BT C&W Ofcom comments on responses Alleged discrimination Alleged discrimination: the allegation Alleged discrimination: legal framework Alleged discrimination: analysis Responses to draft decision: Ofcom’s analysis of alleged discrimination C&W Thus Ofcom comments on responses Alleged excessive pricing Alleged excessive pricing: the allegation Alleged excessive pricing: legal and economic framework Alleged excessive pricing: assessing BT’s prices compared to costs FAC profitability analysis: introduction FAC profitability analysis: source of information FAC profitability analysis: basis of preparation FAC profitability analysis: results FAC profitability analysis: summary of results Alleged excessive pricing: profitability analysis based on SAC SAC profitability analysis: methodology Modelling the hypothetical single product firm’s network costs Modelling the hypothetical single product firm’s non-network costs Results Model input analysis Alleged excessive pricing: conclusion Were BT’s prices excessive compared to costs? Were the prices notified in NCCN 500 unfair in themselves or when compared with competing products? Conclusion on excessive pricing Responses to draft decision: excessive pricing C&W CPW Thus Ofcom comments on responses Other allegations made by C&W NCCN 500 increased C&W’s costs of providing a competing service NCCN 500 increased BT’s market power NCCN 500 was part of a concerted strategy to dilute competition Responses to draft decision: impact on BT’s competitors and BT’s intentions C&W CPW Thus Virgin Media Ofcom comments on responses Conclusion Margin squeeze Discrimination Excessive pricing Retail calls to NTS numbers FAC profitability analysis: cost basis for wholesale services FAC profitability analysis: stages of analysis 2 136 137 137 138 141 141 141 141 142 144 144 145 147 151 151 152 153 155 155 155 160 163 164 164 165 168 169 169 170 179 181 184 185 186 187 190 191 191 193 194 195 199 199 200 201 204 204 206 206 207 207 209 209 210 210 216 236 237 FAC profitability analysis: differences between voice and data Annex 238 Page 1 NCCN 500 ................................................................................................... 211 2 NCCN 651 ................................................................................................... 213 3 Calculation of critical elasticity for calls to 0845 numbers .................... 215 4 Related markets......................................................................................... 216 Other wholesale calls markets ................................................................................ 220 5 ROCE and ROS .......................................................................................... 224 6 The importance of capital employed in BT’s retail services.................. 226 7 Alleged margin squeeze in call origination: data sources and calculations................................................................................................ 229 8 Alleged margin squeeze in call origination: sensitivity analysis .......... 231 NTS calls account for a larger proportion of all calls ............................................... 231 Sensitivity: business customers generate same proportion of data calls as voice calls.................................................................................................... 232 Sensitivity: C&W time of day weights ...................................................................... 234 9 FAC profitability analysis: basis of preparation ..................................... 236 3 Glossary of terms Act: the Competition Act 1998. Article 82: Article 82 of the EC Treaty, which provides that “any abuse by one or more undertakings of a dominant position within the common market or in a substantial part of it shall be prohibited as incompatible with the common market in so far as it may affect trade between Member States”. Average total cost (“ATC”): average total cost is the sum of fixed costs (costs that do not vary with changes in output) and variable costs (costs that vary with changes in output). Average total cost is total cost divided by output. Average variable cost (“AVC”): variable costs are costs that vary with changes in output, e.g. materials, energy, direct labour, supervision, repair and maintenance and royalties. Average variable cost is variable cost divided by output. BT: BT Group plc and its subsidiary undertakings. BT Retail: a retail business of BT providing a range of retail services to residential, business and corporate customers. Commission: the European Commission. Communications provider: a person who provides an electronic communications network or an electronic communications service, as defined in the General Conditions of Entitlement. CPS: carrier pre-selection, a mechanism that enables a customer to transfer some or all of his calls to an alternative communications provider, while retaining his existing telephone line, without having to dial additional codes or use special equipment. CPSO: carrier pre-selection operator, a communications provider that interconnects with BT’s network and that has put in place the necessary arrangements to allow customers who have a BT line to subscribe to CPS services that it provides. Chapter II Prohibition: Section 18 of the Competition Act 1998, which provides that “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”. CAT: the Competition Appeal Tribunal, which was created by Section 12 and Schedule 2 to the Enterprise Act 2002 which came into force on 1 April 2003. CFI: the European Court of First Instance. Dial-up internet access: internet access that uses a dial-up connection over an analogue or ISDN (Integrated Services Digital Network) telephone line. DLE: digital local exchange, the telephone exchange to which customers are directly connected, often via a remote concentrator unit. DQ: directory enquiries. 4 Discounted cashflow (“DCF”): an approach to measuring profitability under which future costs and revenues are expressed as a present value by applying an appropriate discount rate. Enterprise Act: the Enterprise Act 2002. ECJ: the European Court of Justice. FAC: fully allocated cost, an accounting method for attributing all the costs of the company to defined activities such as products and services. Typically this method is guided by the principle of cost causation. FRIACO: flat rate internet access call origination, a wholesale product purchased by TCPs from BT that acts as an input for retail flat-rate dial-up internet access. Freephone: Freephone numbers are numbers starting 0800 and 0808 which are generally free to call from BT lines. Geographic numbers: geographic numbers are telephone numbers that relate to a particular geographic location, indicated by a geographic area code starting 01 or 02. ISP: internet service provider, a company that provides business and residential customers with access to the internet and other related services. IP: Internet Protocol, the packet data protocol used for routing and carriage of messages across the internet and similar networks. IP network: a network that uses IP, for example the internet. Inter-tandem transit: an interconnection service that involves the use of two tandem switches and one inter-tandem transmission link. Long run incremental cost (“LRIC”): the costs caused by the provision of a defined increment of output, taking a long run perspective, assuming that some output is already produced. The ‘long run’ means the time horizon over which all costs (including capital investment) are variable. Net present value (“NPV”): the result from a DCF analysis that discounts costs and revenues using the applicable discount rate and sums these together. NCCN: Network Charge Change Notice, the mechanism by which BT notifies other communications providers of changes to its charges, including charges for the termination of calls to NTS numbers. National Telephone Numbering Plan (“NTNP”): a document published by Ofcom that provides details of all telephone numbers administered by Ofcom and specifies restrictions on the adoption or use of numbers that Ofcom considers appropriate for each range. Non-geographic numbers: a non-geographic number means any telephone number that is not a geographic number. NTS: Number Translation Services, telephone services using prefixes listed in the NTNP as “Special Services”, which start with 08 and 09. 5 NTS call origination condition: a requirement imposed on BT that requires it to originate and retail calls to NTS numbers on behalf of other TCPs, which was imposed by Oftel (the agency responsible for regulation of telecommunications in the UK before the establishment of Ofcom) in 2003 after it concluded that BT had SMP in the wholesale call origination market. NTS call termination market review consultation: Ofcom’s consultation document NTS call termination market review, published 22 October 2004. NTS retail uplift: part of the cost incurred by BT in originating an NTS call is the cost of retailing, which BT is allowed to recover under the conditions of the NTS call origination condition. NTS service provider: a provider of voice or data services to callers who dial NTS numbers. OCP: originating communications provider, the communications provider that owns/operates the network where a call begins. Ofcom: the Office of Communications. OFT: the Office of Fair Trading. POC: point of connection, the point at which another communications provider’s network interconnects with that of BT, often at a Digital Main Switching Unit (“DMSU”). PRS: premium rate services, services supplied over premium rate (09) numbers, used mainly to provide access to competitions, TV voting lines, scratchcards, adult entertainment, chatlines and some post-sales services such as technical support. PSTN: the public switched telephone network. Relevant period: 1 May 2004 to 1 January 2006, the period for which the prices notified in NCCN 500 were in force Return on capital employed (“ROCE”): the ratio of accounting profit to capital employed. The measure of capital employed can be either Historic Cost Accounting (“HCA”) or Current Cost Accounting (“CCA”). Return on sales (“ROS”): the ratio of operating profit to sales. Service provider: a provider of electronic communications services to third parties whether over its own network or otherwise. SAC: standalone cost, the cost of providing a particular service, which is the sum of the incremental cost of the service (i.e. is the cost of producing a specified additional product, service or increment of output over a specified time period) plus all of the costs which are common between that service and other services. 6 SMP: significant market power, as defined in Article 14(2) of the Framework Directive, which states that: “an undertaking shall be deemed to have significant market power if, either individually or jointly with others, it enjoys a position equivalent to dominance, that is to say a position of economic strength affording it the power to behave to an appreciable extent independently of competitors, customers and ultimately consumers.” Single transit: interconnection service that involves the use of a single tandem switch. TCP: terminating communications provider, the communications provider that owns/operates the network where a call ends. Weighted average cost of capital (“WACC”): a company's WACC measures the rate of return that a firm needs to earn in order to reward its investors. It is an average representing the expected return on all of its securities, including both equity and debt. WLR: wholesale line rental, a mechanism that enables alternative suppliers to rent access lines on wholesale terms from BT, as mandated by Ofcom’s Review of the fixed narrowband wholesale exchange line, call origination, conveyance and transit markets, and to resell the access lines to end consumers. 7 Section 1 1 Executive summary 1.1 The Office of Communications (“Ofcom”) has concluded that BT Group plc (“BT”) has not infringed section 18 (“the Chapter II prohibition”) of the Competition Act 1998 (“the Act”) or Article 82 of the EC Treaty (“Article 82”) in relation to its prices for Number Translation Services (“NTS”) call termination between 1 May 2004 and 1 January 2006, as notified in Network Charge Change Notice 500 (“NCCN 500”) issued on 1 April 2004. 1.2 Ofcom opened an investigation on 8 April 2005 in response to a complaint submitted to Ofcom on 15 March 2005 by Cable & Wireless Communications plc (“C&W”) alleging that BT was dominant in a number of relevant markets including the market for NTS call termination, and that BT had abused its dominant position in relation to the price increases for NTS call termination on certain number ranges, as notified in NCCN 500.1 1.3 Ofcom reached a provisional conclusion that the available evidence did not support the view that BT’s conduct over the relevant period infringed the Chapter II prohibition and/or Article 82 and that there were therefore no grounds for action. On 23 July 2007 Ofcom provided a non-confidential version of its provisional decision to BT, C&W and four other parties supporting C&W’s complaint and invited them to submit written responses on its provisional decision. 1.4 After considering respondents’ comments on its provisional decision, Ofcom has reached the conclusion that the available evidence does not support the view that BT’s conduct over the relevant period infringed the Chapter II prohibition and/or Article 82 and that there are therefore no grounds for action. Market definition and dominance 1.5 Ofcom has concluded that the relevant market for the consideration of C&W’s allegations is the market for the termination/hosting of NTS calls on all number ranges, by all terminating communications providers (“TCPs”) in the UK. Ofcom’s reasoning on the relevant market is set out at Section 4 below. 1.6 BT’s share of the NTS call termination/hosting market, based on total volumes, was below a level that would typically be associated with a position of dominance in the relevant market. However, Ofcom found that BT is able to act independently of other TCPs, its competitors, its customers and ultimately of consumers due to the particular structure and features of the market. 1.7 Ofcom found that BT was able to introduce price increases for NTS call termination without its TCP competitors, or its originating communications provider (“OCP”) and NTS service provider customers being able to constrain its pricing behaviour. 1 The complaint that generated this investigation was submitted by Energis Communications Ltd (“Energis”, UK company number 02630471). On 16 August 2005, Cable & Wireless plc (“C&W”, UK company number 00238525) announced an agreement to acquire Energis. On 25 October 2005, the OFT announced that it would not refer the acquisition to the Competition Commission. The complainant is referred to throughout this document as “C&W”, except where it is necessary for clarity to make a distinction between the two entities that existed prior to C&W’s acquisition of Energis. 8 1.8 1.9 BT’s competitors and customers were unable to constrain BT’s pricing behaviour because of: • BT’s position in retail markets, which meant that OCPs were reluctant to pass on the price increase to their retail customers (which would not in any case have undermined BT’s ability to increase its NTS call termination charges); • BT’s significant market power (“SMP”) in the market for call origination which, combined with the National Telephone Numbering Plan (“NTNP”) requirements on retail prices, prevents other TCPs from raising their gross termination charges on BT-originated calls to NTS numbers; • BT’s SMP in the market for single transit, which prevents other TCPs from raising their gross termination charges on non BT-originated calls to NTS numbers; and • the fact that BT would have been able to match or better any outpayment to an NTS service provider that a competing TCP could have made. Ofcom has therefore found that BT was dominant in the market for NTS call termination/hosting in the UK between 1 May 2004 and 31 December 2005, the period of the investigation. Ofcom’s reasoning on BT’s position in the relevant market is set out at Section 5 below. BT’s conduct 1.10 C&W alleged that the price increases notified in NCCN 500: • imposed a margin squeeze on C&W; • discriminated in favour of BT and against C&W on price; • were excessive; • increased C&W’s costs of providing a competing service or forced it to provide an inferior service to that of BT; • increased BT's market power in the NTS call origination and NTS call termination markets; and • formed part of a concerted strategy to dilute competition. 1.11 Before finding an infringement of Article 82 and/or the Chapter II prohibition, Ofcom must satisfy itself on the evidence that an infringement is more likely than not to have occurred.2 The undertaking being investigated is entitled to the presumption of innocence and to any reasonable doubt that there may be. Ofcom must therefore proceed on the basis of an analysis that is robust and soundly based.3 1.12 Ofcom’s analysis of BT’s conduct is set out at Section 6 below. In summary, beginning with the question of an alleged margin squeeze, Ofcom analysed whether an operator as efficient as BT would be able to compete in the relevant downstream markets identified in the investigation, under the cost constraints imposed by NCCN 2 3 Napp Pharmaceutical Holdings Ltd v Director General of Fair Trading [2002] CAT 1, paragraph 109. JJB Sports plc v OFT [2004] CAT 17. 9 500. Ofcom also considered a stricter test using the costs of a reasonably efficient competitor and BT also passed this stricter test.4 1.13 Ofcom found that BT earned high margins across all calls (even discounting all the returns associated with local calls), and that, had NCCN 500 applied to all calls originated by BT, the impact on BT’s margin across the relevant product set would have been insufficient to lead to a margin squeeze. 1.14 In addition, Ofcom carried out an analysis of whether the additional revenue that BT gained from NCCN 500 enabled it to fund higher payments to its NTS service providers, leading to a margin squeeze in NTS hosting. Ofcom found that BT’s conduct did not lead to a margin squeeze in NTS hosting. 1.15 On the question of discrimination by BT, in launching this investigation Ofcom was concerned by public statements made by BT which appeared to indicate that BT had charged different rates to other OCPs than to BT. BT also appeared to consider internally that it had “discriminated” in its charge to other OCPs. Ofcom investigated BT’s regulatory costing systems to assess the actual internal transfer charge paid by BT. There was not robust evidence of a “hard” charge to BT that was different from the charge that applied to BT’s competitors. Ofcom considered whether BT’s charges for NTS call termination, as notified in NCCN 500, might amount to “applying dissimilar conditions to equivalent transactions with other trading parties, thereby placing them at a competitive disadvantage”.5 1.16 Ofcom considers that in the circumstances of NCCN 500 the relevant test for determining whether BT was discriminating in favour of itself in an anti-competitive manner is whether BT would have been able to make a profit had it paid the charges notified in NCCN 500, taking into account profits earned on all the relevant services, i.e. whether BT’s conduct led to the operation of a margin squeeze on other OCPs. As already noted, Ofcom found insufficient evidence to support a finding of a margin squeeze. Notwithstanding BT’s statements on NCCN 500, given the results of the margin squeeze tests, Ofcom did not find compelling evidence to demonstrate that BT had discriminated between itself and other OCPs putting them at a competitive disadvantage. 1.17 Ofcom did not find any evidence that BT’s conduct led to a distortion of competition, and found limited evidence of any adverse effect on consumers. Ofcom concluded that BT’s intention was to increase revenue and draw Ofcom’s attention to anomalies in current NTS regulation, not to reduce competition. 1.18 Ofcom considered that it would have needed compelling evidence of excessive pricing, according to the tests set out in the case law. Ofcom considered these tests, to the extent that they were relevant in this case, offered conflicting evidence as to whether the prices notified in NCCN 500 were excessive. Ofcom found that BT’s prices appeared to be significantly above its fully allocated cost (“FAC”), but below another relevant measure of cost, standalone cost (“SAC”). 4 Ofcom notes the tests set out by the Court of Appeal in Albion (Dwr Cymru Cyfyngedig and Albion Water Limited and Water Services Regulation Authority [2008] EWCA Civ 536) and by the CFI in Deutsche Telekom (Case T-271/03 Deutsche Telekom v Commission Judgement of CFI 10 April 2008). See paragraph 6.23 et seq below. 5 Article 82 2(c). 10 1.19 Overall, in the circumstances of this case and in light of the need for compelling evidence, there is insufficient evidence to demonstrate to the required standard of proof that BT’s conduct constituted an abuse of its dominant position. 11 Section 2 2 Background NTS 2.1 NTS numbers are those number ranges listed in the NTNP as “Special Services” numbers.6 NTS numbers start with 08 or 09.7 2.2 An NTS number does not relate to a specific geographic location, but to a particular service. The NTS number dialled by a caller is ‘translated’ by the network to a geographic number to deliver the call to its destination. 2.3 The NTNP lists the numbers available for allocation and also specifies such restrictions on the adoption or use of numbers that Ofcom considers appropriate for each range. The designations for NTS numbers are summarised in Table 1 below. In the case of NTS numbers the designations specify retail prices for calls from BT lines.8 2.4 NTS numbers are used to provide access to a range of voice and data services. Research undertaken by Ofcom as part of its review of NTS regulatory policy (see paragraph 2.34 et seq below) suggested that different NTS number ranges are, to some extent, associated with different types of services: 6 • Freephone numbers (0800 and 0808) are mainly used to access private sector voice services such as sales lines and helplines; and telephony services provided by two-stage indirect access service providers; • 0844/0843 and 0845 numbers are used extensively for metered dial-up internet access (predominantly using 0845 numbers), and also support a wide range of other services, including pre- and post-sales enquiry lines, public sector services, transaction services and information services; • 0870 and 0871/0872/0873 numbers are mainly used to provide access to preand post-sales enquiry lines, some public sector services and international telephony services provided by resellers; and • Premium Rate (09) numbers are used mainly to access competitions, TV voting lines, scratchcards, adult entertainment, chatlines and some post-sales services such as technical support.9 The current version of the NTNP was published on 17 June 2008 at: http://www.ofcom.org.uk/telecoms/ioi/numbers/numplan170608.pdf. 7 NTS also includes calls to the legacy 0500 (Freephone) and 0345 (local rate) ranges, which are not available for new allocations and are no longer listed in the NTNP. NTS does not include calls to 0844 04 numbers for Surftime internet access or calls to 0808 99 numbers for unmetered dial-up internet access based on FRIACO (flat rate internet access call origination). 8 In the consultation Changes to 0870, published on 2 May 2008 (see http://www.ofcom.org.uk/consult/condocs/0870calls/0870condoc.pdf) Ofcom has proposed changes to the charges that communications providers can make for calls to 0870 numbers. The designations set out in Table 1 below and referred to elsewhere in this document are those that applied during the period under investigation. 9 See Number Translation Services: Options for the future, 22 October 2004, published at: http://www.ofcom.org.uk/consult/condocs/ntsoptions/nts_future/nts_future_op.pdf. 12 2.5 0820 numbers are reserved for Schools Internet Caller, a legacy service for schools that enables them to make free dial-up calls to the internet during school hours, reverting to normal tariffs in the evenings and at weekends (see footnote 243). Table 1: The NTS number ranges Range Designation 0800 and 0808 Free to caller (unless charges are notified to the caller at the start of the call) 0820 Schools internet 0844/0843 Special Services basic rate: charged at up to and including 5p per minute or per call for BT customers, set by TCP including VAT (the price charged by other OCPs may vary) 0845 Special Services basic rate: charged (before discounts and call packages) at BT’s Standard Local Call Retail Price for BT customers including VAT (the price charged by other OCPs may vary) 0870 Special Services higher rate: charged (before discounts and call packages) at BT’s Standard National Call Retail Price for BT customers including VAT (the price charged by other OCPs may vary) Special Services higher rate: up to and including 10p per minute or per 0871/0872/0873 call for BT customers including VAT (the price charged by other OCPs may vary) Special Services at a Premium Rate: charged to BT customers 090 and 091 • Generally at more than 10p per minute, up to and including £1.50 per minute including VAT; or • at more than 10p per call, up to and including £1.50 per call including VAT. Sexual Entertainment Services at a Premium Rate: charged to BT customers 098 • generally at more than 10p per minute, up to and including £1.50 per minute including VAT; or • at more than 10p per call, up to and including £1.50 per call including VAT. Source: NTNP. 2.6 There may be several communications providers involved in conveying an NTS call from the caller to the organisation or individual receiving the call (“the NTS service provider”), including an OCP, on whose network the call originates, and a TCP, on 13 whose network the NTS number terminates.10 The OCP and the TCP may be the same for some calls. A third communications provider (a “transit” provider) may carry the call between the OCP and the TCP. 2.7 These relationships are shown in Figure 1 below: Figure 1: Parties involved in an NTS call retail price OCP Transit CP (if used) revenue share payment termination charge TCP NTS service provider NTS hosting and revenue sharing 2.8 The regulatory framework for NTS enables the TCP to share the revenue from calls to NTS numbers with NTS service providers. NTS thereby acts as a micro-payment mechanism for the various services that can be accessed via 08 and 09 numbers. The caller pays the OCP for the call. The OCP passes on a terminating payment to the TCP, who is then able (subject to commercial viability) to share some of this revenue with the NTS service provider. 2.9 The TCP has two distinct roles in this arrangement. First, it provides call termination to the OCP. Second, it provides various services to NTS service providers, which we refer to collectively as “NTS hosting” (see also paragraph 4.28 below). 2.10 NTS hosting includes the payment of revenue shares. It also includes the provision of a range of value added services. 2.11 NTS service providers vary hugely in their scale and in the services that they offer to callers, and therefore have very different NTS hosting requirements. Some voice services may consist of simple number translation and routing to a particular geographic number (e.g. an 0845 number allocated to a small organisation or an individual, at a single geographic number). At the other end of the scale, the customer may be a large national organisation (for example a retailer, high street bank or government department), receiving very large volumes of calls, with call centres at multiple locations, and will require a wider range of call routing and call management services. 2.12 NTS hosting may therefore include a range of value added services, including: a) conditional call routing to one or more destination numbers based on static or dynamic parameters such as: o the caller’s telephone number or location; 10 For the purposes of this decision, Ofcom is using the definition of “communications provider” in section 405 of the Communications Act 2003, which is “a person who provides an electronic communications network or an electronic communications service”. Section 32 of the Communications Act 2003 defines an electronic communications network as “(a) a transmission system for the conveyance, by the use of electrical, magnetic or electro-magnetic energy, of signals of any description; and (b) such of the following as are used, by the person providing the system and in associated with it, for the conveyance of the signals – (i) apparatus comprised in the system; (ii) apparatus used for the switching or routing of the signals; and (iii) software and stored data.” 14 o time or date; o routing plans defined by the NTS service provider; o call distribution to multiple sites based on static rules, or call centre load; or o customer keypad selections or voice commands; b) recorded announcements; c) call management functionality such as live or historical traffic information and dynamic customer control of call routing parameters. 2.13 In the case of data services, NTS hosting may consist of termination of calls on a suitable modem, user authentication, internet protocol (“IP” – see Glossary) traffic aggregation and management, and IP traffic conveyance. The NTS regulatory regime 2.14 The NTS regulatory regime is based on two formal regulatory instruments, which are described in the following paragraphs. One of these (the NTS call origination condition) applies to BT only. The other (the NTNP) applies to all communications providers. 2.15 The current NTS regulatory regime dates back to 1996. The framework originally put in place by Oftel aimed to encourage the provision of services over the telephone.11 It did this by enabling TCPs to receive the profit from calls to NTS numbers and share it with NTS service providers, who could in turn use that revenue share to fund new and innovative services. 2.16 This model, and the key elements of the supporting policy, were retained following the introduction of the new regulatory regime for electronic communications networks and services that came into force on 25 July 2003. 2.17 Under the new regime, Oftel carried out a series of market reviews, one of which covered the wholesale market for call origination.12 As a result of this review, Oftel concluded that BT has SMP in the wholesale call origination market and imposed a number of SMP conditions on BT, including a requirement to originate and retail calls to NTS numbers on behalf of other TCPs known as the “NTS call origination condition”.13 2.18 The NTS call origination condition allows BT to deduct the costs it incurs in originating the call, retailing the call (the “NTS Retail Uplift”) and making provision for bad debt for Premium Rate Services (“PRS”) calls (over and above the amount already allowed for by the NTS Retail Uplift) before passing the remainder of the retail charge (the amount that the caller pays) on to the relevant TCP, which can use it to fund revenue share payments to its NTS service providers. BT’s call origination 11 Oftel (the Office of Telecommunications) was the authority responsible for the regulation of telecommunications in the UK prior to the creation of Ofcom at the end of 2003. 12 Review of the fixed narrowband wholesale exchange line, call origination, conveyance and transit markets: Identification and analysis of markets, determination of market power and the setting of SMP conditions. Final Explanatory Statement and Notification, 28 November 2003, published at: www.ofcom.org.uk/consult/condocs/narrowband_mkt_rvw/nwe/. 13 Review of the fixed narrowband wholesale exchange line, call origination, conveyance and transit markets, statement of 28 November 2003 (see footnote 12), SMP Services Condition AA11. 15 charges and NTS Retail Uplift charge are regulated by a charge control – another of the remedies imposed by Ofcom to address BT’s SMP in the wholesale call origination market (see previous paragraph).14 The other deductions that BT makes are also regulated.15 2.19 The second element of the NTS regulatory regime is the NTNP. As set out in Table 1 above, the NTNP specifies what the various different NTS number ranges can be used for. Communications providers to which Ofcom has allocated NTS numbers assume a responsibility to ensure that those numbers are used in accordance with the designations given in the NTNP. NTS wholesale charging arrangements 2.20 For every end-to-end NTS call, there may be a number of transactions between the different parties involved, as set out in Figure 1 above. 2.21 The caller is billed for the call by his OCP, which retains a proportion of the retail call charge to cover the cost of originating the call and providing retail services (such as billing and customer service) to the customer (although only when BT is the OCP is the amount retained for origination explicitly cost-based). The OCP passes the residual amount to the TCP as an NTS call termination payment. Finally, the TCP may have a revenue sharing arrangement with the NTS service provider. Where the call is transited by a third party, there is an additional transaction in the chain: the transit provider retains a proportion of the retail price for the provision of the transit service before handing the call on to the TCP for completion (see paragraph 2.31 et seq below). NTS call termination 2.22 BT’s charges for the termination of calls to NTS numbers are not regulated. Nor are those of any other TCP. 2.23 Non-BT TCPs’ charges for the termination of calls to NTS numbers are, however, determined to a large extent by regulation. As set out above, the maximum retail price of a call by a BT customer to an NTS number is subject to the limits set out in the designations in the NTNP. The amount in pence per minute (“ppm”) that BT can retain out of the retail charge is limited by the NTS call origination condition and the charge controls as discussed in paragraph 2.18 above. The amount that is passed through to the TCP (the termination charge) is therefore equivalent to the retail charge, minus BT’s retention, minus any payment to a transit provider. For a further discussion of TCPs’ ability to influence the level of their charges for the termination of NTS calls, see Section 5 below. 2.24 Similarly where a TCP other than BT terminates calls that have originated on another OCP’s network and transited via BT, BT pays the same termination charge as if the call originated on BT’s network. This is because BT’s wholesale billing systems are unable to take account of other OCPs’ retail prices or originating retentions and therefore can only pay the same termination charge for all calls to any one TCP. 14 Review of BT’s network charge controls: Explanatory Statement and Notification of decisions on BT’s SMP status and charge controls in narrowband wholesale markets, 18 August 2005, published at: www.ofcom.org.uk/consult/condocs/charge/statement/. 15 BT’s charges for NTS Retail Uplift and the provision for PRS bad debt are set out in Charges between Communications Providers: Number Translation Services Retail Uplift charge control and Premium Rate Services bad debt surcharge, 28 September 2005, published at: www.ofcom.org.uk/consult/condocs/NTSfin/statement_nts_uplift/. 16 2.25 BT’s charges for the termination of NTS calls originated by other OCPs, on the other hand, are not influenced by regulation in the wholesale call origination market. No OCP other than BT is subject to regulation of charges for originating NTS calls, so the proportion of the retail price that is passed through to BT (i.e. BT’s termination charge) is equivalent to the unregulated retail price of the call, minus the OCP’s retention, minus any payment to a transit provider (but see discussion at paragraph 5.56 et seq below). 2.26 Before it issued NCCN 500 (see paragraph 2.41 et seq below), BT behaved as though its charges for NTS call termination were regulated, in that it applied the same termination charges for OCP-to-BT calls as its competitors charged for BT-to-TCP calls (the latter being indirectly set by the combination of the NTNP and regulation on BT in the wholesale call origination market). NCCN 500 raised BT’s charges for terminating OCP-to-BT calls above TCPs’ charges for terminating BT-to-TCP calls. The NTS Calculator 2.27 In order to assist TCPs in calculating their NTS termination charges BT has made available, on its BT Wholesale website, an online tool called the NTS Calculator.16 This is an extensive Excel spreadsheet which enables TCPs to input their own call routing parameters and to calculate how much they can expect to receive for terminating any 08 or 09 call. The NTS Calculator is maintained both as an up-todate tool, with the latest wholesale network charges, discounts, NTS Retail Uplift and PRS bad debt surcharge, and as a historic tool which enables TCPs to assess how these factors have changed over time and measure the resultant effect on their termination charges. 2.28 Charges for the termination of calls to NTS numbers differ by number range, by time of day, and in some cases by call length. 2.29 Calls are classed as either “Daytime”, “Evening” or “Weekend”. “Daytime” calls are calls made by residential customers between 6am and 6pm, or by business customers between 8am and 6pm. “Evening” calls are calls made after 6pm and “Weekend” calls are calls made on Saturday and Sunday. 2.30 During the period under investigation, calls to some NTS number ranges (0845, its legacy equivalent 0345, and 0820) were classed as either “short” or “long”. “Short” calls were defined during the period under investigation as calls shorter than the call duration corresponding to BT’s minimum call charge; “long” calls as calls equal to or longer than the call duration corresponding to the minimum call charge.17 Transit 2.31 Payment arrangements for transit depend on number range. Where BT acts as transit provider, the OCP pays for transit for calls to number ranges starting 0844 and 0871, whereas the TCP pays for transit for calls to number ranges starting 0845 and 0870, as shown in Figure 2 below. 16 http://www.btwholesale.com/pages/static/service_and_support/service_support_hub/online_pricing_ hub/cpl_hub/cpl_pricing_hub/number_translation_services.html. 17 As of 1 October 2006, the minimum call charge was replaced with a “call set-up fee” of 3p. As a result the concept of “short” and “long” NTS calls no longer applies. 17 Figure 2: NTS transit arrangements D = BT’s discounted retail price C = BT’s retention D-C P OCP D-C BT T Originator pays transit (for 0844/0871) TCP NTS SP T Terminator pays transit (for 0845/0870/PRS) P = OCP’s retail price T = BT’s transit charge 2.32 BT’s transit charges are, in some cases, set by regulation. Ofcom has determined that BT has SMP in the market for single transit (see paragraphs A4.47-A4.49 below).18 As a remedy, BT’s charges for single transit are subject to a charge control. 2.33 Ofcom has determined that no communications provider has SMP in the market for inter-tandem transit (see paragraphs A4.50-A4.53 below). Communications providers are free to negotiate charges for inter-tandem transit. However, the incentives to do so are limited, as explained in paragraph 5.58 et seq below. NTS regulatory policy 2.34 Since its inception, Ofcom has conducted a major review of NTS regulatory policy and implemented a number of changes to the NTS regulatory regime. 2.35 On 22 October 2004 Ofcom published the consultation document, Number translation services: options for the future, which set out a number of options for future regulation of NTS (“the NTS consultation”).19 On 28 September 2005 Ofcom published a further consultation document, Number translation services: a way forward which proposed a number of detailed changes to the NTS regime.20 2.36 On 19 April 2006 Ofcom published the statement Number translation services: a way forward (“the NTS statement”) setting out a number of changes to the NTS regulatory regime, as follows: • 18 retail prices for calls to 0870 numbers will be linked to prices for geographic calls, so that OCPs including BT must charge their customers no more for calls to 0870 numbers than for national calls to geographic numbers. OCPs who wish to charge higher rates for 0870 calls will be required to make a free preannouncement at the beginning of the call, telling the caller how much the call will cost; Review of BT’s network charge controls, statement of 18 August 2005 (see footnote 14). Number translation services: options for the future, 22 October 2004, published at: http://www.ofcom.org.uk/consult/condocs/ntsoptions/#content. 20 Number translation services: a way forward, 28 September 2005, published at: http://www.ofcom.org.uk/consult/condocs/nts_forward/nts_way_forward.pdf. 19 18 • the NTS call origination condition (see paragraph 2.17) will no longer apply to calls to 0870 numbers; • the PRS regulatory framework will be extended to 0871 calls and to all adult PRS services, regardless of price; and • communications providers will be required to give their customers clearer information about the cost of calling NTS (and PRS) numbers.21 2.37 On the same date, Ofcom published a separate statement entitled Providing citizens and consumers with improved information about Number Translation Services and Premium Rate Services, which notified modifications to General Condition 14 requiring communications providers to give their customers clearer information about the cost of calling NTS and PRS numbers, in line with the last of the changes set out in the previous paragraph.22 These new requirements came into effect on 19 August 2006. 2.38 The NTS statement explained that the changes to the regulation of the 0870 range would be implemented 18 months after the publication of Ofcom’s separate statement on numbering policy, which was published on 27 July 2006 and confirmed, among other things, a number of policy decisions relating to the NTNP that were necessary to give effect to the changes to the 0870 regulatory framework set out in the NTS statement.23 2.39 When Ofcom introduced a requirement for pricing pre-announcements on 070 (Personal Numbers) in an unrelated policy initiative, the pre-announcements interfered with a small number of social alarm services (typically used by the elderly and infirm to summon assistance).24 Ofcom also became aware there were a large number of security alarm services using 0870 numbers that might also be susceptible to interference from pre-announcements, causing service disruption. 2.40 As a result of these discoveries, Ofcom took extra time to reconsider whether it should require the use of pricing announcements. Instead of pre-announcements, it has, at the time of publication, proposed that communications providers wishing to charge more for 0870 calls than for geographic calls would be subject to new requirements to display their 0870 charges prominently on price lists and advertising and promotional materials. The details of Ofcom’s revised proposals are contained in the consultation document Changes to 0870, published by Ofcom on 2 May 2008.25 NCCN 500 2.41 On 1 April 2004 BT issued NCCN 500, notifying a number of increases to its charges to third parties (i.e. other OCPs) for the termination of calls to NTS numbers with effect from 1 May 2004. 21 Number translation services: a way forward, 26 April 2006, published at: http://www.ofcom.org.uk/consult/condocs/nts_forward/statement/statement.pdf. 22 Providing citizens and consumers with improved information about Number Translation Services and Premium Rate Services, 19 April 2006, published at: http://www.ofcom.org.uk/consult/condocs/nts_info/statement/statement. 23 Telephone Numbering: Safeguarding the future of numbers, 27 July 2006, published at: www.ofcom.org.uk/consult/condocs/numberingreview/statement/. 24 See the Ofcom statement Removal of the requirement for pre-call announcements on 070 numbers at http://www.ofcom.org.uk/consult/condocs/numbering03/070precall/. 25 Changes to 0870: Changes to 0870 calls and modifications to the supporting regulations, 2 May 2008, published at: http://www.ofcom.org.uk/consult/condocs/0870calls/0870condoc.pdf. 19 2.42 The scale of the price rises notified in NCCN 500 differed by number range, by call duration and by time of day. Price rises ranged between zero (for daytime calls to 0820 numbers) and 37.8% (for long duration evening calls to 0845 numbers). The most significant rises applied to long duration calls to 0845 numbers. 2.43 NCCN 500 is reproduced in Annex 1.26 2.44 NCCN 500 notified price rises for the termination of calls originated by OCPs. Prior to NCCN 500, BT had set its charges for NTS call termination at the same level as other TCPs’ charges for NTS call termination – which other TCPs had set with reference to BT’s regulated origination and transit charges. NCCN 651 2.45 On 2 November 2005 BT issued NCCN 651, notifying a number of changes to its charges to third parties (i.e. other OCPs) for the termination of calls to NTS numbers with effect from 1 January 2006. 2.46 The effect of NCCN 651 was to bring BT’s charges for NTS call termination back into line with those of other TCPs, as they had been prior to 1 May 2004. This is shown in Figure 3 below. Figure 3: The effect of NCCN 500 and NCCN 651 Termination charge (ppm) 2.0 31 December 2005 1 May 2004 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 24/07/1998 06/12/1999 19/04/2001 01/09/2002 0845: terminated by other TCP 14/01/2004 28/05/2005 10/10/2006 0845: BT-terminated Source: NCCN 500, NCCN 651, NTS Calculator 2.47 26 [].27 Ofcom has calculated NCCN 500 price rises with reference to BT’s NTS call termination charges as at 1 April 2004 for OCP–BT calls using the NTS Calculator. Ofcom notes that BT has relied on NCCN 500 itself (as reproduced at Annex 1 below) when deriving estimated price increases, which suggests a considerably higher price increase than that implied by the comparison of the charges notified in NCCN 500 against NTS Calculator generated prices for the period immediately preceding the introduction of NCCN 500. Ofcom does not consider that the charges listed as “current” in NCCN 500 are a true reflection of the charges paid by OCPs to BT immediately prior to the introduction of NCCN 500. Prior to NCCN 500, Ofcom understands that both OCPs and BT used the NTS Calculator to determine the level of charges payable for NTS call termination. 20 2.48 The price increases that are the subject of this investigation were therefore in effect from 1 May 2004 to 31 December 2005 – a period of 20 months. The first complaint 2.49 On 4 May 2004 C&W submitted to Ofcom a complaint about NCCN 500 (“the first complaint”). 2.50 In the first complaint, C&W argued that the price increases notified in NCCN 500 constituted an abuse of BT’s dominant position in the market for NTS call termination, contrary to the Chapter II prohibition, in that they would: • impose a margin squeeze on C&W; • discriminate in favour of BT and against C&W on price; • increase C&W’s costs of providing a competing service or force it to provide an inferior service to that of BT; and • increase BT’s market power in the NTS call origination and NTS call termination markets.28 2.51 Ofcom considered the first complaint and reached the view that the issues raised therein would be addressed by Ofcom’s ongoing review of the NTS regulatory regime (which is described at paragraph 2.34 et seq above). Ofcom therefore decided to concentrate its resources on completing its review of the NTS regulatory regime, rather than diverting resources to running a parallel investigation. Ofcom wrote to C&W on 25 May 2004 informing C&W of this decision. 2.52 Ofcom wrote to C&W again on 7 July 2004 indicating that it would keep the matters raised in the first complaint under review and that it would not be precluded at some point in the future from considering a possible breach of the Act or Article 82 from the date that the alleged breach had occurred. The NTS call termination market review 2.53 As discussed in the previous section, Ofcom did not open an investigation into the first complaint, as it considered that the issues raised by C&W in that submission would be addressed by its ongoing review of the NTS regulatory regime. 2.54 NCCN 500 prompted Ofcom to include, in its ongoing review of the NTS regulatory regime, a detailed consideration of NTS call termination. 2.55 On 22 October 2004 (the same date that it published the first NTS consultation), Ofcom published the consultation document NTS call termination market review (“the NTS call termination market review consultation”).29 2.56 In the NTS call termination market review consultation, Ofcom proposed to identify a market for NTS call termination in the UK and set out its view that BT has SMP in that market. Ofcom considered a number of options for action, and indicated that its 27 []. C&W submission of 4 May 2004, paragraph 1.5. 29 NTS call termination market review, 22 October 2004, published at: http://www.ofcom.org.uk/consult/condocs/ntsctmr/. 28 21 preferred option was to impose two new SMP conditions on BT, requiring it to provide network access, and not to exercise undue discrimination in relation to the provision of network access, in the NTS call termination market.30 2.57 The NTS call termination market review consultation closed on 7 January 2005. Respondents to the consultation included BT and UKCTA, the trade association of competitive fixed line carriers in the UK, which represented a number of OCPs and TCPs with interests in NTS call termination. Non-confidential responses to the consultation were published on Ofcom’s website.31 2.58 As discussed in Section 3 below, on 7 April 2005 Ofcom opened an investigation into a further complaint from C&W on 15 March 2005 relating to BT’s charges for NTS call termination. In order to avoid duplication of analysis, Ofcom decided not to proceed separately with the market review while the investigation was ongoing, and has not therefore concluded on the proposals set out in the market review consultation. Some of the respondents to the draft decision (see paragraphs 3.28 et seq below) commented that Ofcom should not have suspended the market review pending the completion of its investigation under the Act. However, Ofcom considered that the market definition exercise and, to some extent, the dominance analysis undertaken as part of this investigation would in effect have been duplicated had Ofcom progressed with the market review and the investigation in parallel. In Ofcom’s view, this would have represented an inefficient and inappropriate use not only of Ofcom’s resources, but the resources of the parties to the investigation. 30 31 NTS call termination market review consultation, Section 5. http://www.ofcom.org.uk/consult/condocs/ntsctmr/resntcctr/. 22 Section 3 3 Ofcom’s investigation The complaint 3.1 On 15 March 2005 C&W submitted its second complaint to Ofcom about NCCN 500 (“the complaint”). 3.2 In the complaint, C&W alleged that NCCN 500 represented an abuse of BT’s dominant position in a number of markets, including the market for NTS call termination. 3.3 C&W submitted that the relevant markets for the purposes of the complaint were the markets for: “NTS call origination, NTS call termination, NTS call transit and NTS service hosting.”32 3.4 C&W submitted that BT holds a dominant position in the market for call origination on fixed public narrowband networks, noting that Ofcom had found BT to have SMP in that market (see paragraph 2.17 above). 3.5 C&W submitted that; “Call origination to NTS number ranges forms a subset of the wider call origination market and can be distinguished from, and is not substitutable with, geographic call origination.”33 3.6 C&W submitted in addition that BT has market power in the market for NTS call termination to services hosted on the BT network. 3.7 C&W submitted that it agreed with Ofcom’s proposals in relation to the NTS call termination market, as set out in the market review consultation.34 3.8 However, C&W submitted that even if BT were not dominant in the NTS call termination market, BT’s conduct in issuing NCCN 500 would equally constitute an abuse of BT’s dominant position in the market for wholesale call origination, since: “It is only BT’s control over and overwhelming market share in, call origination that renders BT’s conduct in issuing NCCN 500 profitable.”35 3.9 C&W submitted that NCCN 500 constituted an abuse of BT’s dominant position in both the market for NTS call termination and the market for wholesale call origination.36 3.10 C&W alleged that the price increases notified in NCCN 500: 32 C&W submission to Ofcom of 15 March 2005, paragraph 5.5. C&W submission to Ofcom of 15 March 2005, paragraph 5.10. 34 C&W submission to Ofcom of 15 March 2005, paragraph 5.13. 35 C&W submission to Ofcom of 15 March 2005, paragraph 5.14. 36 C&W submission to Ofcom of 15 March 2005, paragraph 5.16. 33 23 3.11 • imposed a margin squeeze on C&W; • discriminated in favour of BT and against C&W on price; • were excessive; • increased C&W’s costs of providing a competing service or forced it to provide an inferior service to that of BT; • increased BT's market power in the NTS call origination and NTS call termination markets; and • formed part of a concerted strategy to dilute competition.37 C&W requested: • that Ofcom adopt a decision pursuant to section 31 of the 1998 Act that the price increases notified in NCCN 500 constitute an abuse by BT of its dominant position; • that Ofcom issue directions to BT under section 33 of the 1998 Act requiring it to take such steps as are necessary to bring those abuses to an end, in particular: a) that NCCN 500 be withdrawn; b) that BT be required to reimburse C&W in respect of any payments it had made to BT based on the higher charges notified in NCCN 500; and c) that BT be required to charge other OCPs for calls terminated on its network only in accordance with the NTS Formula;38 and d) that BT be required to pay a penalty under section 36 of the Act.39 3.12 3.13 37 C&W indicated that its complaint was supported by: • Centrica; • ntl; and • Tiscali.40 Thus also contacted Ofcom in March 2005 and asked to be considered as a stakeholder in this case. C&W submission to Ofcom of 15 March 2005, paragraph 5.24. I.e. the methodology of calculating charges using the NTS Calculator (see paragraphs 2.27 et seq above). 39 C&W submission to Ofcom of 15 March 2005, paragraph 12.2. 40 C&W submission to Ofcom of 15 March 2005, paragraph 13.1. Centrica has since been acquired by The Carphone Warehouse Group plc (“CPW”). ntl is now part of Virgin Media. The complaint, which was made by Energis prior to C&W’s acquisition of Energis (see footnote 1), also listed C&W as supporting the complaint. 38 24 The investigation 3.14 On 8 April 2005 Ofcom opened an investigation into the complaint under the Act and Article 82 and published an entry on its Competition Bulletin setting out the details of the investigation.41 Period of investigation 3.15 This investigation relates to the period of the alleged abuse i.e. the period of time that NCCN 500 was in place, which was 1 May 2004 to 31 December 2005, i.e. 20 months. Evidence gathered 3.16 In order to investigate C&W’s allegations, Ofcom gathered evidence from a number of parties, as set out in the following paragraphs. Evidence gathered from BT 3.17 3.18 Ofcom requested information from BT in connection with its investigation in several ways: • formal Notices under section 26 of the Act; • follow-up questions to BT’s responses to Notices under section 26 of the Act; and • informal requests for information. Before reaching its provisional decision, Ofcom sent BT 14 section 26 Notices, as set out in Table 2 below: 41 www.ofcom.org.uk/bulletins/comp_bull_index/comp_bull_ocases/open_all/cw_823/. On 7 April 2004 Ofcom wrote to the Office of Fair Trading (“OFT”), informing the OFT that it intended to exercise prescribed functions (as defined under Regulation 2(j) of the Competition Act 1998 (Concurrency) Regulations 2004) in relation to the complaint and asked the OFT to confirm that it agreed that Ofcom was the competent person to exercise prescribed functions in relation to the complaint. The OFT replied on the same date confirming that Ofcom was the competent person best placed to consider the complaint. 25 Table 2: Details of section 26 Notices sent to BT during the course of the investigation Date of Notice Date of BT’s response(s) Purpose of Notice 22 April 2005 13 May 2005 Provision of documents related to BT’s decision to impose the price increases notified in NCCN 500. 19 May 2005 27 May 2005 6 June 2005 Details of volumes of NTS calls terminated by BT and revenues received by BT in respect of those calls. 11 July 2005 Description of BT’s NTS business. 12 July 2005 Details of BT’s top 10 NTS service provider customers. 28 June 2005 14 July 2005 21 July 2005 12 July 2005 14 July 2005 Copies of BT’s management accounts and business cases relevant to its NTS call termination business. Details of volumes of NTS calls originated by BT. 21 July 2005 9 September 2005 13 September 2005 24 August 2005 26 August 2005 Clarification of relevant transfer charges in BT’s management accounts. 25 August 2005 9 September 2005 Details of volumes of NTS calls originated by BT. 11 October 2005 11 October 2005 17 October 2005 Financial Information derived from BT’s regulatory costing systems (see Section 6) required for Ofcom’s investigation of C&W’s allegations of excessive pricing and margin squeeze. Information on volumes of calls transited by BT. 21 October 2005 4 November 2005 Financial information derived from BT’s regulatory costing systems required for Ofcom’s investigation of C&W’s allegations of excessive pricing. 10 November 2005 18 November 2005 Financial information derived from BT’s regulatory costing systems information required for Ofcom’s investigation of C&W’s allegations of excessive pricing. 14 November 2005 2 December 2005 Provision of documents related to BT’s decision to impose the price changes notified in NCCN 651. 14 December 2005 26 Date of Notice Date of BT’s response(s) Purpose of Notice 14 December 2005 19 December 2005 Financial information derived from BT’s regulatory costing systems required for Ofcom’s investigation of C&W’s allegations of excessive pricing. 24 January 2006 3 February 2006 Financial information derived from BT’s regulatory costing systems required for Ofcom’s investigation of C&W’s allegations of excessive pricing. 6 February 2006 6 March 2006 10 March 2006 24 March 2006 24 April 2006 28 April 2006 3 May 2006 Financial information derived from BT’s regulatory costing systems required for Ofcom’s investigation of C&W’s allegations of excessive pricing. Provision of documents related to BT’s contracts for the provision of NTS hosting. 5 May 2006 3.19 Ofcom requested further information from BT by way of follow-up questions to BT’s responses to Notices under section 26 of the Competition Act 1998 on a number of occasions, as set out in Table 3 below: Table 3: Follow-up information requested from BT during the course of the investigation Date of Ofcom’s request Details of Ofcom’s request Date of BT’s response 23 May 2005 Letter from [] (Ofcom) to [] (BT) requesting further information on BT’s response of 13 May 2005 to Ofcom’s Notice of 22 April 2005. 27 May 2005 13 December 2005 Letter from [] (Ofcom) to [] (BT) requesting further information on BT’s response of 2 December 2005 to Ofcom’s Notice of 14 November 2005. 14 December 2005 20 December 2005 Letter from [] (Ofcom) to [] (BT) with follow-up questions on BT’s response of 13 October to Ofcom’s Notice of 11 October, BT’s response of 17 October to Ofcom’s Notice of 11 October; BT’s response of 4 November to Ofcom’s Notice of 21 October and BT’s response of 18 November to Ofcom’s Notice of 10 November. 30 December 2005 3.20 13 January 2006 20 January 2006 Ofcom requested information from BT informally on a number of occasions, as set out in Table 4 below: 27 Table 4: Other information requested during the course of the investigation Date of Ofcom’s request Details of Ofcom’s request Date of BT’s response 27 January 2006 Letter from [] (Ofcom) to [] (BT) inviting BT’s comments on Ofcom’s financial analysis of BT’s voice NTS call termination/hosting business. 16 February 2006 12 May 2006 Letter from [] (Ofcom) to [] (BT) asking BT to clarify how it accounts within its regulatory costing systems for the effect of non-geographic number portability. 23 May 2006 28 July 2006 8 September 2006 3.21 3.22 BT made a number of other submissions that Ofcom took into account in its investigation, as follows: • BT’s response of 7 January 2005 to the market review consultation; • letter from [] (BT) to [] (Ofcom) of 1 April 2005 setting out BT’s comments on the complaint; and • letter from [] (BT) to [] (Ofcom) of 20 April 2006 setting out BT’s comments on the continuing investigation, including two Annexes providing details of other OCPs’ retention on calls to NTS numbers, differences between the charges notified in NCCN 500 and NTS Calculator rates, and BT’s analysis of retail prices for calls to NTS numbers. Ofcom and BT discussed various aspects of Ofcom’s investigation, either face to face or by telephone, on the following dates: • 4 July 2005; • 18 August 2005; • 2 September 2005; • 29 September 2005; • 11 October 2005; • 19 October 2005; • 2 December 2005; • 7 April 2006; and • 16 June 2006. Evidence gathered from C&W 3.23 28 In addition to C&W’s submission of 15 March, Ofcom considered evidence provided by C&W in various different ways on a number of occasions. 3.24 3.25 3.26 Ofcom sent C&W five section 26 Notices over the course of the investigation: • 19 May 2005 (to Energis and C&W) requesting calls volumes for the purposes of enabling Ofcom to calculate market shares; • 28 June 2005 (to Energis and C&W) requesting details of their respective businesses and the impact of NCCN 500; and • 23 March 2006 (to C&W only) requesting clarification of the impact of nongeographic number portability (“NGNP”) on the investigation (see following paragraph). C&W made a number of other submissions that Ofcom took into account in its investigation: • an email of 22 April 2005 setting out in writing a number of comments made by C&W in a meeting with Ofcom 6 April 2005 relating to issues raised in the complaint; • a letter of 18 August 2005 setting out in writing a number of comments made by C&W in a meeting with Ofcom 11 August 2005 relating to issues raised in the complaint; • a letter of 6 September 2005 setting out C&W’s comments on the appropriate approach to Ofcom’s analysis of the alleged margin squeeze; • an email of 9 December 2005 attaching C&W’s comments on the impact of NCCN 651 on the investigation; • a letter of 1 March 2006 about the potential impact on the investigation of issues relating to NGNP; • a further letter of 10 March 2006 about the potential impact on the investigation of issues relating to NGNP; • a letter of 21 April 2006 clarifying C&W’s position with regard to issues relating to NGNP; • a letter of 12 May 2006 containing a review of previous submissions made by C&W in light of its letters to Ofcom of 1 March 2006 and 10 March 2006; and • a letter of 23 June 2006 commenting on the relevance of issues relating to NGNP to Ofcom’s decision. Ofcom and C&W met to discuss various aspects of Ofcom’s investigation on the following dates: • 6 April 2005; • 11 May 2005; • 8 June 2005; • 11 August 2005; 29 • 29 November 2005; • 15 March 2006; • 12 April 2006; • 7 June 2006; and • 16 June 2006. Evidence gathered from third parties 3.27 Ofcom gathered evidence from third parties as follows: • section 26 Notices of 19 May 2005 to COLT Telecom Ltd, Centrica Telecommunications, Easynet Group Plc, Gamma Telecom Holdings Ltd (“Gamma”), Global Crossing (UK) Telecommunications Ltd, Kingston Communications Ltd, MCI Worldcom Ltd, ntl Group Limited (“ntl”), Telewest Communications Plc (“Telewest”), Thus Plc (“Thus”), Tiscali UK Limited (“Tiscali”) and Your Communications Ltd and section 26 Notice of 29 June 2005 to Opal Telecom Limited, requesting volume information for the purpose of enabling Ofcom to calculate market shares; • section 26 Notices of 28 June 2005 to Centrica, C&W, Energis (see footnote 1); Gamma; Hutchison 3G UK Ltd (“3”); ntl; O2 Plc (“O2”); Orange Limited (“Orange”); T-Mobile Limited (“T-Mobile”); Telewest; Thus; Tiscali; Vodafone Ltd (“Vodafone”). • section 26 Notice of 16 May 2006 to FlexTel Ltd and section 26 Notice of 18 May 2006 to IV Response Ltd. The draft decision 3.28 On 23 July 2007 Ofcom provided a draft decision to BT, C&W and four other parties supporting C&W’s complaint (Carphone Warehouse (“CPW”), Thus, Tiscali, and Virgin Media, together “the supporting parties”) for comment (the “draft decision”) and invited written responses on its proposed findings by 10 September 2007. Following requests for extensions from a number of the recipients, five responses were received in the weeks commencing 8 October 2007 and 15 October 2007. 3.29 Responses were received from all the companies listed in paragraph 3.28 above apart from Tiscali. These comments are set out in the relevant sections throughout this document. 3.30 C&W, CPW, Thus and Virgin Media expressed concerns on the length of time taken for this investigation. 3.31 As set out in this decision, Ofcom has undertaken an extensive amount of work and evidence gathering on the relevant market and on the allegedly abusive conduct in order to come to this decision. This has included additional work in response to new 30 information provided by C&W during the investigation and in response to comments on the draft decision (see below).42 3.32 As set out above, Ofcom has sent extensive information requests, in particular to BT, to obtain the evidence necessary to carry out its analysis. In some cases, particularly in relation to BT’s costs, information has been unavailable, or not available in a readily accessible or sufficiently disaggregated form. In these instances Ofcom has engaged in an iterative process with BT to obtain the best information that it can on which to base its decision. Evidence gathered following the draft decision 3.33 Ofcom met C&W, CPW, Thus and Virgin Media on 11 December 2007. At the meeting, C&W and the supporting parties commented on a number of issues that it believed warranted further analysis by Ofcom. 3.34 In a letter to Ofcom of 14 December 2007, Thus (on behalf of itself and C&W, CPW and Virgin Media) stated that it wished to propose an alternative to Ofcom’s methodology for the analysis of C&W’s excessive pricing allegation, and requested access to unredacted versions of Tables 18, 19, 20 and 22 of the draft decision (which are also Tables 18, 19, 20 and 22 of this document) to enable them to carry out this work. Thus also asked Ofcom for access to its case file, in order to understand any analysis conducted by Ofcom that might not have been detailed in the draft decision.43 3.35 In order to inform its consideration of some of the comments made by C&W and the supporting parties, Ofcom sent BT a further section 26 Notice on 21 January 2008 requesting details of call traffic routed via the 1280 prefix (which enables CPS (carrier pre-selection: see Glossary) customers to select BT for some of their calls on a callby-call basis) and of BT’s marketing activity over the period under investigation. 3.36 On 28 January 2008 Ofcom provided C&W, CPW, Thus and Virgin Media with copies of a file containing a number of non-confidential documents on which it had based its draft decision (the “non-confidential documents file”). All the documents in the nonconfidential file were documents provided by BT. The file included documents that were non-confidential in full, and redacted versions of documents containing confidential information. 3.37 On 30 January 2008, Ofcom met with C&W, CPW, Thus and Virgin Media to discuss the case. Thus expressed the view, as stated in a follow up email, that it felt: “unable to provide a robust contribution due to the amount of redacted information both in the consultation itself and the recently delivered case file [i.e. the non-confidential documents file].”44 3.38 In order to keep C&W and the supporting parties closely associated with the proceedings, Ofcom considered whether it could provide further information to C&W and the supporting parties, in particular: 42 C&W provided additional information in March 2006 relating to numbers that were ported from BT to C&W, []. Work related to this information resulted in a delay to Ofcom’s ongoing work in order to investigate the new information. 43 Letter from [] (Thus) to [] (Ofcom),14 December 2007. 44 email from [] (Thus) to [] (Ofcom), 31 January 2008. 31 • whether it could share with C&W and the supporting parties BT’s reasons for asserting confidentiality over particular information; • whether it could tell C&W and the supporting parties which inputs in its SAC model were derived from BT’s regulatory accounts, which were derived from confidential information provided by BT, and which were from other sources; • whether it could give C&W and the supporting parties access to a version of its SAC model, redacted to remove confidential information; and • whether it could give third party economic advisors employed by C&W and the supporting parties access to confidential information (including a full version of the SAC model) subject to a suitable non-disclosure agreement. 3.39 Ofcom took into account its obligations under Part 9 Enterprise Act in assessing what it was necessary to disclose to the complainants in order to keep the complainants closely associated to assist in Ofcom carrying out its functions.45 Ofcom also had regard to procedures set out in The Competition Act 1998 (Office of Fair Trading’s Rules) Order 2004 (SI 2004/2751).46 Ofcom asked BT to provide further explanations of its claims to confidentiality. BT responded to Ofcom’s requests by consenting to certain information being disclosed and providing explanations of why other information remained confidential. Ofcom carefully considered BT’s claims to confidentiality to apply the considerations set out in section 244 Enterprise Act. Using this test, Ofcom considered that disclosure of BT’s confidential information might significantly harm BT’s legitimate business interests but did not consider that the disclosure of this confidential information to C&W and the supporting parties was necessary for the purpose of facilitating the exercise of Ofcom’s functions under the Act. 3.40 In addition Ofcom considered whether it was necessary to facilitate its functions to disclose a full version of the model, including BT confidential information, to third party economic advisors retained by C&W and the supporting parties, under a nondisclosure agreement.47 Ofcom concluded, having regard to its obligations under Part 9 Enterprise Act, that disclosure to C&W’s third party economic advisors was not necessary.48 3.41 On 4 March 2008 Ofcom therefore provided C&W with a version of its SAC model, redacted to remove all confidential information supplied by BT and third parties (see 45 Ofcom notes the OFT’s guideline Involving third parties in Competition Act investigations (OFT 451), 2006, published at: http://www.oft.gov.uk/advice_and_resources/publications/guidance/competition-act/oft_451. 46 The Competition Act 1998 (Office of Fair Trading’s Rules) Order 2004 (SI 2004/2751), published at: http://www.opsi.gov.uk/si/si2004/20042751.htm 47 Ofcom notes that complainants may be provided with confidential information in the context of appeals of decisions made under the Act to the CAT (usually within confidentiality rings which are established by order of the CAT). However, Ofcom also notes that the CAT is not bound by Part 9 Enterprise Act (see section 237 (5)) and has wide powers to give directions to parties involved in proceeding before it, including with regard to disclosure of documents (rule 19 The Competition Appeal Tribunal’s Rules 2003 S.I. 1372/2003). 48 Ofcom notes Article 8(1) of Commission Regulation 773/2004 relating to the conduct of proceedings by the Commission pursuant to Article 81 and 82 of the EC Treaty where in relation to access to information for complainants, the Commission states: “the complainant may however not have access to business secrets and other information belonging to other parties involved in the proceedings”. Ofcom also notes the European Commission’s Access to File Notice (2005/C325/07), paragraphs 30 to 32 and 39 to 43 and 49. 32 paragraph 3.27 above, final bullet). On the same date Ofcom wrote to C&W setting out its decision on the other requests made by C&W and the supporting parties, as set out in paragraph 3.38 above. The letter included an annex explaining in more detail the derivation of the non-confidential inputs to the SAC model. Ofcom asked C&W to submit any further comments on Ofcom’s excessive pricing methodology, and on the contents of the non-confidential documents file, by 27 March 2008. 3.42 On 4 March 2008 Ofcom provided BT with a version of its SAC model, redacted to remove all confidential information supplied by third parties (see paragraph 3.27 above, final bullet) and asked BT to submit any further comments on Ofcom’s excessive pricing methodology by 27 March 2008. BT did not submit any further comments on Ofcom’s excessive pricing methodology. 3.43 C&W wrote to Ofcom on 27 March 2008 stating that it did not intend to submit further comments on the redacted SAC model or on the contents of the non-confidential documents file. 33 Section 4 4 The relevant market Summary 4.1 The Chapter II Prohibition provides that any conduct on the part of one or more undertakings which amounts to the abuse of a dominant position in a market within the UK is prohibited if it may affect trade within the UK. 4.2 Article 82 prohibits the abuse of a dominant position if it affects trade between Member States. 4.3 For the purposes of the Chapter II prohibition and Article 82, dominance is assessed within a relevant market.49 This section therefore considers the relevant market for the purposes of assessing whether BT holds a dominant position (BT’s market position is considered at Section 5) and whether it has abused that position (BT’s conduct is considered at Section 6). 4.4 For the reasons set out below Ofcom concludes that in the context of its assessment of C&W’s allegations the relevant market is a single market for the termination/hosting of NTS calls in the UK. 4.5 In identifying the relevant market in this case Ofcom relied on some of the evidence and findings in recent relevant market reviews where appropriate, and where Ofcom was not aware of material changes to the findings in those reviews.50 Identifying markets 4.6 The relevant market normally has two dimensions: the relevant goods or services (the product market) and the geographic extent of the market (the geographic market).51 4.7 These dimensions may be broadly defined as follows: 49 • a relevant product market comprises all those products and/or services which are regarded as interchangeable by any reason of the products’ characteristics, prices and intended use.52 • the relevant geographic market is the area over which substitution takes place. See OFT, Market Definition (OFT 403), 2004, published at: http://www.oft.gov.uk/shared_oft/business_leaflets/ca98_guidelines/oft403.pdf and European Commission Notice on the definition of the relevant market for the purposes of Community competition law, OJ [1997] C372/5. 50 Ofcom notes that in Joined Cases T-125/97 etc Coca Cola v Commission [2000] ECR II 1733, paragraph 82, the Court of First Instance (“CFI”) stated that in the course of any decision applying Article 82 “the Commission must define the relevant market again and make a fresh analysis of the conditions of competition which will not necessarily be based on the same considerations as those underlying the previous finding of a dominant position.” See also paragraph 5.7 of OFT 403 and Aberdeen Journals Ltd v DGFT (No 1) [2002] CAT 4, paragraph 139. 51 See OFT 403, paragraph 2.15. 52 Commission Notice on market definition (see footnote 49), paragraph 7. 34 Identifying the relevant product market 4.8 In Aberdeen Journals, the Competition Appeal Tribunal (“CAT”) stated that, in assessing the relevant product market: “The key idea is that of competitive constraint: do the other products alleged to form part of the same market act as a competitive constraint on the conduct of the allegedly dominant firm?” 53 4.9 The European Commission (“the Commission”) has provided guidance on how it applies the concept of the relevant product market in practice in its Notice on market definition, which describes the sorts of information that may be used to define relevant product markets.54 These include product characteristics and intended use, evidence of substitution in the recent past, the views of customers and competitors, consumer preferences, barriers and costs associated with switching demand to potential substitutes and different categories of customers and price discrimination. Supply-side substitution may also be relevant to a definition of the relevant market, where its effects are sufficient to constrain pricing behaviour within the relevant timescale. 4.10 The “hypothetical monopolist test” is a useful tool for identifying close demand-side and supply-side substitutes.55 A product is considered to constitute a separate market if a hypothetical monopoly supplier could impose a small but significant, nontransitory relative price increase (in the range of 5% to 10%) above the competitive level without losing sales to such a degree as to make this unprofitable (known as the “SSNIP test”). If such a price rise would be unprofitable because of demand-side or supply-side substitution, then the market definition should be expanded to include the substitute products. 4.11 Demand-side substitution occurs if consumers would switch to other products in response to the price increase. Supply-side substitution occurs if suppliers of other products would begin to compete with the hypothetical monopolist in the supply of the product under investigation in response to the price increase. 4.12 The OFT defines supply-side substitution as follows: “If prices rise, undertakings that do not currently supply a product might be able to supply it at short notice and without incurring substantial sunk costs. This may prevent a hypothetical monopolist profitably sustaining prices 5 to 10 per cent above competitive levels. This form of substitution is carried out by suppliers and hence is known as supply-side substitution. Supply-side substitution can be thought of as a special case of entry – entry that occurs quickly (e.g. less than one year), effectively (e.g. on a scale large enough to affect prices), and without the need for substantial sunk investments. Supply-side substitution addresses the questions of whether, to what extent, and how quickly, undertakings 53 Aberdeen Journals Ltd v DGFT [2001] CAT 5, paragraph 97. Commission Notice on the definition of the relevant market for the purposes of Community competition law OJ [1997] C372/5, paragraphs 36-43. 55 Commission Notice on market definition (see footnote 49), paragraphs 15-19, OFT 403, paragraphs 2.8-2.9 and Aberdeen Journals (see footnote 53), paragraph 99. 54 35 would start supplying a market in response to a hypothetical monopolist attempting to sustain supra competitive prices.”56 4.13 Standard practice is first to consider whether demand-side substitution would be sufficient to constrain the ability of a hypothetical monopolist in the supply of the relevant good or service to sustain a SSNIP above the competitive level. Supply-side substitution possibilities are then assessed to consider whether they provide any additional constraints on the pricing behaviour of the hypothetical monopolist which have not been captured in the demand-side analysis. In this assessment, supplyside substitution is considered to be a low cost form of entry which can take place within a reasonable time frame (e.g. up to 12 months). 4.14 The key point is that, for supply-side substitution to be relevant, not only must suppliers be able, in theory, to enter the market quickly and at low cost by virtue of their existing position in the supply of other services or areas, but there must also be an additional competitive constraint arising from such entry into the supply of the service in question. Where supply side substitutes are not included within the relevant market, their impact will be taken into account in the assessment of market power as potential entry. 4.15 A third factor that may be an additional consideration is whether there exist common pricing constraints across customers, services or areas, such that they should be included within the same market. A common pricing constraint is generally used to define the market when there are indications that there are additional constraints on firms’ behaviour that override the absence of demand- or supply-side substitutability. 4.16 Such constraints can include, for example, transactions costs associated with differentiating prices in different areas, or consumers’ preferences for purchasing a bundle of services from a single supplier rather than on a service-by-service basis. Failure to consider the existence of a common pricing constraint could lead to unduly narrow markets being defined. 4.17 A fourth factor which may also be relevant is the homogeneity of competitive conditions. It may be appropriate to consider two products as forming a single market if competitive conditions in their supply are sufficiently homogeneous and sufficiently distinct from other products potentially in the market. Homogeneity of competitive conditions may be particularly relevant for geographic market definition. Identifying the relevant geographic market 4.18 The geographic boundary of the relevant market is generally defined using the same approach as the product market definition, that is, by using the hypothetical monopolist test to inform judgments as to demand- and supply-side substitutability, and, in the absence of demand- or supply-side substitutability, considering evidence as to the existence of a common pricing constraint. 4.19 For geographic markets, the hypothetical monopolist test asks whether a price increase in the narrowly defined area would encourage suppliers outside the area to begin to offer services to customers in the area and/or whether customers could switch to suppliers located outside the area. In certain circumstances, as discussed in paragraph 4.15 above, the existence of a common pricing constraint can be relevant to the consideration of relevant geographic market boundaries. 56 OFT 403, paragraphs 3.13 and 3.15. 36 4.20 In many communications markets, demand- and supply-side substitution possibilities between different geographic areas are limited. It may be appropriate to consider the area in which competitive conditions are homogeneous in order to determine the geographic extent of the market. The Commission’s guidelines on SMP state: “According to established case law, the relevant geographic market comprises an area in which…the conditions of competition are similar or sufficiently homogeneous and which can be distinguished from neighbouring areas in which the prevailing conditions of competition are appreciably different. The definition of the geographic market does not require the conditions of competition between traders or providers of services to be perfectly homogeneous. It is sufficient that they are similar or sufficiently homogeneous, and accordingly, only those areas in which the conditions of competition are ‘heterogeneous’ may not be considered to constitute a uniform market.”57 Markets relevant to this investigation Introduction 4.21 4.22 For the purposes of this investigation, Ofcom has considered: • the relevant market in which BT may hold a dominant position and in which the conduct in question has taken place, i.e. the market in which NTS call termination/hosting falls (see paragraphs 4.52-4.151 below); • any other upstream markets, in which BT’s position may influence the extent to which it faces competitive constraints in the relevant market (see paragraphs 4.42-4.51 below); and • downstream markets in which any effect of BT’s conduct in NTS call termination/hosting may be felt (see paragraphs 4.36-4.41 below). Ofcom will define the downstream market before considering the relevant market. This is because the demand for upstream services is a derived demand, i.e. the level of the demand for upstream inputs depends on the demand for downstream services. In this case, the demand for NTS call termination/hosting is derived from callers who want to buy services by calling NTS numbers. If the upstream input accounts for a sufficiently large proportion of the downstream price, the range of available substitutes at the downstream (retail) level will to some extent determine the likely range of substitutes for the upstream (wholesale) service.58 NTS markets 4.23 The following table lists NTS number ranges and summarises the retail and wholesale pricing arrangements applicable to each. 57 Commission guidelines on market analysis and the assessment of significant market power under the Community regulatory framework for electronic communications networks and services (2002/C 165/03), paragraph 56. 58 Where the costs incurred in the upstream market do not form a sufficiently large proportion of the downstream market price, distinct upstream markets may sometimes be defined on the basis that a SSNIP for a narrower service in the upstream market would not lead to sufficient substitution to the other upstream inputs. 37 Table 5: Wholesale payment arrangements for different NTS number ranges Range Retail price59 Wholesale payments 080 Generally free to caller 0844 Up to and including 5p per minute or per call for BT customers (the price charged by other OCPs may vary). 0845 BT’s standard local call charge for BT customers (the price charged by other OCPs may vary).61 0870 BT’s standard national call charge for BT customers (the price charged by other OCPs may vary).62 • NTS service provider makes payment to TCP • TCP makes payment to OCP • TCP-OCP payment is, in effect, set by regulation for BT-originated calls • TCP chooses a price point (for BT customers – see Table 1 above) • OCP makes a payment to TCP • TCP may make a revenue share payment to NTS service provider • for BT-originated calls, payment to TCP is the difference between the BT retail price and BT’s regulated retention60 • OCP makes payment to TCP • TCP may make a payment to NTS service provider • for BT-originated calls, payment to TCP is the difference between the BT retail price and BT’s regulated retention Same as 0845 0871 Up to and including 10p per minute or per call for BT customers (the price charged by other OCPs may vary). Same as 0844 09 PRS Same as 0844 4.24 4.25 As discussed at paragraph 2.6 et seq above, any given NTS call will involve a number of participants, including some or all of: • the retail end user (i.e. the caller); • an OCP (which may be BT); • a transit provider (which may be BT); • a TCP (which may be BT); and • an NTS service provider. There are therefore a number of possible transactions that may take place, i.e. between: • 59 retail end users and OCPs; See Table 1 for full details of call prices as specified in the NTNP. Strictly the “Deemed Retail Price”, which is the actual retail price less discounts. 61 The requirement to charge at BT’s standard local call rate applies only to BT and derives from the NTS call origination condition (see paragraph A9.27 of the NTS consultation). 62 The requirement to charge at BT’s national call rate applies only to BT and derives from the NTS call origination condition (see paragraph A9.27 of the NTS consultation). 60 38 4.26 • OCPs and transit providers; • transit providers and TCPs; • OCPs and TCPs; • TCPs and NTS service providers; and • end users and NTS service providers. Figure 4 below provides an illustration of the most important of these transactions for the purposes of Ofcom’s investigation. For simplicity, it does not describe the situation in which traffic is transited via a third party transit provider. It also reflects the more typical situation where an OCP makes a payment to the TCP who then shares revenue with the service provider.63 Figure 4: Transactions in an NTS call – predominant case 4.27 The transaction that takes place between OCPs and TCPs is of particular relevance to this investigation, since this was the transaction directly affected by NCCN 500. The following discussion refers to this service as NTS call termination, where the per-minute revenue paid to the TCP is the (gross) charge for termination (the TCP then retains the net termination charge which is the gross termination charge less any revenue share paid the service provider). Ofcom focuses on the predominant case (revenue is collected by the OCP), since this is the mechanism used in the termination of calls to the number ranges affected by NCCN 500.64 4.28 NTS hosting is the term used to describe the additional services purchased by NTS service providers that enable them to use NTS as a micro-payment mechanism by which they receive a share of call charges paid by callers (see paragraph 2.12). 63 Some TCPs share no revenue with NTS service providers and (in the past) NTS service providers have made top-up payments to TCPs. Note that for 0800 the flow of money is entirely in the other direction. 64 This is not the only payment model. For example, for 0800 numbers, payments are made by TCPs to OCPs (but not vice versa) so that an increase in the price of termination would mean a TCP starting to hand over smaller payments to OCPs. 39 4.29 TCPs sell NTS call termination to OCPs (and ultimately the caller) and NTS hosting to NTS service providers. They sell call termination both directly and to resellers which act as intermediaries to sell NTS call termination/hosting on to NTS service providers.65 The balance of the gross termination charge not passed to the NTS service provider or reseller (referred to as the net termination charge) is retained by the TCP. Both TCPs and resellers may provide value added services (such as call centre solutions) in addition to basic termination/hosting. 4.30 As shown in Figure 4 above there are three significant payments in the chain of transactions that makes up an NTS call: a) the retail price paid by callers to OCPs; b) the gross termination charge paid by OCPs to TCPs; and c) revenue share payments made by TCPs to NTS service providers. 4.31 In some cases, there is a positive revenue share payment from the TCP to the NTS service provider. This is typically the case on the relatively expensive 087 and 09 number ranges, and for large ISP (internet service provider) customers on 084. In other cases, the NTS service provider makes a payment to the TCP, that is, a negative revenue share. This has often been the case for voice calls terminated/hosted on 084 (for example the “other revenue” shown in Table 18 below, which includes payments from NTS service providers to BT). 4.32 C&W has alleged that NCCN 500 created a margin squeeze between the gross termination charge paid by OCPs to BT and the retail price paid by callers to OCPs for calls to NTS numbers. NCCN 500 notified changes to BT’s gross termination charges, BT’s revenue from which might have been used to fund the revenue share payments BT makes to NTS service providers. 4.33 The market in which BT sells retail calls (and in which C&W has alleged a margin squeeze has been created) is considered in the following section. 4.34 In considering the competitive pressures faced by BT and its competitors in NTS markets, it is important to also take into account BT’s position in the markets for call origination and transit, which are considered in paragraph 4.42 et seq below. 4.35 The market in which NTS call termination/hosting falls is defined in paragraph 4.52 et seq below. Downstream markets: retail NTS calls 4.36 The definition of the relevant retail market is discussed in detail in Annex 4. The discussion follows the standard approach to market definition in which the starting point is an analysis of the constraints imposed on prices by demand- and supplysubstitution. In addition, the existence of common pricing constraints across a number of potentially distinct products is taken into account. On this basis, Ofcom believes that it would be possible to define two distinct product markets for: • 65 NTS calls made by business users; and For example, during the course of the investigation [] and [] (both undertakings operating on the NTS hosting side of the market) presented Ofcom with anecdotal evidence that indicated resellers typically provide NTS call termination in addition to NTS hosting. [] and [] both provide NTS call termination to OCPs, with [] also providing NTS hosting. 40 • NTS calls made by residential users. 4.37 However, this definition is not necessarily the most relevant to this case. The relevant retail market is that market (or those markets) in which competition could be affected by BT’s conduct. Ofcom’s definition of the relevant market should therefore reflect the way in which competition actually works. As market definition is not an end in itself but a tool to identify and define the boundaries of competition between firms, it is also appropriate to take account of the nature of the conduct under investigation. 4.38 Downstream retail market definitions are not material to assessing two of the abuses alleged in this case: whether the charges notified in NCCN 500 charges were excessive or discriminatory and aimed to exclude competitors. This means that the key reason for defining the relevant retail market(s) is to assist with Ofcom’s assessment of C&W’s allegation that NCCN 500 imposed a margin squeeze on OCPs operating in retail markets. 4.39 Ofcom has therefore taken account of the fact that a number of products are supplied together as a bundle in the definition of the relevant retail market. As described in Annex 4, in other circumstances, for example in the context of an analysis of market power for the purposes of imposing ex ante regulation, other market definitions might be appropriate. Ofcom’s view is that, for the purposes of this case, the relevant downstream market may be regarded as no narrower than the smallest bundle of services in which retail NTS calls may be offered as a component, i.e. a bundle of the following retail products (with distinct markets for services sold to residential and business customers): • local calls; • calls to mobiles; and • NTS calls. 4.40 This bundle represents the additional services purchased as part of the CPS “all calls” option compared to the (combined) national and international calls only options (see paragraph 6.40et seq below). 4.41 Ofcom’s view is that it is not necessary formally to conclude on the relevant retail market for the purposes of this decision, as it is sufficient to conclude that the relevant market in which to assess the adequacy of the retail margin is no narrower than this. Ofcom’s analysis across the variety of combinations of calls and access that could form the relevant retail market is sufficient to allow it to reach the view that NCCN 500 did not create a margin squeeze either on OCPs in retail markets or in NTS hosting (see Section 6). Upstream wholesale markets 4.42 BT’s position in wholesale markets other than NTS call termination/hosting plays an important part in this investigation. In particular: • BT’s position of SMP in the market for wholesale call origination (see Annex 4) is a key factor in preventing TCPs other than BT from increasing the gross termination charges that they levy on BT when they terminate BT-originated traffic (see paragraph 5.46 et seq below); and 41 • BT’s role as a transit provider is, similarly, a key factor in preventing other TCPs from increasing their gross termination charges for traffic that is transited (but not originated or terminated) by BT (see paragraph 5.56 et seq below). 4.43 This combination of factors means that BT is uniquely in a position to increase its gross termination charges above the level that is charged by its competitors. In the NTS call termination/hosting market, non-BT TCPs are not in a position to successfully negotiate over their gross termination charges either to BT or to other OCPs (see paragraph 5.46 et seq below). 4.44 The economic markets in which these services fall, together with BT’s position within them, are discussed at Annex 4 below. 4.45 In summary, the wholesale markets relevant to this investigation are: • wholesale call origination in the UK excluding the Hull area, in which Ofcom has designated BT as having SMP; • single transit in the UK excluding the Hull area, in which Ofcom has designated BT as having SMP; and • inter-tandem transit in the UK excluding the Hull area, in which Ofcom has determined that BT no longer has SMP. 4.46 Ofcom analysed these markets in March 2003 and following a detailed market review and consultation, made SMP findings in these markets in August 2003 in accordance with the requirements of the legal framework of the EC Directives. The findings were implemented from October 2003.66 These findings remained in place until they were reviewed again in August 2005 following a detailed market review and consultation in relation to network charge controls.67 BT was found to have SMP in relation to call origination and single transit, and was made subject to network change controls for the subsequent four-year period from October 2005 to September 2009. In relation to inter-tandem transit, BT was found to no longer have SMP from October 2005. 4.47 In Ofcom’s August 2005 review of BT’s network charge controls, call origination and single transit were found to have stable market conditions. BT was found to have a share of approximately 80% of call origination and 100% of single transit and entry barriers were high. In respect of call origination, Ofcom concluded in its “forward look” that: “BT’s current SMP in the market is unlikely to be eroded until such time that other direct access networks expand their customer base and are able to compete in a significant manner with BT. New entry into this market is constrained by the high entry barriers in the form 66 Review of the fixed narrowband wholesale exchange line, call origination, conveyance and transit markets: Identification and analysis of markets, determination of market power and setting of SMP conditions: Explanatory Statement and Notification, 26 August 2003, published at: http://www.ofcom.org.uk/static/archive/oftel/publications/eu_directives/2003/fix_narrow_retail0803.pdf, paragraph S.12 and Chapters 4, 5 and 6; and Review of the fixed narrowband wholesale exchange line, call origination, conveyance and transit markets: A consultation document issued by the Director General of Telecommunications, 17 March 2003, paragraph S.6 and analysed at Chapters 5, 6 and 7, published at: http://www.ofcom.org.uk/static/archive/Oftel/publications/eu_directives/2003/eu_narrow/narrow0303.p df. 67 Review of BT’s network charge controls, statement of 18 August 2005 (see footnote 14). 42 of sunk costs. Therefore Ofcom believes that BT is likely to have SMP in wholesale call origination for the duration of the next NCCs.”68 4.48 In respect of single transit, Ofcom concluded that: “Ofcom is of the view that, for the foreseeable future, [other communications providers] will continue to depend on BT for single transit where direct interconnection is not economically viable. Therefore, Ofcom believes that BT is likely to have SMP in single transit for the duration of the next NCCs.”69 4.49 Ofcom believes therefore that it is not necessary to conduct a further review of BT’s position in these markets for the purposes of this investigation, for which the relevant time period ended in December 2005, only two months after the start of the new controls on BT’s call origination and single transit charges. 4.50 Ofcom notes the falling away of BT’s SMP in the inter-tandem transit market two months prior to the cessation of NCCN 500 (which ceased immediately after December 2005). However, if conditions in this market were to change such that BT once again had a dominant position, this would not affect Ofcom’s analysis in this case. 4.51 In addition, Ofcom’s July 2006 review of BT’s retail price controls confirmed BT’s market power in (certain) retail markets and the need for continuing regulation to address BT’s market power in wholesale network services. 70 The relevant market Introduction 4.52 As described in Table 5 above, given the link between retail prices for calls to NTS numbers and the price of geographic calls made by BT customers for 0845 and 0870 calls, NTS service providers select the retail price to be paid by BT’s retail customers for calls to NTS numbers, by selecting a particular number range (or, in the case of 0844, 0871, and 09 number ranges, a particular price point within a range). 4.53 NTS calls may be divided into four categories depending on the identity of the OCP and TCP. What all four call types have in common is that the gross termination charge levied by TCPs on OCPs is outside the control of the TCP (except where BT is the TCP), for the reasons set out in paragraphs 4.54-4.58 below.71 4.54 First, for calls from BT to another TCP, BT’s retail price is linked to the relevant geographic price, and other TCPs are unable to affect BT’s cost-based originator retention (the maximum charge is set by regulation – see paragraph 2.18 above). Further, BT’s market power in wholesale call origination means that it can resist attempts to force it to accept a lower retention, meaning that in effect the gross 68 Review of BT’s network charge controls, statement of 18 August 2005 (see footnote 14), paragraph A5.46. 69 Review of BT’s network charge controls, statement of 18 August 2005 (see footnote 14), paragraph A5.68. 70 Retail Price Controls: Explanatory Statement, 19 July 2006, published at: http://www.ofcom.org.uk/consult/condocs/retail/statement/. 71 As most calls between two communications providers other than BT are transited via BT, the case of direct connection between communications providers other than BT is not described here. 43 termination charge is selected by the NTS service provider from a limited menu at the time at which it obtains a number. 4.55 Second, for calls between communications providers other than BT, while other OCPs may offer higher (or lower) retail prices than BT, the fact that most such calls are delivered via BT as a transit provider means that other TCPs’ gross termination charges are set as though the call were originated, as well as transited, by BT, because BT pays the same rate for termination in either case. This arrangement minimises the complexity of interconnection billing arrangements by allowing BT to have a single set of transit charges and termination payments that do not vary according to the OCP or TCP involved in each call. 4.56 The exception to the predominant case where a TCP cannot influence the level of its gross termination charge is where the OCP is not BT, and BT is not involved in the transit of the call. BT (as an OCP) is subject to maximum retail price regulation (as set out in the NTNP) and BT is the only OCP found to have SMP in call origination. Accordingly, a non-BT OCP with no SMP in the call origination market is likely to have to accept a lower call origination retention (see paragraphs 2.22-2.26 above). In addition, transit providers other than BT are likely to be able to bill OCPs at different rates, thereby providing TCPs with more flexibility to vary their gross termination charge accordingly (see paragraphs 2.24 above and 5.61 below). However, Ofcom considers that calls made under this special case are not sufficient in number or share to materially affect Ofcom’s findings on market definition and dominance (see paragraphs 5.51 and 5.56 above). 4.57 Third, for BT-to-BT calls, the gross termination charge would be purely an internal transfer charge, not necessarily directly reflected in charges to customers, and therefore has no necessary commercial or economic significance (see Ofcom’s analysis of alleged discrimination in Section 6). 4.58 Finally, for calls from another OCP to BT, the retail price is unregulated and BT is able to set the termination charge subject only to any competitive constraints that may apply. No other OCP has been found to be dominant in retail calls markets or wholesale call origination in the UK (outside the Hull area).72 Analysis of NTS call termination/hosting as a two-sided market 4.59 In this section, Ofcom considers, first, whether NTS call termination and NTS hosting can be regarded as a “two-sided” market and secondly whether, if so, it is appropriate to analyse both sides of the market together, as a single whole, or as if they were two distinct markets. The two (candidate) sides are marked “side 1” and “side 2” in Figure 4 above. These are, respectively, NTS call termination and NTS hosting. Features of two-sided markets 4.60 72 Certain features are necessary for a market to be regarded as two-sided. One, in particular, is the notion of “getting two sides on board”, i.e. acting as an intermediary to create an indirect relationship between the two sides (for example advertisers and readers in the case of newspapers). Review of BT’s network charge controls, statement of 18 August 2005 (see footnote 14),and Retail price controls – explanatory statement, 19 July 2006, http://www.ofcom.org.uk/consult/condocs/retail/statement/. 44 4.61 For instance, as part of its market review into classified directory advertising services (“CDAS”), the Competition Commission noted that classified directory providers operate in a two-sided market in which success depends on their ability to attract both users and advertisers to their directory. The Competition Commission went on to argue that: “This interdependence of advertiser and user demand for directories gives rise to a network effect or ‘virtuous circle’; a directory provider that has built up high levels of usage and advertising is more attractive to new advertisers and users than a competitor with less usage and less advertising whose offer, in terms of price, quality and service, may otherwise be the same.”73 4.62 In a recent merger case involving equestrian magazines, the OFT identified the existence of a two-sided market using similar reasoning: “Equestrian magazines operate in a two-sided market, in which readers (who demand information of one kind or another) and advertisers (who demand space to advertise/promote their goods and/or services) are closely inter-linked. The willingness of advertisers to pay for advertising space in a magazine will depend on the number and type of readers. Readers' willingness to read a particular magazine may in certain cases depend on the amount and type of advertising and/or editorial content within that magazine.”74 4.63 Recently, some economists have tried to define ‘two-sidedness’ in a rigorous way. The definition proposed by Rochet and Tirole is based on the idea that, in a twosided market, the “platform” provider’s total volume of transactions depends on the price structure (that is the respective share of the total charge borne by each side), as well as the level of the combined price (that is the sum of the charges to each side).75 In single-sided markets, total revenue depends only on the latter. A number of two-sided markets feature prices on one side of the market that do not cover costs or are even negative. 4.64 Newspaper advertising is often regarded as an example of a two-sided market.76 Advertisers contribute to the recovery of the cost of the paper by paying for advertising space. The most extreme form of this exists where the paper is free of charge to end consumers (i.e. readers). The volume of total newspaper readership is likely to be sensitive to the share of payments made between readers and 73 Competition Commission, Classified Directory Advertising Services market investigation, 21 December 2006, published at http://www.competitioncommission.org.uk/rep_pub/reports/2006/521cdas.htm, paragraph 6.2. 74 OFT, Anticipated acquisition by IPC Media Ltd of Horse Deals Ltd: The OFT's decision on reference under section 33(1) given on 16 August 2006, 1 September 2006, published at: http://www.oft.gov.uk/shared_oft/mergers_ea02/2006/ipc.pdf. 75 Jean-Charles Rochet and Jean Tirole, Defining Two-Sided Markets, 2004, presented at UCL twosided markets conference, May 2006. 76 For instance, the OFT has stated: “Newspapers and magazines operate in so-called ‘two-sided markets’, i.e. markets in which there are two sets of customers, each of which benefits from the other type of customer buying the product concerned. In this case, each title competes against other titles to attract readers, on the one hand, and advertisers, on the other. Publishers receive revenue from both readers (through cover prices) and advertisers. Publishers take account of the interaction between these two customer groups when determining the pricing strategy for their titles, as readership (circulation) is an important factor in determining advertising rates.” See Newspaper and Magazine Distribution: Public consultation on the draft opinion of the Office of Fair Trading (OFT 851), 31 May 2006, published at: http://www.oft.gov.uk/shared_oft/reports/comp_policy/oft851.pdf. 45 advertisers. For example, if readers of a free paper were required to pay a cover price or subscription, and the amount paid by advertisers fell by an equivalent amount, overall circulation of the newspaper might fall. 4.65 The effect on total sales arising from a change in the structure of payments is a key feature of two-sided markets. NTS and the characteristics of a two-sided market 4.66 In the case of NTS calls, TCPs always provide both termination (to the OCP and ultimately the caller) and hosting (to the NTS service provider), and so perform the function of bringing "the two sides of the market together" which, as discussed above, is a feature of two-sided markets. In addition, the structure of charges is crucial to the way NTS markets operate, since a TCP’s ability to attract NTS service providers (and generate calls from OCPs’ retail customers) depends on the level of the payment that it can offer NTS service providers. 4.67 Callers typically pay a price for NTS calls that is high relative to the costs of providing the underlying telecoms service, while NTS service providers pay little or receive a revenue share payment. For the purposes of illustration, a BT retail customer may pay up to 10ppm to make a daytime 0870/0871 call, of which only some 6.7ppm would be retained by the provider(s) involved in the call to cover the cost of origination, conveyance and termination.77 The remaining 3.3ppm would be passed to the NTS service provider as a revenue share. 4.68 TCPs compete with each other to attract NTS service providers, with greater success in doing this being rewarded by receiving an increased volume of calls, on which a termination payment is earned. A firm that sells NTS hosting to a particular NTS service provider always terminates traffic on behalf of that NTS service provider. The role played by NTS service providers is key since, as noted in paragraph 4.52 above, they select the retail price paid by callers, by choosing a particular number range (or, in the case of 0844, 0871, and 09 number ranges, a particular price point within a range: see Table 1 above). 4.69 On the face of it, the market has some of the characteristics of two-sided markets, in particular the dependency of call volumes on the structure of charges (with callers paying charges which are high relative to TCPs’ costs and NTS service providers paying relatively little or receiving a payment, as described in paragraph 4.67 above).78 4.70 In the next two sections we consider the implications of this for the most appropriate market definition in the circumstances of this case. Single versus separate markets: the general case 4.71 77 Ofcom agrees that in most two-sided markets it is likely be appropriate to examine the two sides separately, as, in effect, distinct (but inter-related) markets and is fully aware that this approach has been followed in a number of cases which have been examined by competition authorities. For example in assessing the market for mobile Estimates derived from BT price list and information provided by BT on hosting contracts. Jean-Charles Rochet and Jean Tirole, Defining Two-Sided Markets, 2004, presented at UCL twosided markets conference, May 2006, and as referred to in The Challenges of a digital World and the need for a new regulatory paradigm, Leonard Waverman, London Business School, essay in Communications - The next decade, a collection of essays prepared for the UK Office of Communications, November 2006, Chapter 3, page 173. 78 46 call termination, both Ofcom and the Competition Commission defined distinct markets for termination, on the one hand, and for retail mobile calls (that is, calls from a mobile) and mobile access combined, on the other.79 4.72 In Aberdeen Journals the CAT upheld the OFT’s market definition which focused on the relevant product market for advertising space in local newspapers (as opposed to a single two-sided market also encompassing the provision of newspapers to readers).80 4.73 In Carlton/Granada, the Competition Commission defined separate relevant product markets for advertising and broadcasting activities.81 In the report of its investigation of the Classified Directory Advertising Services (CDAS) market, the Competition Commission explained its market analysis in the following terms: “Industries such as CDAS are described as ‘two-sided markets’ – providers of CDAS compete both for users and for advertisers. The value of CDAS for advertisers depends on the level of usage by consumers, which, in turn, depends on the total amount of useful advertising they find in these publications. This creates a ‘network effect’, by which more advertising attracts more usage which in turn attracts more advertisers. Because users do not pay directly for the use of CDAS, CDAS providers’ profitability will be determined by advertiser revenues (although these, in turn, may be affected by usage); therefore we examined the market at the advertiser level.”82 4.74 When assessing whether it is appropriate to treat the relevant market as a single market or to define distinct markets for each of the two sides, careful consideration must be given to the circumstances of the case. Single versus separate markets: features of NTS hosting/termination 4.75 Ofcom believes that, in the particular circumstances of NTS call termination, it is appropriate to consider both sides of the market simultaneously and not necessary to analyse the two sides of the market separately, as two distinct markets (one for NTS call termination and one for NTS hosting).83 79 See Wholesale mobile voice call termination - Preliminary consultation on future regulation, Ofcom, June 2005, http://www.ofcom.org.uk/consult/condocs/termination/260123/ and Competition Commission, Vodafone, O2, Orange and T-Mobile, January 2003, published at: http://www.competition-commission.org.uk/rep_pub/reports/2003/475mobilephones.htm#full. 80 Aberdeen Journals Ltd v OFT [2003] CAT 11 at [301]-[308]. 81 Competition Commission, Carlton Communications Plc and Granada Plc: a report on the proposed merger (Cm 5953), October 2003, published at: http://www.competitioncommission.org.uk/rep_pub/reports/2003/482carlton.htm. The implications of two-sidedness do not appear to have been explicitly considered although at paragraph 5.64 it is noted that “money [i.e. advertising] follows viewers”. 82 Competition Commission CDAS market investigation, statement of 21 December 2006 (see footnote 73), paragraph 5.9. 83 This seems to be the approach adopted by the Competition Commission in its November 2000 report on the merger between Regional Independent Media and others. See for example the discussion of market share measures at paragraph 4.14, and the discussion of market power at paragraph 4.60. The Competition Commission considers whether in a local market it is likely that “the merged entity would wish to exploit its position, for example by raising cover prices or advertising rates, within that market”. See http://www.competitioncommission.org.uk/rep_pub/reports/2000/447rim.htm. 47 4.76 In a two-sided market, there may be an optimum pricing structure (the balance of charges between the two sides), which may involve prices on each side of the market that reflect considerations in addition to the costs of serving that side in isolation. Market power on one or other side of the market may distort the structure away from the optimum. In such cases, it may be appropriate to analyse the two sides of a two-sided market separately. 4.77 However, the specific features of NTS mean that the characteristics we generally expect to find in two-sided markets are either not present or are relatively unimportant in this case. 4.78 • The pricing structure (the balance of charges between the two sides) is determined by regulation on the termination side of the market, and competition on the hosting side. This means that the market outcome is unlikely to produce a price structure which is a long way from the optimum, and structure is not therefore an issue in this case. • There is no obvious means of determining the “competitive” level of the charge on each side of the market, which varies according to the NTS number range, which is selected by the service provider. A variety of different price structures therefore exist in the market (compare 0800 with 09 for example). • Callers to some ranges benefit from a low or negative price for hosting, that is, the outpayment to the service provider, as this enables the NTS revenue sharing mechanism to function as a micro-payment mechanism and encourages the development of new services (e.g. dial-up internet access). • Service providers are likely to be directly concerned with the price on both sides of the market. These four specific features of NTS are considered in more detail in the following paragraphs. We set out below how these features, and in particular the role of NTS service providers in choosing the retail price, mean that it is appropriate in this case to analyse the two sides as a single market. Pricing structure is in effect determined by regulation 4.79 A key feature of NTS markets – the structure of prices (that is, the relative levels of the gross termination charge and the revenue share outpayment) – is not a concern being examined in this investigation. Ofcom is not, in this case, investigating whether the structure of charges, with retail callers paying a relatively high charge and service providers paying relatively little or even receiving a revenue share, is inappropriate.84 4.80 Existing regulation (the NTNP and the NTS call origination condition) already provides a mechanism that allows issues of pricing structure between the two sides to be addressed to a large extent, by enabling service providers to choose from a menu of retail NTS call prices. While it might therefore be possible to define a separate market for each of the two sides in which an NTS terminating provider is involved (termination and hosting), this would not illuminate the analysis of the case. 84 As noted earlier, this pattern is in fact an intentional characteristic of the NTS regime. Ofcom has considered the range of call types which may appropriately fall within this regime elsewhere, in the NTS consultation for example, where it also proposed measures to address concerns about the level and transparency of certain charges. 48 Therefore, a key advantage (in general) of analysing the two sides of the market separately, the ability to consider the balance of prices between the two sides, is not relevant for the purpose of this investigation. No means for accurately assessing competitive prices separately on each of the two sides of market 4.81 Because the optimum pricing structure does not necessarily reflect the underlying costs on either side of the market, it would be difficult to assess the “competitive price”, and hence the right starting point for a SSNIP test, for the two sides of the market separately. Identifying the competitive price level for the net termination charge (where the net price is the total unit revenue retained by the TCP following its transactions with both OCPs and NTS service providers and, where relevant, transit providers, as opposed to the gross charge or price, which is the price that the TCP charges to OCPs, as noted at paragraph 4.27 above) in the single-market approach is less difficult, because it reflects the combined underlying costs of servicing both sides of the market i.e. termination and hosting. Similar issues may arise in other two-sided markets such as classified directories, as noted in paragraph 4.73 above. In some cases the price for hosting is zero or negative 4.82 As noted at paragraph 4.31 above, in some cases the price of hosting is negative, that is, the TCP pays the service provider a revenue share in return for which a service such as a customer helpline or internet access is provided (see paragraph 2.12 above). As explained in paragraph 4.99 below, the level of the outpayment reflects the pressure of competition to attract service providers. A range including zero or negative prices appears therefore to be consistent with a competitive outcome yet does not, by itself, bear any obvious relation to the costs of hosting. This further illustrates the difficulty in analysing pricing constraints for NTS hosting separately from pricing constraints on NTS termination. Service providers are directly concerned with the price on both sides of the market 4.83 A number of NTS service providers operate primarily as retailers, including a disparate range of consumer markets such as banking, white goods etc, and use NTS numbers to deliver after-sales service. A relevant factor in the competition between NTS service providers in these consumer markets may be the levels and balance between the price for making NTS calls and the price for the primary consumer service (banking, white goods etc). This means that the NTS service provider will be concerned directly with the price on both sides of the market, including the charge to the caller. An implication of these features is that it is not necessary to analyse the two sides of the market separately. 4.84 There is a parallel between the service provider’s ability, in this case, to choose from a menu of retail NTS call prices, and the role of “side payments” in two-sided markets.85 As pointed out by Rochet and Tirole, side payments can enable the externality between customers on the two sides to be internalised.86 This means that, 85 Side payments are payments directly between the parties. Callers and called parties (NTS service providers) are generally unable to make such side-payments among themselves. 86 Op cit, page 11. Rochet and Tirole refer to “side transfers” and “side contracting”. One source of externality in this case is that the called party benefits from a call being made but the caller will not (in general) take this into account when deciding to make the call. Another externality may arise because the cost that a party on one side (of the two sided market) is willing to bear depends in turn on the number of customers on the other side of the market (and vice versa). 49 where side payments enable the optimum price structure to be achieved, there is no need to analyse the two sides separately, and the market may be analysed as if it were a single market (for NTS call termination/hosting). We consider that a service provider's ability to choose from a menu of NTS call prices can approximate to a side-payment mechanism, albeit imperfectly. 4.85 For example, suppose a service provider on an 0870 number decides that the desired structure in fact entails it paying for the call in order to encourage calling (because of a positive externality). If explicit side-payments were possible, the service provider could agree with the caller for a side payment of the requisite amount to be paid to the caller (which could vary by caller). Such payments are not feasible but, within the NTS regime, there is another option open to the service provider. It can choose a different number range which does feature the desired structure – in the example above, 0800 would be more appropriate. In this sense, the service provider's ability to choose from a number of different price points for the call appears to have a similar role to the side-payment, in enabling the externality between the caller and called party to be internalised, and in allowing a relatively efficient price structure to be achieved by the market. 4.86 In the light of the above discussion, we believe it may be helpful to identify three different cases for the purposes of market definition in this decision. We set out below the terminology that we have used to describe these cases in this section. Terminology of course is not in itself important except insofar as it aids understanding. But with this in mind, and with the caveat that it is proposed for the purposes of this case only, we propose the following typology of markets: i) a single-sided market is one in which there are no interdependencies. This would include “standard” cases, such as vegetables purchased from a market stall; ii) a two-sided market is one in which there are interdependencies between the two sides. The sort of interdependency which is relevant here is found for example in the classified directories, newspapers and other markets described in paragraphs 4.60-4.65above. The key feature of such markets is that total sales depend on the structure of charges, not just their level, and in particular on the balance of charges between the two sides (for example, on the fact that a particular newspaper may be free to the reader, with costs being covered by charges made to advertisers). It is possible in such cases that market power on one side of the market leads to a distortion in the price structure and therefore it is generally appropriate to analyse the two sides of the market separately; and iii) a special case of ii) in which the interdependencies are internalised, for example, by means of a side payment mechanism. It might be helpful to think of such markets as “single(-sided) platform” markets. For the reasons set out above, notably the ability of service providers to choose the retail call price from within a range subject to regulation, NTS call termination/hosting is an example of this kind of market. In these cases, it may be appropriate to analyse the two sides of the market together, because the side payment or equivalent mechanism means that a reasonably efficient price structure can be achieved by the market. Questions of market power, or alleged abusive behaviour can then be analysed by looking at the market as a whole. 50 NTS call termination/hosting market: analysis 4.87 In the discussion of market definition that follows, Ofcom considers the constraints that apply to NTS call termination and NTS hosting jointly, based primarily on the net, rather than gross, price of termination.87 4.88 Where the TCP makes revenue share payments to NTS service providers in order to attract them to its network, the net charge is lower than the gross charge. For some calls, such as 0845 voice, the service provider may make a payment to the TCP rather than receiving a revenue share. In such cases, the net termination charge is higher than the gross charge. 4.89 The market in question is referred to as NTS call termination/hosting. In the analysis that follows, Ofcom therefore applies the hypothetical monopolist test based on the net price of NTS call termination/hosting. In other words, it considers the response of customers following a 5-10% increase in the net price of NTS call termination. 4.90 In the following section (paragraphs 4.92-4.151), the hypothetical monopolist test is used to define the boundaries of the market for NTS call termination/hosting. To this end, the effect of a SSNIP in the net termination charge is considered in order to test, in turn, each of the following propositions: a) the relevant market encompasses NTS call termination/hosting services sold by all TCPs (see paragraphs 4.92-4.99); b) the relevant market spans all NTS number ranges (see paragraphs 4.100-4.107); c) the relevant market includes both voice and data termination (see paragraphs 4.108-4.141); and d) other non-geographic call types fall outside the relevant market (see paragraphs 4.142-4.150). 4.91 The geographic extent of the market is considered from paragraph 4.152 et seq. The relevant market encompasses NTS call termination/hosting services sold by all TCPs 4.92 The net termination charge levied by TCPs varies by number range and by customer. For example, considerably larger amounts of money are available for revenue sharing on the more expensive number ranges (087 and 09), while, on a given number range and other things being equal, NTS service providers that receive greater volumes of incoming traffic tend to receive higher per minute revenue share payments. 87 Similarly, in the OFT consultation Newspaper and Magazine Distribution: Public consultation on the draft opinion of the Office of Fair Trading (OFT 851), 31 May 2006, published at: http://www.oft.gov.uk/shared_oft/reports/comp_policy/oft851.pdf, footnote 203, the OFT states: “Consider a publisher which increases prices to one side of the market, for example by demanding a higher share of the cover price (thereby reducing wholesalers’ and retailers’ margins and possibly leading to an increase in [carriage service charges]). This will result in: (i) fewer copies being sold to final consumers (for example, because some retailers exit the market leading to reduced availability and thus fewer sales); and (ii) higher profits on the copies that are still sold. Analogously to a firm in a one-sided market, the publisher would balance effect (i) against effect (ii). The difference in a twosided market is that effect (i), as well as including the lost revenue on foregone sales, is reinforced by a fall in advertising revenue (resulting from the fall in circulation)”. 51 4.93 Following a 5-10% increase in the net termination charge levied on a particular NTS service provider by a TCP (i.e. a fall in revenue share payments), it would be relatively straightforward (subject to any contractual obligations that may apply, for example) for the NTS service provider to switch to another TCP for NTS call termination. 4.94 This is based on Ofcom’s understanding that investments that need to be undertaken to supply additional NTS number ranges are small (see paragraph 5.31 below).88 4.95 Different TCPs also compete with each other to provide NTS call termination at the stage when NTS service providers entering the market are deciding which TCP to use to enable them to provide services to callers. Ofcom considers that substitution by service providers between TCPs is likely to be sufficiently large to make the SSNIP unprofitable. Therefore all TCPs should be included in the relevant market. 4.96 There are a number of precedents for markets for call termination services being defined on a network specific basis.89 In such markets, the calling party pays arrangement means that a SSNIP in the price of call termination is borne by the caller, who is unable or does not have the incentive to influence the choice of TCP and therefore does not respond to the SSNIP by choosing an alternative TCP (as this choice is made by the called party). 4.97 This is sometimes referred to as the “calling party pays externality” (because the cost or benefit arising from the choice of terminating operator falls on parties who do not participate in it). This tends to lead to network-specific call termination markets. 4.98 In general, the calling party pays externality does not play a key part in determining the competitiveness of NTS call termination. However, as explained in paragraph 5.46 et seq below, non-BT TCPs are not able to exert an influence on gross termination charges or retail prices (except to the extent that they select the retail price point for calls to 0844 and 0871 numbers) – whereas BT is. 4.99 The key source of competitive pressure therefore comes from NTS service providers, who have an incentive to choose the TCP that offers them the highest revenue share (i.e. imposes the lowest net termination charge at the desired retail price point). As noted above in paragraph 4.83 above, in some cases this price point may reflect competitive pressure in final goods markets in which NTS service providers operate and, unlike fixed geographic and mobile calls, the called party (in the case of NTS, the NTS service provider) has a role in setting the retail price. NTS service providers therefore provide a competitive constraint across all TCPs (see paragraph 4.66 et seq above).90 The relevant market spans all NTS number ranges 4.100 There are a number of limitations to demand-side substitution between different number ranges. 4.101 Consumer research undertaken by Ofcom shows that consumers’ perceptions regarding the relative cost to them of calling NTS numbers depends on the number range. These perceptions reflect real differences in the prices of calls to different 88 NTS call termination market review consultation, paragraph 4.17 et seq. See footnote 79. 90 Some indication of the importance of the revenue share outpayment, though in a different context, was collected for Ofcom’s review of NTS: see for example the NTS statement, paragraph 3.56. 89 52 NTS number ranges and the types of services that tend to be offered (see Table 1 and paragraph 2.4 above).91 4.102 This means that, facing a 5-10% increase in net NTS call termination charges, NTS service providers offering services terminated on a particular number range might be reluctant to switch to another number range (whether or not they switch to another TCP at the same time) as this could have an impact on call volumes. The extent of this reluctance is difficult for Ofcom to determine. However, Ofcom’s view is that it is not necessary to examine this issue in detail given the scope for supply-side substitution between the provision of termination/hosting services using different number ranges, which, as will be argued below, is significant in this market, and the homogeneity of competitive conditions. 4.103 If a hypothetical monopolist supplier of NTS call termination/hosting on (for example) the 084 range increased the net charge levied for its services, TCPs providing NTS call termination/hosting on the 080, 087, and 09 ranges would be able to switch quickly into supplying the same service on 084 (subject to allocation of numbers by Ofcom and compliance with the NTNP) without having to make substantial additional investments (see paragraph 5.31 below). Once a TCP has established the network platform needed to terminate NTS calls, the platform can be used equally for all types of NTS voice calls. In addition, the marketing and distribution arrangements for the various types of NTS are very similar. 4.104 Supply-side substitution between number ranges (for example by terminators of 0870 calls into the termination of 0845 calls) will, however, only place additional constraints on TCPs beyond those already captured by demand-side analysis to the extent that there are additional players not already materially present in the supply of 0845 terminations. Such a situation is likely to exist in the case of NTS call termination/hosting, since, to some extent, some TCPs specialise in the termination of a certain type of traffic (e.g. only 0845 traffic). This is demonstrated in Figure 5 below, which shows that, currently, TCPs terminate quite different mixes of each traffic type. Figure 5 shows the percentage split, by number range, of total terminated minutes of the six largest terminators of NTS traffic based on 2004/05 data. 4.105 It is clear from Figure 5 below that different TCPs terminate significantly different mixes of traffic, and in particular that there are a number of terminators of 0845 traffic that do not terminate significant volumes of the other traffic types. Given that, for the reasons set out in paragraph 4.103 above, little additional cost need be incurred to terminate calls on additional number ranges (at least in the case of voice calls – see paragraph 5.33 below), a monopoly supplier of termination/hosting on the 09 or 087 ranges could be additionally constrained by supply-side substitution from communications providers currently providing termination on 0845 number ranges.92 91 Ofcom commissioned residential and business consumer research in July 2004 among GB adults and Oftel residential consumer research carried out in August 2003 among UK adults. The research was intended to examine awareness and perceptions of the cost of calling 084 and 087 numbers, perceptions of the cost of ‘local rate’ and ‘national rate’ calls why SME’s use NTS numbers and reactions to changing price structure. See Annex 7 of the NTS consultation, e.g. paragraph A.7.3. 92 The figures in the table below are all rounded to the nearest whole percentage point, meaning that in cases where less than 0.5% of a TCP’s total traffic is to a particular number range the figure is listed as 0% (and also that, in some cases, the row entries in a particular column do not sum to 100%). However, some TCPs generate negligible amounts of traffic on certain ranges, and do not therefore exert a significant competitive constraint. 53 Figure 5: Traffic mixes of different TCPs, 2004/0593 [] Source: Responses to section 26 Notices of 19 May 2006 (see paragraph 3.27). 4.106 In the above discussion, we provided the example of TCPs who specialised in 0845 and could additionally constrain the 09 and 087 number ranges, since the supplier could easily provide termination services in these number ranges given the small investments needed. It is also the case that some TCPs who specialise in termination in 0845 are also present in the provision of NTS termination in 09 and 087 number ranges, albeit as small players. The potential for expansion by suppliers with a small but material presence in more than one number range (as set out in Figure 5), is suggestive of a degree of homogeneity of competitive conditions not necessarily reflected in market shares. 4.107 Given this scope for supply-side substitution and homogeneity of competitive conditions, Ofcom considers that the services affected by NCCN 500 form a single product market, namely that for NTS call termination/hosting, which covers all NTS number ranges. The relevant market includes both voice and data termination 4.108 Ofcom has considered whether termination of voice and data calls to NTS numbers are both part of the relevant market.94 There are a number of considerations which have informed Ofcom’s view that a broader market including both data and voice exists. These are: • the NTS regulatory framework; • common pricing constraints; and • supply-side substitution. The NTS regulatory framework; 4.109 The key regulations governing the pricing of NTS call termination include: • the NTS call origination condition; and • retail pricing arrangements for NTS calls. 4.110 Ofcom does not consider that these regulations constrain BT (or other TCPs) to set the same gross termination charge for both NTS voice and data calls, or to charge other OCPs for termination at the same rate as they pay to other TCPs. Importantly, these regulations do not constrain the net termination charge retained by the TCP which will also reflect any variations in the (unregulated) outpayment to service providers. 93 [] supplies mostly 084x for data NTS calls related to its dial-up ISP business. C&W also provides 084x data NTS calls for other ISP services. 94 Data calls to NTS numbers are mainly for dial-up internet access. 54 4.111 Under the first regulation (the NTS call origination condition) BT is only permitted to make cost-related charges for originating and retailing NTS calls and must pass the remaining revenues over to the TCP. 4.112 Under the second (the NTNP), the various NTS number ranges have different designations for use, some of which relate to retail pricing (others relate to content, for example 098 which is reserved for sexual entertainment services). For example, the 080 range is designated as a range where no charges are made to customers unless those charges are notified to the customer at the beginning of the call. In selecting a particular NTS number block, a TCP takes on an obligation to use the numbers in accordance with the designations given in the NTNP. In this sense, the TCP is restricted in its use of NTS numbers. 4.113 Ofcom notes that NTS calls on the number ranges 0870/0871 are primarily voice calls (although not exclusively), whereas NTS calls to 0845/0844 include both voice and data calls. 4.114 Neither of the above regulations require TCPs to price voice and data call termination at the same rate. Common pricing constraint 4.115 Ofcom considers that for the period under investigation, data and voice NTS call termination/hosting were subject to a common pricing constraint, thereby placing them both in the relevant NTS call termination/hosting market for the reasons set out below. In particular: • technical constraints make it unfeasible for BT to introduce different gross termination charges for voice and data traffic without disproportionate cost and disruption; • other TCPs are in any case not able to vary their gross termination charges because of the combined effects of the regulation of retail NTS prices and BT’s market power in call origination and (single) transit, as set out in paragraph 5.46 et seq below. The amount BT pays other TCPs does not differentiate between voice and data traffic due to retail price regulation and BT’s market power in call origination and the call origination condition that applies to BT as a remedy to this market power (see paragraph 4.54 above); • differences in average (gross and net) termination charges for voice and data which are apparent in BT’s accounting data reflect different patterns of usage at different times of day (and therefore volume weights applied to the same parameter gross termination charges).95 In addition, differences in average net termination charges for voice and data service providers also reflect a tendency for larger service providers to receive higher outpayments (for example because of economies of scale and negotiating strength). 4.116 Ofcom considers that there are two aspects to this constraint: • whether or not TCPs can distinguish between data and voice calls; and 95 The reason for differences in average termination charges between voice and data arises from different patterns of usage at different times of day, and therefore volume weights applied to the same parameter prices (BT responses of 4 November 2005 and 16 December 2005 in response to Ofcom section 26 Notices of 21 October 2005 and 14 December 2005, respectively. 55 • whether or not TCPs can impose different charges for different types of calls. 4.117 Ofcom considers that it could be technically feasible for a telecoms network to distinguish between voice and data calls by identifying data calls. At least two methods could potentially be used. For example: • TCPs could identify a data call once the call is established. This could be accomplished by monitoring the call for the set-up tones that are exchanged between modems; or • A TCP might be able to distinguish data calls if it can tell (from the destination number, or another routing parameter) which calls are destined for its IP network, which is used to terminate only data calls. 4.118 In respect of the first option, Ofcom understands from BT that such technology is not currently incorporated in BT’s network. BT stated: “BT does not have a way of splitting long and short calls between voice and data.”96 4.119 In either case, in order to establish differential charging, the information identifying data calls would need to be passed to the interconnection billing systems, for example via the Call Detail Records. 4.120 For interconnection purposes, charges for calls to 0844 and 0871 numbers are set for each block of 100,000 numbers. Ofcom generally allocates numbers to communications providers in blocks of 10,000.97 This means that communications providers must specify the price point they require when they apply for numbers so that Ofcom can allocate a block within a group of 100,000 that has the required price point. Calls to 0800, 0845 and 0870 numbers each have a single price point for interconnection purposes. 4.121 Since voice and data services have historically been mixed up within 08 number blocks (in other words 0845 and 0844 are both used to provide voice and data NTS call termination), BT would need to depart from established interconnection billing arrangements in order to bill voice and data services differently. Differential billing for voice and data services therefore implies greater tariffing granularity unless a significant effort is made to separate voice and data services onto separate number blocks. 4.122 A TCP could adopt several strategies to introduce differential termination charges for 08 calls without greater billing granularity: 96 • it could introduce new number blocks, reserved for voice or data services and apply differential price points; or • it could modify its NTS call termination charges for existing number blocks that are predominantly used for data services – for example the 0844 09XX blocks that are designated for metered internet services in the NTNP. BT response to Ofcom section 26 Notice, 18 November 2005. Certain non-geographic ranges are allocated in blocks of 1,000, e.g. 0800 Freephone numbers and some Internet-only ranges. 97 56 4.123 The first approach would be the least disruptive but would of course only work for new allocations unless the TCP is able to persuade its customers to move to the new ranges. 4.124 The second approach is possible, but in practice most NTS data calls are terminated on numbers in ranges that contain a mix of voice and data services, so that this approach is not practical, and Ofcom is not aware of any providers that have applied this approach.98 4.125 Ofcom considers that it would be feasible for a TCP to depart from the current interconnection tariffing convention and start applying tariffs to smaller number blocks, or even individual numbers. However, increasing tariff granularity increases both the complexity and load on billing systems both for the TCP and other OCPs, and the TCP might therefore need to establish multilateral agreement with OCPs in order to introduce such a solution (although it might also attempt to introduce differential changes on a unilateral basis). In practice, TCPs other than BT have limited ability to negotiate with OCPs in any case.99 4.126 In conclusion Ofcom considers that there are currently technical constraints preventing TCPs from introducing differential termination charges for voice and data NTS calls that would involve significant cost to overcome. Moreover, Ofcom considers that the potential need to migrate existing, as well as new, customers onto different number blocks would be disruptive as well as costly for customers. Accordingly, Ofcom considers that, in practice, communications providers have limited incentives to price separately for voice and data customers (as evidenced by the fact that no TCP attempted to do so over the period under investigation). Ofcom concludes from this that TCPs’ charges for voice and data NTS call termination are subject to a common pricing constraint. 4.127 There are however differences in the average net termination charges for voice and data calls. As noted above, these reflect different patterns of usage at different times of day (and therefore volume weights applied to the same parameter gross termination charges). In addition, average revenue share payments for data calls may tend to be higher for a given gross termination charge and this could result in a lower net charge for termination. This may reflect economies of scale in serving larger service providers, and their negotiating strength. However, net termination charges for data and voice calls will tend to move together in response to any change in the gross termination charge for the reasons given above. 98 In the NTS call termination market review consultation, the following question was put to NTS call operators “Do stakeholders agree with Ofcom’s initial view that the termination of all NTS call types (0845/0870, etc, also voice and data calls) should be included within the same market? In response, several suppliers made statements that supported Ofcom’s view that there was no evidence of NTS data and voice calls being charged at different rates. For instance, BT stated that “BT agrees that substitutability patterns and pricing constraints suggest a single NTS call termination market covering all of the NTS number ranges and including voice and data traffic.” BT’s response to Ofcom’s Consultation Documents Translation Services: Options and Translation Services Call Termination, 7 January 2005, Annex C, page 48. Centrica stated that “Centrica is not aware of any pricing implications for differentiating between voice and data calls; therefore we also agree that it is acceptable for voice and data calls to be included within the single market for NTS call termination.”, NTS Call Termination Market Review: A response by Centrica, 5 January 2005, page 7. Stakeholder Reponses to the NTS call termination market review consultation are published at: http://www.ofcom.org.uk/consult/condocs/ntsctmr/resntcctr/. 99 This argument is explored further in paragraph 4.43, i.e. due to a combination of BT’s position of SMP in call origination, and the relationship between BT’s retail NTS prices and geographic call prices. 57 Supply-side substitution 4.128 As described above, supply-side substitution is relevant where suppliers that are not currently supplying data (voice) NTS call termination/hosting to a material extent, but are supplying voice (data) NTS call termination/hosting, would enter the supply of data (voice) within one year and without incurring significant cost in response to a SSNIP, and where a sufficient number of customers would switch to those new suppliers to make that SSNIP unprofitable.100 4.129 First, this section examines the competitive structures (including entry barriers) related to the provision of voice and data. Second we consider the ability of firms present in either voice or data to enter the ‘reciprocal’ segments of the market (i.e. if providing data the ease of supplying voice NTS call termination and vice versa). Third, this section assesses whether entry would be additional to that already considered as part of the demand side analysis (i.e. would there be new entry by firms not already present in the market). 4.130 Ofcom obtained anecdotal qualitative evidence during meetings with providers of NTS call termination that suggests the structure of the two segments differs materially. For example, Ofcom met with two small voice call termination businesses, [] and [], but it appeared from the discussions that there were few small businesses offering data NTS call termination.101 A discussion with [], an ISP (and one of C&W’s customers) suggested to Ofcom that it was more often larger providers such as BT and C&W who provide data NTS call termination, due to the fact they provide IP network and routing.102 4.131 Ofcom therefore understands that data NTS call termination is now typically provided by larger TCPs (including, for example, BT and C&W) that own an IP network, and that there are relatively few (if any) TCPs that provide data NTS call termination/hosting by interconnection to larger providers’ IP networks. This appears to be because a number of operators have constructed such networks primarily for the provision of unmetered termination. Given the continued demand for metered termination as well, the most efficient way to provide it is for operators with IP networks to use this infrastructure, enabling NTS data traffic to be taken off the PSTN at the earliest possible point. 4.132 This contrasts with the voice segment of the market, where entry requires potentially less investment and fewer points of interconnection. For example, there appears to be a proliferation of operators that interconnect with larger networks, and in some cases at only a few points of interconnection (“POCs”).103 4.133 This qualitative analysis suggests that it may be more difficult to expand from voice into data than from data into voice. In the case of a supplier of NTS data termination, it is likely that this operator could commence supplying NTS voice call termination relatively quickly in response to a 10% increase in voice termination charges, since the entrant could start off at a small scale (e.g. one POC) and could scale its investment over time as their market share grew. Therefore, on the basis of the 100 To conduct a SSNIP test, we set aside the assumption of a common pricing constraint to enable us to raise the prices of NTS data and voice call prices independently for the purpose of undertaking the test. 101 []. 102 Teleconference with [], 23 May 2006. 103 Teleconference with [], 12 April 2006 and meeting with [], 15 May 2006. 58 evidence presented above, the entrant would only need to interconnect with an existing network operator’s PSTN network, and this could be done at very few POCs. 4.134 However, it is likely to be more difficult for providers of NTS voice call termination, in response to a 10% increase in NTS data call net termination charges, to supply NTS data call termination. For instance, entry would require investment in IP networks. These network investments are likely to be substantial since digital local exchange (“DLE”) interconnection is required and would suggest entry within a year would be unlikely. Even if entry within a year were possible, suppliers operating in voice that are not operating in data tend to be small-scale, and the likelihood of such operators placing an effective constraint on existing operators in data therefore appears low. 4.135 Accordingly, Ofcom considers that there is likely to be an asymmetry between likely entry from data to voice, with possible barriers for those operating in voice, who seek to supply data termination but not vice versa. 4.136 Ofcom has given consideration to whether the likely entry from data to voice, in response to a hypothetical 10% SSNIP increase in the net price of voice represents an additional supply side constraint. One way of considering whether supply side substitution will provide additional constraints on voice is to consider whether firm shares differ between the two segments in order to identify suppliers of data terminations not materially present in the supply of voice terminations who could switch to the provision of the latter. 4.137 From the market share data available to Ofcom (see Figure 5 above), it appears that a firm (such as []) that was primarily a terminator of metered data traffic could impose an additional constraint on the pricing decisions of a hypothetical monopolist of the termination/hosting of NTS voice traffic.104 As [] is, in the UK, an ISP, it is reasonable to assume that reported TCP terminated traffic in the 084x range in Figure 5 relates to data. On this basis Ofcom considers that [], which is not significantly active in voice call termination, would provide an additional constraint were it to enter the voice segment. 4.138 Although there are some suppliers (such as BT) who are already significantly present in both voice and data, and therefore could not be expected to impose a material additional constraint on the behaviour of providers of voice NTS call termination, Ofcom considers that other suppliers (such as []) who are present in NTS data call termination could be expected to enter the NTS voice call termination segment in response to a hypothetical 10% increase in the net price of voice termination charges (i.e. asymmetric supply side substitution from data to voice call termination for calls to NTS numbers). 4.139 Ofcom acknowledges that, given the asymmetry in supply-side substitution and low likelihood of entry from providers of voice NTS call termination, there might be a case for regarding data NTS call termination as a separate market, if the alleged behaviour related to the provision of termination of data calls alone. However, given the common pricing constraint described above, Ofcom considers that the relevant market in this case spans both voice and data. 4.140 In conclusion, Ofcom has considered evidence of common pricing constraints and constraints arising from supply-side substitution. Ofcom finds that there is evidence that over the period of the investigation, termination of voice and data calls to NTS numbers were subject to a common pricing constraint. Although there may be scope 104 [] response to Ofcom section 26 Notice of 19 May 2005. 59 for BT to introduce separate prices for voice and data (through migrating existing customers and allocating new customers to new number blocks) BT did not attempt to do so during the period under investigation. 4.141 Ofcom also considers that the possibility of asymmetric supply-side substitution from data to voice combined with the evidence on common pricing constraints suggests that voice and data NTS call termination can be regarded as part of a single market for the purposes of analysing the conduct alleged in this case.105 Termination/hosting of unmetered internet traffic and calls to directory enquiries fall outside the relevant market 4.142 Ofcom does not consider that the market should be broadened to include termination/hosting of unmetered internet traffic or directory enquiries (“DQ”) services. Ofcom has considered the possibility of demand- and supply-side substitution and the homogeneity of competitive conditions in each case. Termination/hosting of unmetered internet traffic 4.143 Demand for unmetered (internet) termination is derived from the demand for retail internet access. There are separate markets for metered and unmetered internet access at the retail level because of the different characteristics of metered and unmetered access and the different needs of customers that use them. 4.144 Ofcom considers that there is limited demand side substitution between dial-up (i.e. metered) and unmetered internet access services. Facing a 5-10% increase in retail dial-up internet charges, Ofcom considers it unlikely that customers will switch to take up unmetered internet access in sufficient numbers to make the dial-up price increase unprofitable. 4.145 Metered services suit light users and those wanting flexible payments, whereas unmetered services are suited to heavier users (who are therefore likely to have a higher willingness to pay) and those wanting fixed monthly payments. Ofcom notes that dial-up customers are less likely to have this higher willingness to pay and therefore are less likely to be attracted to the flat rate price structure.106 4.146 On the supply side, there are significant differences between the requirements for networks supporting metered and unmetered traffic with the implication that there will be significant costs for metered suppliers wishing to start supplying unmetered services, even in the presence of regulated products such as FRIACO (flat rate internet access call origination).107 For instance entry barriers arise from the need for entrants to build out their networks to a high proportion of BT’s DLEs in order to supply unmetered termination economically (i.e. based on FRIACO) although, as noted earlier, such a network could also be used to provide metered terminations. 4.147 As far as Ofcom is aware, there are no suppliers of unmetered data NTS call termination that are not already active in the NTS call termination/hosting market. 105 106 See also footnote 98. Review of the wholesale unmetered narrowband Internet termination market, consultation, 17 March 2003, paragraphs 2.8 to 2.13, and Wholesale unmetered narrowband Internet termination services – UK excluding Hull area market November 2003, paragraphs C.14-C24, both published at: http://www.ofcom.org.uk/static/archive/Oftel/publications/eu_directives/index.htm. 107 FRIACO is a wholesale product purchased by TCPs from BT that acts as an input for retail flat-rate dial-up internet access. 60 Broadening the definition of the relevant market from NTS call termination to one for both NTS and unmetered termination/ hosting would therefore not necessarily identify any additional constraints beyond those already captured by the NTS-wide market definition. Calls to DQ services 4.148 Similarly, Ofcom’s view is that the relevant market should not be broadened to encompass the termination/hosting of DQ services. 4.149 Facing a 5-10% increase in net termination charges, NTS call customers would be unlikely to switch to calling numbers in the 118 range, which is reserved for DQ services. The different nature of the services offered by NTS and directory enquiry service providers makes such demand-side substitution between them unlikely. 4.150 Facing a 5-10% increase in net termination charges, it is unlikely that NTS service providers would switch to supplying services offered on the 118 number range, as 118 is reserved for the provision of DQ services only. While there may be services operating on other number ranges that would meet the criteria for the allocation of a 118 code, this does not apply to the majority of services currently offered on other NTS ranges so switching would not be an option. Conclusion 4.151 Ofcom concludes that in the context of its assessment of C&W’s allegations the relevant product market is a single two-sided market for the termination/hosting of NTS calls on all ranges, by all TCPs. Relevant geographic market 4.152 A key feature of the NTS call termination/hosting market is that the relationships NTS service providers have with TCPs are not based on geographic considerations (see Annex 4). 4.153 An NTS service provider that is physically located in any area of the UK, including Hull, is able to buy services from any TCP, and is not restricted to TCPs located in the same area. Similarly, callers in any part of the UK, including the Hull area, can reach service providers wherever they are located. From the service provider’s point of view therefore, termination/hosting services provided anywhere in the UK are likely to be seen as demand-side substitutes. This substitutability may also be expected to be reflected in relative homogeneity of competitive conditions throughout the UK. 4.154 For some types of telecoms service, separate geographic markets have been defined for the Hull area (in which BT does not operate and Kingston Communications is the legacy incumbent) and the rest of the UK. However, service providers and callers in the Hull area are in a comparable position (as far as NTS call termination is concerned) to service providers and callers in other parts of the UK. 4.155 The implication of the above factors is that termination/hosting on Kingston’s network is substitutable for termination on the network of any other TCP. Ofcom therefore 61 concludes that the geographic area of the NTS terminations/hosting market is the UK.108 Conclusion 4.156 Ofcom concludes that in the context of its assessment of C&W’s allegations the relevant market is the market for NTS call termination/hosting in the UK. Responses to draft decision: Ofcom’s approach to market definition 4.157 Four respondents to Ofcom’s draft decision commented on Ofcom’s approach to market definition. BT 4.158 BT states that: “Generally BT concurs with the approach that Ofcom took to defining the market but differs from Ofcom in its conclusions.”109 4.159 BT considers that, rather than a single two-sided market, there are two separate markets: “(i) for NTS call terminations in the UK, and (ii) NTS hosting in the UK.”110 4.160 BT argues that: “There are significant differences between the structure of a true two-sided market as exemplified by the Newspaper/publishing industry and the structure of the market for NTS calls.”111 4.161 Specifically: “In a true two-sided market there is a direct correlation between the behaviour of the purchasers of the products (i.e. a published periodical and the advertising space within that newspaper) forming the two sides of the market.”112 4.162 According to BT, this means that the price charged to purchasers impacts on circulation (an increase in price leads to a decline in circulation and vice versa). However, in the case of NTS, if BT were to offer higher revenue shares in the NTS hosting market, it does not follow that it would then be able to charge more for NTS call termination.113 108 Because the geographic market definition does not materially affect the dominance assessment or conduct analysis, Ofcom does not undertake a detailed analysis of it here (for further detail see Annex 4, particularly A4.26). 109 BT’s comments on the draft decision, paragraph 2.4. 110 Ibid. 111 BT’s comments on the draft decision, paragraph 2.2. 112 BT’s comments on the draft decision, paragraph 2.3. 113 BT’s comments on the draft decision, paragraph 2.4. 62 4.163 BT argues a SSNIP analysis would suggest that NTS service providers will be inclined to switch to the TCP that pays the highest revenue shares (as indeed Ofcom suggests at paragraph 4.79 of its draft decision) and concludes that: “This suggests that the market for NTS hosting should be as wide as all TCPs providing NTS hosting to NTS service providers.”114 4.164 However, according to BT, a SSNIP analysis of NTS call termination shows that substitutability for a particular NTS number is limited, which suggests that: “the market for NTS call termination is only as wide as calls that may be terminated on an individual TCP’s network.”115 4.165 BT concludes that: “As the product market for NTS hosting encompasses all TCPs providing NTS hosting and the market for NTS call termination is limited to individual TCPs networks then this suggests that there are two separate markets as opposed to a two-sided market.”116 4.166 BT argues that in fact there may be a number of narrower (network-specific or rangespecific) markets for NTS call termination. BT notes that the termination charge for a particular NTS number (unlike revenue share outpayments): “does not vary depending on the NTS service provider customer and is fixed to the number range in question or in the case of 0844, 0871 and 090 the price point chosen by the NTS service provider.”117 4.167 In particular BT notes, in its introductory remarks on market definition, that it may not be appropriate to include 0845 and 0870 in the same market as other NTS ranges: “as the effect of the NTS call origination [condition] may be such as to lead to the conclusion that substitutability and competitive constraints differed for these number ranges when contrasted with other number ranges.”118 4.168 Nevertheless, BT does not consider that it is necessary, for the purposes of reaching a view on C&W’s complaint, to conclude whether there is a single two-sided market or two separate markets, or whether the relevant market definition for NTS call termination is narrower than that proposed by Ofcom: “the main purpose of the market definition is to identify the competitive constraints that BT faced…BT is of the view that these constraints would be the same whether the relevant market for BT’s NTS call termination products and hosting products is one two-sided market or two markets each consisting of one side of the doublesided market as defined by Ofcom.”119 114 BT’s comments on the draft decision, paragraph 2.32. BT’s comments on the draft decision, paragraph 2.36. 116 BT’s comments on the draft decision, paragraph 2.37. 117 BT’s comments on the draft decision, paragraph 2.39. 118 BT’s comments on the draft decision, paragraph 2.7. 119 BT’s comments on the draft decision, paragraph 2.6. 115 63 C&W 4.169 C&W states that Ofcom’s approach to market definition departs from standard practice in two key respects: “According to the OFT’s Market Definition Guidelines an analysis of the relevant market starts with the product or products relevant to the complaint, expanding the market depending on demand and supply side substitutability – Ofcom does not follow this approach. Ofcom’s approach to market definition in two-sided markets does not appear to be supported by case law and appears generally inconsistent with the standard economic views on how the market definition exercise should be conducted in relation to two-sided markets.”120 4.170 C&W submits that Ofcom’s approach to market definition “results in serious errors in the subsequent analysis of BT’s conduct thereby raising questions as to the validity of that analysis”.121 4.171 C&W submits that Ofcom should have analysed the two sides of the two-sided market separately (which is the approach taken in the case law quoted in the draft decision), and that Ofcom does not adequately explain why it examined them simultaneously. C&W cites, in support of its view, an article by David S Evans on the analysis of two-sided markets and a presentation by Amelia Fletcher, the OFT’s Chief Economist.122 4.172 C&W submits that Ofcom’s approach to market definition has implications for its subsequent analysis of its excessive pricing allegation. The market definition adopted, according to C&W: “does not allow an identification of the customer who suffers the excessive price, i.e. whether they are hosting or termination customers.”123 4.173 C&W believes that this: “leads towards an analysis based on profitability, rather than an analysis of pricing. This produces a one-dimensional assessment of excessive pricing, rather than viewing it in the round.”124 4.174 C&W further submits that Ofcom’s approach to market definition has implications for its analysis of C&W’s margin squeeze allegation.125 4.175 On the retail side of the market, C&W argues, a different approach to market definition (i.e. the application of the SSNIP test as set out in the OFT’s guidelines) “would have identified the products relevant to the margin squeeze allegation, i.e. NTS calls subject to NCCN 500”.126 In other words, C&W seems to be arguing that 120 C&W’s comments on the draft decision, page 10. Ibid. 122 C&W’s comments on the draft decision, page 12-13. 123 C&W’s comments on the draft decision, page 14. 124 Ibid. 125 C&W’s comments on the draft decision, page 15. 126 Ibid. 121 64 Ofcom’s failure to properly conduct a SSNIP test led it to apply the retail margin squeeze test on the wrong product set. 4.176 C&W states that, “with regards to the hosting market the errors in market definition may be even more marked”. C&W submits that Ofcom’s market definition exercise fails to take into account: a) “The competitive nature of the hosting side of the market; b) How service providers choose which TCP to use for hosting purposes; and c) Whether the market may be further segmented by type of NTS number, or by type of customer”.127 4.177 The impact of this, C&W argues, is that Ofcom has applied the hosting margin squeeze on the wrong product set. C&W argues in relation to this point that Ofcom’s analysis of the hosting side of the market fails to recognise that BT’s product mix may be very different to that of its competitors.128 CPW 4.178 CPW contends that Ofcom has “departed from the conventional method of defining the relevant product market leading to a fundamental flaw in the provisional conclusions”.129 4.179 CPW “does not believe that current case-law supports Ofcom’s approach in analysing this market as one single two-sided market”, and submits that the standard approach to two-sided markets is to conduct a SSNIP test on each side of the market separately and to examine the “dynamic linkages” between the two to help explain how one side may constrain the other. CPW cites a number of cases in which the two sides of a two-sided market are analysed separately.130 4.180 CPW believes that Ofcom’s approach “leads to an over-emphasis on BT’s overall profitability” in its analysis of C&W’s excessive pricing allegation, and that it “does not allow an identification of the customer who suffers the excessive price, i.e. whether they are hosting or termination customers.”131 Thus 4.181 Thus submits that: ”Despite apparently setting out the correct approach to follow, Ofcom does not appear to have followed the suggested logic through and as a result has […] failed to define the market correctly.”132 4.182 Thus submits that Ofcom should have analysed the two sides of the market separately, and that: 127 C&W’s comments on the draft decision, page 16. Ibid. 129 CPW’s comments on the draft decision, page 1. 130 CPW’s comments on the draft decision, page 2. 131 CPW’s comments on the draft decision, page 3. 132 Thus’s comments on the draft decision, page 7. 128 65 “had the analysis been conducted in this way it would have been clear to Ofcom that there is no demand side substitution, because originating providers are not in a position to switch to another provider (i.e. NTS termination on the relevant number ranges is a bottleneck service, similar to geographic call termination, which is subject to an SMP condition).”133 4.183 Thus submits that NCCN 500 may have had an effect on call volumes and that this may have imposed an indirect pricing constraint on certain NTS calls, which “could have an impact on the relative attractiveness of individual communications providers’ retail offerings”.134 4.184 Thus also believes that Ofcom should apply the same principles to non-geographic call termination as it has to its analysis of geographic call termination, i.e. that each TCP’s own network constitutes a separate market. 4.185 In addition, Thus states that Ofcom has dealt exclusively with the retail market and has ignored the primary (wholesale) market where the price change was notified.135 4.186 Thus contends that Ofcom’s analysis of the hosting market was also deficient, and failed to go into sufficient detail.136 4.187 In its comments on Ofcom’s margin squeeze analysis, Thus mentions the 1280 prefix, which enables CPS customers to revert to BT on a call-by-call basis and would, in theory, enable them to use BT only for calls to 0845 and 0870 numbers. Thus contends that: “Whether or not this happened in practice, this undoubtedly acted as a competitive constraint on OCPs’ behaviour and, as such, demonstrates that the market is not as broadly bundled as Ofcom suggests.”137 4.188 Thus cites, as evidence for the relevance of 1280 to Ofcom’s analysis, online advice for consumers explaining how they can use 1280 on a call-by-call basis, and the wording of the letter sent by BT to customers who had switched from BT Retail to a CPSO at the time that NCCN was in force, which invited them to “check which calls are cheaper with BT”.138 Ofcom comments on responses 4.189 The respondents’ main concerns with Ofcom’s approach to market definition are that: 133 • although it describes the SSNIP test approach to market definition, Ofcom does not in fact carry out a SSNIP test; • Ofcom should have analysed the two sides of the market separately and then examined the “dynamic linkages” between them; • Ofcom’s analysis of the market on the hosting side is insufficiently detailed; Ibid. Ibid. 135 Ibid. 136 Thus’s comments on the draft decision, page 8. 137 Ibid. 138 Thus’s comments on the draft decision, page 8-9. 134 66 • Ofcom’s approach to market definition leads to flaws in its analysis of (and conclusions on) excessive pricing; and • Ofcom’s approach to market definition leads to flaws in its analysis of (and conclusions on) retail and hosting margin squeeze. Application of the SSNIP test 4.190 Ofcom’s approach to market definition reflects the approach set out in the OFT guidelines by: a) starting with the product or products relevant to the complaint, and b) expanding the market depending on demand and supply side substitutability. 4.191 C&W’s complaint relates to the increased prices for NTS call termination notified in NCCN 500 (see Annex 1). Ofcom therefore starts its analysis by considering the narrowest possible set of NTS call termination products, including NTS call termination provided on the network of a single TCP, termination of calls to a single NTS number range, voice NTS calls only and data NTS calls only, and exclusive of other non-geographic calls. 4.192 Ofcom has explicitly applied the SSNIP methodology to assessing the product dimensions of the relevant market. Specifically, Ofcom has adhered to the OFT guidelines when assessing market definition by considering the likely demand and, where appropriate, additional supply side substitution response to a 5%-10% SSNIP by a hypothetical monopolist. Ofcom has then interpreted whether the likely demand or supply side substitution would make the SSNIP profitable, thereby enabling it to assess whether the nearest substitute(s) fall within the relevant market or not. For instance, Ofcom has applied SSNIP tests to determine whether the relevant market: • encompasses NTS call termination/hosting services sold by all TCPs (see paragraphs 4.92-4.99 above); • spans all number ranges (see paragraphs 4.100-4.107 above); • includes both voice and data termination (see paragraphs 4.108-4.141 above); and • includes other non-geographic calls (see 4.142-4.150). 4.193 Ofcom also notes that in respect of Ofcom’s margin squeeze test, the relevant market for the purpose of applying the test is determined by the scope of retail competition, including an analysis of the product bundles across which competition takes place (see paragraph 6.40 et seq below). Ofcom’s analysis of the market as a single market, rather than as two separate markets 4.194 Ofcom sets out its reasoning for why the NTS termination/hosting market can be both defined and analysed as a single market at paragraphs 4.66-4.85 above. Ofcom believes that this analysis remains appropriate and has expanded on the arguments set out in the draft decision to address respondents’ comments. 67 4.195 C&W cites an article by David S Evans in support of its view that the two sides should be analysed separately in this case. However, it is clear that the point of Evans’s warning not to ignore interdependencies is precisely to avoid the error that might result if the market were defined in the way proposed by C&W and analysed without taking account of the interdependencies identified by Evans. 4.196 For example, in a more recent article than that quoted by C&W, Evans writes: “Price equals marginal cost (or average variable cost) on a particular side is not a relevant economic benchmark for two-sided platforms for evaluating either market power, claims of predatory pricing, or excessive pricing under EC law…thus it is incorrect to conclude, as a matter of economics, that deviations between price and marginal cost on one side provide any indication of pricing to exploit market power or to drive out competition.”139 4.197 C&W also cites the view of the OFT’s Chief Economist, Amelia Fletcher, in support of its view. However, it is clear that the presentation C&W refers to in its submission (made by Amelia Fletcher in a personal capacity) refers to two-sided markets in general. Ofcom acknowledges that it is appropriate to analyse the two sides of the market separately in most cases, but not in this one, and does not believe that Amelia Fletcher’s intention was to rule out in all cases the approach that we have taken here. 4.198 As set out at paragraphs 4.77-4.85 above, Ofcom considers that it is appropriate in this case to assess the two sides of the market together. In particular, the “dynamic linkages” referred to by some respondents are reflected in the price structure in a relatively efficient way by means of the ability of NTS service providers to choose the retail price, subject to the NTS regulatory regime. Moreover, adopting the alternative market definition proposed by respondents would not, in any case, affect Ofcom’s conclusions on BT’s position on the relevant market or on BT’s conduct (see paragraphs 4.204-4.209 below). 4.199 The literature on two-sided markets appears to acknowledge that the same conclusions may be reached whether a two-sided market is analysed as a single market or as two separate single-sided markets. For example, Evans and Schmalensee argue that the importance of two-sidedness will vary from case to case: “the constraints on market power that result from interlinked demand also affect market definition. Market definition assists in understanding constraints on business behaviour and assessing the contours of competition that are relevant for evaluating a practice. In some cases the fact that a business can be thought of as two-sided may be irrelevant. That could happen either because the indirect network effects…are small or because nothing in the analysis of the practices really hinges on the linkages…In other cases, the fact that a business is two-sided will prove important both by identifying the real dimensions of competition and focusing on sources of constraints.” 140 139 David Evans and Richard Schmalensee, The Industrial Organization of Markets with Two-Sided Platforms, Competition Policy International Spring 2007), p.174. 140 Op cit. 68 4.200 Moreover, as Tommaso Valletti says: “For an economist this second approach [i.e. analysis as a single platform] is bound to give the same answer...as the first approach [defining two separate markets].”141 4.201 This is not to say that other approaches would necessarily lead to incorrect conclusions. As Valletti has said, it should be possible to reach the same conclusions with other market definitions provided constraints are correctly identified – but in this case, the analysis of the two-sided market as a single market facilitates the identification and analysis of these constraints. 4.202 Ofcom has considered whether analysis of the two sides of the market separately would affect any of its conclusions on BT’s position in the relevant market or on whether BT’s conduct constituted an abuse, and has concluded that it would not (see paragraphs 4.204-4.209 below). 4.203 Ofcom does not therefore accept C&W’s contention that, had Ofcom adopted C&W’s proposed market definition (i.e. two single-sided markets, one for NTS call termination and one for NTS hosting), it would have reached different conclusions in its analysis of BT’s position on the relevant market. Nor does Ofcom accept BT’s contention that it is not dominant whichever of the two approaches to market definition we adopt. Analysing the two sides separately would not affect Ofcom’s conclusion that BT is dominant in a relevant market 4.204 BT does not agree with Ofcom’s provisional conclusion that it is dominant in a relevant market. BT was the only respondent to disagree with this element of Ofcom’s analysis. Ofcom’s analysis on dominance is set out at Section 5 below. The following paragraphs consider, for completeness, whether a different approach to market definition would have led to different conclusions on dominance. 4.205 Ofcom’s finding that BT is dominant in the market for NTS call termination/hosting would apply equally if the market were more narrowly defined, to consist of termination only, or if additionally restricted to termination only on BT’s network. 4.206 At paragraph 5.67 below Ofcom states that: “the focus…is on the constraints faced by BT as a TCP when it increases its net charges via an increase in gross termination charges [that is, the price on the termination side of the market], because this analysis goes directly to answering the question as to why BT can act independently with respect to price.” 4.207 In other words, Ofcom has considered whether there are any constraints on BT’s ability to increase its termination charges – exactly as would be necessary if the market were defined to include only termination on BT’s network. 4.208 The analysis of the responses available to OCPs and their ability to undermine an increase in BT’s gross termination charge set out in paragraph 5.68 et seq below would apply unchanged, as would the conclusion in paragraph 5.134 below that 141 Tommaso Valletti, Mobile call termination: a tale of two-sided markets published in Communications and Strategies, Q1 2006. 69 “possible strategies of OCP…customers are not capable of constraining the pricing behaviour of BT as a TCP.” 4.209 Ofcom has also considered possible constraints from service providers and other TCPs. As it has found that these do not, in practice, constrain BT’s ability to increase charges, and that BT is therefore dominant in the wider market, it follows that it would also be dominant in the more narrowly defined market. The fact that BT would, by definition have a 100% share of this market would lend further weight to the finding of dominance in this case. Other comments on Ofcom’s approach to market definition 4.210 THUS comments that “NCCN 500 may have impacted on call volumes and this could have led to an indirect pricing constraint”. 4.211 Ofcom notes that if NCCN 500 rates had been passed on in retail prices then there could have been an effect on retail NTS call volumes. However, the evidence indicates that, in general, NCCN 500 was not passed on to retail customers (although see paragraphs 5.114-5.115 below for details of []’s response to NCCN 500), and therefore it would be unlikely that there would be an impact on call volumes (see paragraph 5.45). Ofcom has also considered the explanation for why OCPs did not pass on the price increase, including the possibility that this would make their retail offerings less attractive, which we consider in our analysis of BT’s dominance (see paragraphs 5.95-5.111 below). A different approach to market definition would not have led to different conclusions on alleged excessive pricing 4.212 C&W and CPW submit that Ofcom’s approach to market analysis has led it to concentrate its analysis of C&W’s excessive pricing allegation on BT’s profitability, whereas it should have considered alleged excessive pricing “in the round” and, in particular, identified which consumers might be affected by the alleged excessive price (see paragraphs 4.172 and 4.180 above). 4.213 Ofcom’s approach to the excessive pricing allegation is set out in full in Section 6. Ofcom agrees that excessive pricing should be considered “in the round”, and that it is normally appropriate to consider excessive pricing from a number of different angles (as suggested by case law, notably Napp). 4.214 Ofcom notes C&W’s concern that Ofcom has not conducted its test of excessive pricing based on the gross termination charge: “it is of course the [price to OCPs] that is of concern to C&W in terms of all the allegations it has made i.e. that the…charge paid by C&W to BT is excessive, discriminatory, and causes a margin squeeze”. 4.215 It is also clear from C&W’s reference to Evans that it does not believe that the gross termination charge should necessarily be related to costs on the termination side of the market only, even if these could be identified. As noted at paragraph 4.79 et seq above there is no obvious means of determining the “competitive” level of the charge on each side of the market. This varies according to the NTS number range and is selected by the service provider. A variety of different price structures therefore exist in the market, with a wide range of “competitive” gross termination charges (compare 0800 with 09 for example), and the service provider is able to select the one appropriate for its business model. This does not of course mean that excessive pricing cannot occur in the NTS call termination/hosting market, only that it is 70 appropriate to consider it taking into account costs on both sides of the market including the level of any outpayment to service providers or, if higher termination charges are passed on in higher outpayments, by means of the hosting margin squeeze test. 4.216 In any case, Ofcom has considered whether prices were excessive for calls to 0845 and 0870 number ranges separately. Table 23 shows that prices for 0845 termination (alone) were not excessive and table 24 shows that this is also true of 0845 and 0870 calls together. Paragraph 6.428 below explains that the latter result is more marginal than the former largely because of economies of scale (SAC could be significantly higher for a business specialising in 0870 calls). 4.217 Ofcom does not consider that a different market definition would have led it to approach the excessive pricing allegation in a different way, and does not therefore believe that its approach to market analysis has led to flaws in its analysis of BT’s conduct, as the respondents suggest. A different approach to market definition would not have led to different conclusions on alleged margin squeeze 4.218 Insofar as the two margin squeeze tests are concerned, Ofcom has, in effect, carried out the “single sided” analysis apparently advocated by C&W. On the retail origination side of the market, it has assessed whether BT’s retail margin would have been adequate to cover relevant costs including gross NTS termination charges at the rates levied to OCPs under NCCN 500, while on the hosting side, it has considered whether other TCPs could have matched BT’s outpayments to service providers given the level of gross termination charges paid by BT to those other TCPs. This latter would be equivalent to a test of leverage of dominance into the hosting side of the market from a separate termination side. 4.219 The respondents’ concerns appear to focus on the range of products across which the margin squeeze analyses are carried out. These however do not depend on the analysis of the wholesale market as a single market or two separate markets. 4.220 In fact, the tests that Ofcom has carried out in order to establish the existence or otherwise of a margin squeeze on the hosting side of the market and excessive pricing are the same as those that would be conducted if a narrower definition of the market were adopted. As is clear from Tables 15–17 below and the discussion in subsequent paragraphs, there was no margin squeeze in NTS hosting for any of the relevant service increments. These include 0844/0845 alone, 0844/0845 and 0870/0871 and all number ranges. In each of these cases, the tests were carried out for voice and data separately. 4.221 Ofcom believes its approach to have been reasonable given the competitiveness of NTS hosting (C&W agrees that “there may well be considerable movement between TCPs and/or certain number ranges if there were a SSNIP” and that hosting is “potentially very dynamic”) and the absence of evidence that the product mix changed during the 20 months of NCCN 500.142 4.222 On the retail side, C&W argues (and other respondents also suggest) that the relevant market is “NTS calls subject to NCCN 500”.143 However, the scope of the retail-side margin squeeze test is determined by the scope of competition at the retail 142 143 C&W’s comments on the draft decision, page 16. C&W’s comments on the draft decision, page 15. 71 level, not the definition of the wholesale (NTS) market, and no retail competitor offers just this bundle of services 4.223 Ofcom has carried out the retail margin squeeze test on a number of possible service bundles, and found there to be no margin squeeze on even the narrowest relevant market definition. 4.224 In its guideline on market definition the OFT states that: “In some cases the relevant product market may consist of 'bundles' of what are otherwise distinct products…Whether this is appropriate depends on the investigation…The perspective of customers will be important in assessing the appropriate frame of reference.”144 4.225 In its guideline on assessment of conduct, the OFT states that: “If there is evidence that a vertically integrated dominant undertaking has applied a margin squeeze and that it harmed (or was likely to harm) competition, this is likely to constitute an abuse of that dominant position.”145 4.226 In light of the OFT’s comments as set out in the two preceding paragraphs (in particular, the need for a margin squeeze to be at least capable of harming competition for it to be an abuse) Ofcom believes that its treatment of the retail market is appropriate to the investigation. 4.227 Thus argues that CPS customers’ ability to use 1280 to divert some calls to BT demonstrates that the market (and the relevant bundle for conducting a margin squeeze test) is narrower than Ofcom suggests (see paragraphs 4.187-4.188 above). 4.228 However, if the 1280 function had, as Thus contends, been a significant competitive constraint it would presumably have resulted in pressure to bring retail NTS prices into line in advance of NCCN 500. The fact that some competitors apparently maintained retail NTS call prices at levels above BT’s (see paragraphs 5.115 and 5.119 below suggests that competition did not take place solely for these calls, but at the level of a larger bundle, as set out in the preceding paragraphs and considered more fully in Section 5 below. 4.229 Ofcom has not been able to find any evidence that 1280 is widely used, to the extent that it would represent the kind of competitive constraint Thus suggests. As noted at paragraph 3.35 above, Ofcom sent BT a further section 26 Notice requiring it to provide information about the use of 1280 during the period that NCCN 500 was in force. BT was unable to extract data from its systems that would have provided an accurate estimate of the extent to which 1280 was used over the period. 4.230 However, Ofcom does not believe that it would have been possible, by means of NCCN 500, for BT to have established a dominant position in the retail supply of calls to NTS numbers by encouraging use of 1280. Given that CPS operators (and cable operators to which 1280 is not relevant) could remain in the market, a fact confirmed by Ofcom’s margin squeeze tests, it would be easy for customers to cease using 1280 if BT subsequently raised retail prices (which it is prevented from doing in 144 145 72 OFT 403, paragraph 5.11. OFT, Assessment of conduct (OFT 414), paragraph 6.5. any case by the NTNP). This suggests that BT would not have been able to establish a position of dominance by this means. 4.231 Ofcom does not therefore agree that the availability of the 1280 function affects the relevant market definition or its analysis of margin squeeze. 73 Section 5 5 BT’s position in the relevant market Introduction 5.1 The European Court of Justice (“ECJ”) has defined a dominant position as: “a position of economic strength enjoyed by an undertaking which enables it to prevent effective competition being maintained on the relevant market by giving it the power to behave to an appreciable extent independently of its competitors, customers and ultimately of consumers.”146 5.2 For the reasons set out below, Ofcom has concluded that BT was dominant in the market for NTS call termination/hosting in the UK over the relevant period, given the particular features and structure of that market. 5.3 The key element of Ofcom’s dominance assessment is a detailed assessment of the competitive constraints faced by BT during the period that NCCN 500 was in force. This assessment directly considers the extent to which BT was able, over the relevant period, to act independently of its competitors, customers and ultimately of consumers. 5.4 The assessment is based on: 5.5 • an analysis of the incentives of BT’s customers and competitors following an increase in BT’s prices for NTS call termination/hosting (see paragraphs 5.665.94 below); • survey evidence which assesses the likely reaction of end users following different strategies that might be adopted by OCPs other than BT in the event of such a price increase (see paragraphs 5.95-5.111 below); and • a description of what actually happened following the price increases notified in NCCN 500 (see paragraph 5.112-5.124 below). In addition, Ofcom has considered other factors which might impose constraints on an undertaking’s ability to behave independently. The following section therefore begins with a consideration of two other widely used indicators of market power: market shares (see paragraphs 5.6-5.26 below) and barriers to entry (see paragraphs 5.27-5.34 below) and expansion (see paragraph 5.35-5.49 below). Market shares 5.6 An analysis of dominance often begins with an assessment of the market share of the allegedly dominant company. 5.7 In its guideline Assessment of Market Power, the OFT states that: “There are no market share thresholds for defining dominance under Article 82 or the Chapter II prohibition. An undertaking’s market 146 74 Case 27/76 United Brands v Commission [1978] ECR 207, paragraph 65. share is an important factor in assessing dominance but does not determine on its own whether an undertaking is dominant. For example, it is also necessary to consider the position of other undertakings operating in the same market, and how market shares have changed over time.”147 5.8 The guidelines also emphasise the importance of other competitive constraints including existing competitors, the scope for new entry, buyer power, economic regulation, firm behaviour and financial performance.148 5.9 In Hoffmann-La Roche, the ECJ stated that: “…although the importance of the market shares may vary from one market to another the view may legitimately be taken that very large market shares are in themselves, and save in exceptional circumstances, evidence of the existence of a dominant position. An undertaking which has a very large market share and holds it for some time… is by virtue of that share in a position of strength.”149 5.10 The OFT has stated that it considers it unlikely that an individual undertaking would be dominant with a market share of below 40% but it acknowledges that, if other factors are present such as the weak position of competitors in that market, dominance could be established at market shares below 40%.150 5.11 In Virgin/British Airways, the Commission found that British Airways was dominant in the UK air travel agency services market with a market share of 39.7%.151 This was upheld by the Court of First Instance (“CFI”) on appeal. 5.12 The presumptions that have developed in the case law to assist in assessing the degree of market power held by an undertaking from its market share reflect the (conventional) relationship between market shares and market power which is considered to exist when the effect of competing products is felt directly and the market is relatively stable and mature. Volume market shares 5.13 Figure 6 below shows how total NTS call termination/hosting volumes, and BT’s share of this market have evolved over the three full financial years to 2005. Figure 6: BT’s volume market share, 2002/03 to 2004/05 [] Source: BT response to section 26 Notice of 25 August 2005. 5.14 Figure 6 shows that, over this three-year period, the total size of the market declined from over [] minutes to under [] minutes. Over the same period, BT’s share remained broadly constant ([]% in 2002/2003 and []% in both 2003/04 and 2004/05). BT had the largest market share in 2002/2003. 147 OFT, Assessment of Market Power (OFT 415), 2004, published at: http://www.oft.gov.uk/shared_oft/business_leaflets/ca98_guidelines/oft415.pdf, paragraph 2.11. 148 OFT 415, paragraphs 3.3-3.6. 149 Case 85/76, [1979] ECR 461, [1979] 3 CMLR 211, paragraph 41. 150 OFT 415, paragraph 2.12. 151 OJ [2000] L 30/1, [2000] 4 CMLR 999 Commission Decision BA/Virgin and Case T-219/99 [2003] ECR II-5917 British Airways v Commission 75 5.15 For the period under investigation, BT’s market share by volume was [ (below 31%)]. 5.16 Ofcom does not consider that, in the particular circumstances of the market under investigation in this case, market shares are a reliable indicator of whether or not BT can act independently of its competitors and customers. This is because BT has a unique position in this market among competing suppliers of NTS termination/hosting, as a consequence of the NTS regime and BT’s SMP in wholesale call origination and transit arrangements (as identified at paragraphs 2.142.26 above and summarised in paragraphs 5.46-5.47 below). This unique position does not arise by virtue of BT’s share of the NTS call termination/hosting market. Accordingly, the market shares illustrated above are not particularly informative as to whether BT has a dominant position in the market. 5.17 It is worth noting, however, that in 2004/05, BT was in fact the largest NTS call terminator (by volume) with a []% share of all NTS call terminated traffic (see Figure 7 below). Energis and C&W (prior to their announced merger on 16 August 2005) had []% and []% of the market, respectively. For the period of the investigation prior to the merger of Energis and C&W, BT was the largest NTS call terminator in the market. Figure 7: TCP shares of all terminated traffic, 2004/2005 [] Source: TCP responses to section 26 Notices in May and August 2005. 5.18 Ofcom does not consider that BT’s position in the termination market has been reflected in a higher market share because until NCCN 500, BT had set the same gross termination rates as it paid to other TCPs and so had not chosen to use its position of market power (for example, by raising termination charges and outpayments in order to attract service providers). This changed when BT decided to unilaterally raise its gross termination charge in May 2004. Ofcom considers the lack of competitive constraints on BT’s ability to raise its gross and hence net termination charges (i.e. the inability of competing TCPs and service provider customers to respond to BT raising its gross and hence net termination charges) to be more relevant to its assessment of BT’s position in the relevant market, as set out at paragraph 5.24 et seq below. 5.19 For completeness, Ofcom also considered possible differences between volume- and revenue-based market shares. 5.20 According to the Commission’s Notice on market definition: “As a rule of thumb, both volume sales and value sales provide useful information. In cases of differentiated products, sales in value and their associated market share will usually be considered to better reflect the relative position and strength of each supplier.”152 5.21 152 76 Calculating revenue shares for NTS call termination/hosting is a potentially difficult exercise, and figures calculated in this way must be interpreted with care. As explained in paragraph4.99 above, the chief mechanism by which competitive pressure is exerted on TCPs is via the actions of NTS service providers. This means that, for revenue market shares to provide a better reflection of the competitive Commission Notice on market definition (see footnote 49), paragraph 54. position of a TCP than volume market shares, they would have to be based on the net revenues earned by TCPs, i.e. the sum of all gross charges less revenue share payments. 5.22 Comparisons between the gross revenues earned by different TCPs might be misleading, for a number of reasons. First, different TCPs have different time of day profiles (i.e. the balance between the proportion of daytime, evening and weekend calls that they terminate). As daytime charges for termination are higher than charges for termination at other times, operators with a relatively high proportion of daytime calls will inevitably tend to have a revenue share above their volume share, irrespective of market power. Over the period of the conduct in question, NCCN 500 had a particularly significant effect on the gross revenues earned on daytime calls: BT’s gross termination charges for daytime weekday traffic were over [] times higher than for the next highest category, weekend traffic (see Annex 1). Second, different TCPs terminate different proportions of calls to each NTS number range, and both gross termination charges and average revenue share payments vary significantly by number range. 5.23 On this basis, Ofcom has decided that, given the particular circumstances of NTS call termination/hosting, in which gross termination charges provide limited information about the extent of competitive pressures faced by TCPs, it would not be appropriate for Ofcom to solely base its dominance assessment on volume or revenue market shares.153 This is consistent with Ofcom’s approach to assessing BT’s position within the relevant market, which, as set out at paragraph 5.66 et seq below, is a detailed assessment of the constraints faced by BT. Conclusion on BT’s market share 5.24 BT’s share of the NTS call termination/hosting market, based on total volumes, is below a level that would typically be associated with a position of dominance in the relevant market.154 5.25 However, for the reasons set out at paragraph 5.16 above the market shares above are not particularly informative as to whether BT has a dominant position in the market. Rather, it is necessary to analyse the competitive constraints faced by BT in making its increase in the gross termination charge. 5.26 Therefore, as explained in the following section, a detailed assessment of the constraints faced by BT forms the central part of Ofcom’s assessment of BT’s position within the market. Barriers to entry and expansion Barriers to entry 5.27 The threat of competitors entering a market may prevent firms from raising prices above the competitive level. However, if there are significant barriers to entry, this threat is likely to be weak or absent. A dominant firm may then be able to raise 153 []. DG Competition in Discussion paper on the application of Article 82 of the Treaty to exclusionary abuses: Public Consultation, Brussels (2005), paragraph 31, page 11, and as quoted in John Vickers, Market Power in Competition Cases, European Competition Journal Vol.2 special issue, p 13, seems to countenance a single dominance finding with a share as low as 25%. 154 77 prices sustainably without attracting additional competition that would compel it to lower its prices. 5.28 In its guideline Assessment of Market Power, the OFT states that: “Entry barriers are important in the assessment of potential competition. The lower are entry barriers, the more likely it is that potential competition will prevent undertakings already within a market from profitably sustaining prices above competitive levels… An undertaking even with a large market share in a market with very low entry barriers would be unlikely to have market power. However, an undertaking with a large market share in a market protected by significant entry barriers is likely to have market power.”155 5.29 5.30 In these guidelines, the OFT states that “it is useful to distinguish between the following factors which, depending on the circumstances, can contribute to barriers to entry: • sunk costs; • poor access to key inputs and distribution outlets; • regulation; • economies of scale; • network effects; and • exclusionary behaviour.”156 In a footnote, the OFT adds: “Exclusionary behaviour does not refer only to behaviour that raises entry barriers. Exclusionary behaviour also refers to practices that make it harder for existing competitors to become more forceful competitors…”157 5.31 155 Most of the types of barrier to entry listed above are relatively low in this market. A firm wishing to enter this market and sell NTS hosting to NTS service providers (and NTS call termination to OCPs), in competition with those sold by TCPs already in the market must: • invest in obtaining access to the necessary network infrastructure, including switching equipment; • obtain NTS number ranges from Ofcom; • invest in any additional hardware and software needed to support particular types of NTS services; • interconnect with other communications network providers, notably BT; and OFT 415, paragraph 5.7. OFT 415, paragraphs 5.6-5.28. 157 OFT 415, footnote 24. 156 78 • market its hosting services to NTS service providers.158 5.32 Ofcom considers that the sunk entry costs and economies of scale associated with these activities are unlikely to be prohibitive. As set out in the market review consultation, the technology used to offer NTS is well established and is available in entry-level PC based switches.159 TCPs do not therefore need to build networks out to a large number of end users in order to enter this market, and typically interconnect only with BT.160 5.33 In practice, NTS service providers employing the technology described in the preceding paragraph appear to specialise in the provision of NTS call termination/hosting for voice calls only.161 Termination of metered data calls appears to be provided largely by TCPs with more extensive networks (this is necessarily the case for unmetered data call termination, which does not form part of the relevant market) and this may indicate that somewhat greater sunk costs have to be incurred in order to offer data NTS call termination. 5.34 However, even notwithstanding that entry barriers in the wider NTS call termination/hosting market appear to be low this does not mean that BT faced significant extra constraints from competitors as a result of NCCN 500. This is because of the significant barriers to expansion faced by other TCPs with respect to the NTS traffic terminated by BT, which are discussed in the following sections. Barriers to expansion 5.35 According to the OFT: “New entry is not simply about introducing a new product to the market. To be an effective competitive constraint, a new entrant must be able to attain a large enough scale to have a competitive impact on undertakings already in the market. This may entail entry on a small scale, followed by growth. Barriers to entry are closely related to barriers to expansion and can be analysed in a similar way. Many of the factors discussed above that may make entry harder might also make it harder for undertakings that have recently entered the market to expand their market shares and hence their competitive impact.”162 5.36 Ofcom believes that service providers are likely to be price sensitive and willing to switch TCP in response to price differentials. As with barriers to entry, most potential barriers to expansion appear low. However, the cost of establishing extensive interconnection with BT may represent a barrier to expansion from the perspective of very small players. More importantly, as discussed below, BT’s ability to offer prices which other TCPs cannot match represents the key barrier to expansion in the NTS call termination/hosting market. 158 NTS call termination market review consultation, paragraph 4.19. NTS call termination market review consultation, paragraph 4.17 et seq. 160 NTS call termination market review consultation, paragraph 4.20. 161 Ofcom approached several NTS call providers who indicated that typically smaller operators tended to supply voice rather then NTS data call services. Operators approached by Ofcom included [] (telephone meeting of 23 May 2006), [] (telephone meeting of 12 May 2006) and [] (meeting of 15 May 2006). 162 OFT 415, paragraph 5.37. 159 79 Constraints on BT from competing suppliers of NTS termination/hosting 5.37 In its complaint, C&W suggested that the price increases notified in NCCN 500 constituted prima facie evidence that BT has market power in setting its own NTS call termination rates, unconstrained by competitive pressures from any source.163 In a competitive market, it would be unusual for an undertaking to be able to profitably increase its charges to a level that was over 20% (in the case of 0845 calls) above that of its competitors where it was providing a similar service. The remainder of this section therefore considers this suggestion made by C&W in its complaint in more detail. 5.38 The following discussion first identifies how, in the absence of the regulatory and other economic factors present in this market, TCPs could, in principle, constrain BT’s price increase (paragraphs 5.39-5.40 below). The subsequent paragraphs then go on to explain why in practice, this mechanism is not available to TCPs, and that TCPs are therefore unable to constrain BT’s prices, due to the presence of regulation and other economic factors (see paragraphs 5.42-5.49 below). 5.39 A profitable increase in the gross (and net) termination charge by BT would, in principle, attract entry (or expansion) by other TCPs who could match the increase and pay slightly more to service providers who would then switch to the new TCP entrant in response. This mechanism would return the net termination charge to the competitive level and therefore TCPs could undermine BT’s attempt to raise gross (and net) termination charges. 5.40 Importantly, the ability to fund higher outpayments to service provider customers (who select the terminating network) arises initially from the TCP raising its gross termination charge to OCP customers (on the termination side of the market). In the absence of factors such as SMP in call origination (see paragraph 2.23 above), the higher gross termination charge leaves OCPs with a lower call origination payment (i.e. lower retention) unless the OCP raises its retail prices (which would make it less competitive compared to other OCPs). As illustrated at Figure 9 below, and as discussed at paragraph 5.77 et seq below, the OCP will not switch to another TCP unless the service providers were also to switch, since call origination revenues could be at risk. 5.41 The following paragraphs consider whether in practice non-BT TCPs can influence the level of the gross termination charge. 5.42 NCCN 500 resulted in an increase of up to about 25% in BT’s gross NTS call termination charges to non-BT OCPs (see Figure 3 above), with corresponding increases in the net termination charge (except to the extent that it was offset by increases in payments to NTS service providers). However, as will be seen at the conclusion of this section, this did not cause BT to lose market share. In other words, despite the apparent absence of barriers to entry in this market, new entrants were not able to discipline BT by capturing some of its share of the market, i.e. they faced barriers to expansion. 5.43 When BT increased its NTS call termination charges to OCPs via NCCN 500, this increase did not appear to provide BT’s NTS service provider customers with an incentive to switch away from BT’s NTS hosting services to those of other firms (see paragraph 5.77 et seq below). 163 80 C&W submission to Ofcom of 15 March 2005, paragraph 7.23. 5.44 The main immediate impact of the price increases notified in NCCN 500 was on OCPs (as it took the form of an increase in the gross termination charge), rather than on NTS service providers (who select the terminating network). Therefore in this section we consider possible responses by OCPs first, before going on to consider responses on the other side of the market. 5.45 Ofcom does not consider, based on evidence gathered from BT’s competitors (see Section 3), that any OCP passed on the increase in a way that would encourage substitution into calls to competing NTS service providers.164 Given this, it is highly likely that, relative to a counterfactual in which BT had not increased its charges for NTS call termination, NTS service providers continued to receive the same volume of traffic and the same per minute revenue for this traffic (or possibly more, to the extent that BT passed on some of the higher termination charge in higher revenue share payments). 5.46 In addition, other TCPs were unable to match BT’s increase in the gross termination charge. As discussed at paragraph 2.14 et seq above, this is because of the interaction of BT’s dominance in call origination and the fixed relationship between BT’s retail NTS prices and geographic call prices. The maximum retail price of a call by a BT customer to an NTS number is determined by the NTNP, which fixes the relationship between BT’s retail NTS prices and geographic call prices. The maximum amount that BT, as OCP, can then retain out of the retail charge is limited by the NTS call origination condition (i.e. BT’s retention must be cost-based) and, given BT’s dominance in origination, it is able to charge (no less than) this amount. The amount that is passed through to the TCP is therefore equivalent to the retail charge minus BT’s retention, minus any payments to a transit provider, all of which are outside the control of the TCP. 5.47 In the absence of these factors, a profitable increase in the gross (and net) termination charge by BT would attract entry (or expansion) by other TCPs who could match the increase and pay slightly more to service providers, who would then switch to the new entrant in response. This mechanism would return the net termination charge to the competitive level. However, because competing TCPs can’t match BT’s higher gross termination charge in practice, and therefore cannot fund a strategy of offering higher outpayments in a competitive market, TCPs will not be able to attract service providers away from BT without incurring losses.165 5.48 The absence of barriers to entry to the NTS call termination/hosting market would be a relevant factor if it enabled new entrants to capture market share from BT, i.e. by reducing the volumes of NTS traffic that BT terminates. 164 ntl response to section 26 Notice of 28 June 2005; BT Response of 25 May 2005 to Ofcom section 26 information request of 19 May 2005; “BT interconnect change 20040608.ppt”, PowerPoint presentation forming part of ntl response to section 26 Notice of 28 June 2005; C&W response to section 26 Notice of 28 June 2005; Centrica response to section 26 Notice of 28 June 2005; Energis response to section 26 Notice of 28 June 2005; Telewest response to section 26 Notice of 28 June 2005. 165 As set out at paragraph 4.29, NTS call termination/hosting is sold directly by TCPs to NTS service providers, and via resellers that act as intermediaries between TCPs and NTS service providers. Resellers therefore both compete with TCPs in the NTS call termination/hosting market, and provide them with a channel to market. As with TCPs, resellers’ prices for NTS hosting services are constrained by the possibility of switching by NTS service providers. Since resellers are dependent on the TCP to supply NTS call termination/hosting, the extent of any competitive pressure they are able to apply to TCP charges is limited, perhaps similar to that of large NTS service providers, discussed at paragraphs 5.129-5.132. 81 5.49 In practice, for the reasons given above, BT’s competitors face significant barriers to expansion in this market. The demand faced by BT as a TCP with regard to its termination charges was therefore very inelastic. 5.50 The following section now considers some specific cases that are relevant to the questions as to whether TCPs can influence the level of their gross termination charges. These include: a) BT-originated traffic; and b) traffic transited by BT. Special cases: BT-originated traffic 5.51 BT originates a high proportion ([]) of calls to NTS numbers.166 Ofcom has designated BT as having SMP in the wholesale call origination market.167 Since 1996, BT and other communications providers have been subject to a regulatory regime for NTS which is designed to encourage innovation in the provision of value added services despite BT’s position of dominance in wholesale call origination (see Section 2). 5.52 TCPs other than BT are unable to set their own termination charges for NTS calls originating on BT’s network, because of a combination of the weakness of their bargaining position vis-à-vis BT (which is a result of BT’s dominant position in call origination) and the relationship between BT’s retail NTS prices and geographic call prices.168 In order for other TCPs’ termination charges to increase, BT would have to accept less for call origination than is permitted by the relevant charge control condition. BT’s dominance in call origination allows it to refuse any such proposals (and BT has actually done so where requested by TCPs).169 5.53 The relevant conditions are that: 166 • BT’s retail prices for calls to 0845 and 0870 numbers are linked to its standard local and national call charges, which are both part of the basket of retail services that was subject to the Retail Charge Control while NCCN 500 was in force; and • the NTS call origination condition (see paragraph 2.17 above) requires that BT’s origination retention is cost-based and allows BT to charge up to a limit set by the relevant Network Charge Control.170 BT response of 9 September to Ofcom Notice of 25 August 2005. Review of the fixed narrowband wholesale exchange line, call origination, conveyance and transit markets, statement of 26 August 2003 (see footnote 66), as summarised at paragraph S.12 and analysed at Chapters 4, 5 and 6 and most recently, Review of BT’s network charge controls, statement of 18 August 2005 (see footnote 14). 168 It might be argued that 0844 and 0871 calls, where TCPs can choose from a range of different retail price points, are a partial exception to this. But it remains the case that, for a given retail price point, non-BT TCPs are unable to replicate BT’s NCCN 500 behaviour, i.e. increasing termination charges for a given retail price. 169 See email from [] (Easynet) to [] (BT) of 2 April 2004 and subsequent correspondence, supplied by BT in response to Ofcom section 26 Notice of 22 April 2005, numbered CA172, and email from [] (Telewest) to [] (BT) of 13 April 2004 and subsequent correspondence, supplied by BT in response to the same Notice, numbered CA183. 170 The charge control for certain retail services is set out in SMP Condition D1, Review of Fixed Narrowband Retail Services Markets published by Oftel, 28 November 2003. 167 82 5.54 In the absence of the NTS call origination condition, BT could use its position of SMP in wholesale call origination to raise its charges for originating NTS calls, and with retail prices remaining linked to geographic call prices, TCPs’ receipts would be squeezed. 5.55 In summary, other TCPs cannot increase their current termination charges on traffic originated by BT (either directly in terms of the gross termination charge, or in terms of the net charge through lowering outpayment levels to service providers). Special cases: traffic transited by BT 5.56 Ofcom understands that close to 100% of NTS traffic that is both originated and terminated by communications providers other than BT is transited via BT.171 This is because few communications providers other than BT interconnect directly with each other. Traffic that is originated by an OCP other than BT and terminated by a TCP other than BT is therefore generally passed by the OCP to BT, which then passes it to the TCP. Such traffic is said to ‘transit’ BT’s network. 5.57 Where both the OCP and the TCP are connected to the same BT switch, the service provided by BT is referred to as ‘single transit’. Where the OCP and TCP are connected at different BT switches and conveyance over BT’s network is required, the service is referred to as ‘inter-tandem transit’. 5.58 In theory, competing network operators could interconnect directly with each other instead of using BT as a transit provider. However, there is a fixed cost associated with interconnecting at each interconnection point, which means it is generally uneconomic for competing network operators to establish direct interconnection on a large scale for the relatively small volumes of traffic involved (given that BT has by far the largest share of the wholesale call origination market). This is a contributory factor to BT’s position in the market for single transit where Ofcom has designated BT as having SMP, for the reasons set out in Annex 4 (these reasons include BT’s ubiquitous access network and BT as the only provider that provides single transit to any notable extent, meaning that its share of the market is close to 100%). As a result of BT’s SMP in the market for single transit, BT is required to offer single transit to other communications providers at regulated charges. 5.59 In contrast to single tandem transit, inter-tandem transit is the service a communications provider supplies to others to enable them to convey calls between its tandem exchanges when a call originates and terminates on networks other than its own. In August 2005, Ofcom set out the view that there is a single market for inter-tandem conveyance and inter-tandem transit on fixed public narrowband networks in the UK excluding the Hull area and that this market is effectively competitive (see Annex 4).172 This finding arose because the number of providers with a high level of connectivity to BT’s tandem exchanges means that there are hardly any routes where fewer than three communications providers are present and allows single transit to replace inter-tandem transit to a significant extent as many providers are connected to BT at the same tandem exchange. 5.60 TCPs are unable to set differential gross termination charges for: 171 NTS calls that are transited by BT between non- BT OCPs and TCPs account for around []% of all NTS call traffic. 172 See Review of BT’s network charge controls, statement of 18 August 2005 (see footnote 14). 83 5.61 • calls to 0845 and 0870 numbers that are transited (but neither originated nor terminated) by BT; and • calls originated by BT and terminated by another TCP. This is because BT bills other OCPs at the same rate irrespective of the identity of the TCP and pays other TCPs at the same rate irrespective of the identity of the OCP. Figure 2 above (repeated as Figure 8 below) shows that (excepting NCCN 500), when BT functions as a transit provider, OCPs pay the same amount to BT as BT does to TCPs when it originates a call. It subsequently bills separately for transit. Payment arrangements for transit depend on number range. The OCP pays for transit for calls to number ranges starting 0844 and 0871, whereas the TCP pays for transit for calls to number ranges starting 0845 and 0870.173 Figure 8: NTS transit arrangements D = BT’s discounted retail price C = BT’s retention D-C P OCP D-C BT T Originator pays transit (for 0844/0871) TCP NTS SP T Terminator pays transit (for 0845/0870/PRS) P = OCP’s retail price T = BT’s transit charge 5.62 The existence of this arrangement, together with low levels of direct interconnection between communications providers other than BT, makes it difficult for competing network providers to broker bilateral agreements for gross termination charges.174 In most cases, either the TCP or the OCP will lose out by not transiting traffic via BT. When providing service to OCPs other than BT, TCPs need to make these OCPs an offer that is at least as attractive as the one that the OCP can get if it uses BT as a transit provider, i.e. an amount that is no more than the OCP would have to pay to BT as a transit provider ((D-C) in the diagram above). This is also the amount that BT will pay the TCP if the OCP chose to route the calls to BT for onward transit to the TCP. The margin that the TCP can earn over and above the payment that BT will make to it is therefore restricted to being equal in size to the transit payment. 5.63 This means that transiting via BT is the norm for NTS calls neither originated nor terminated by BT (close to 100%). Of NTS traffic transited via BT, in 2004/05 []% was conveyed via single transit and []% by “double tandem” (i.e. inter-tandem) transit.175 While Ofcom has determined that BT no longer has SMP in the market for inter-tandem transit, BT continues, at the time of writing, to provide this service at the regulated charges that applied while it was designated as having SMP, which gives 173 These payment arrangements are as set by Ofcom in paragraphs 3.52–3.56 of the NTS consultation. 174 See Review of BT’s network charge controls, statement of 18 August 2005 (see footnote 14)., Annex 5. 175 Source: BT response of 13 October 2005 to Ofcom section 26 Notice of 11 October 2005. 84 OCPs and TCPs the incentive to transit traffic via BT. Single transit, in which BT has SMP, is also available at controlled charges. Initial conclusion: traffic originated and transited by BT 5.64 TCPs other than BT are constrained in the gross charges they can levy on OCPs for NTS call termination, as a result of BT’s dominance in the market for call origination, commercial incentives (that do not support the establishment of direct interconnection between altnets), and the way in which non-BT TCPs bill BT for calls that transit via BT’s network (which does not identify the OCP). TCPs other than BT are therefore in effect unable to increase gross charges above their current level. This has implications for the constraints that BT faces in terminating traffic that is originated by other OCPs (see paragraph 5.40 et seq above). Constraints that apply in the case of BT-terminated traffic 5.65 On balance, Ofcom’s conclusion is that BT’s ability to increase prices, such as its NCCN 500 charge increases, was not constrained by competition between TCPs for NTS service providers. Constraints on BT from customers 5.66 This section sets out Ofcom’s analysis of the likely sources of competitive pressure on TCPs (including BT) from customers on both sides of the market for NTS call termination/hosting in the UK, i.e. OCPs and NTS service providers. 5.67 The focus of this section is on the constraints faced by BT as a TCP when it increases its net charges via an increase in gross termination charges, because this analysis goes directly to answering the question as to why BT can act independently with respect to price. Since, however, the demand for NTS call termination/hosting is two-sided, Ofcom’s analysis considers constraints from both sides of the market, i.e. both on OCPs/callers and on NTS service providers. OCP responses to a price increase by BT 5.68 The responses potentially available to OCPs following NCCN 500 were: • to refuse to originate calls destined for BT-terminated NTS numbers; • to pass the price increase on to their own retail customers; and • to absorb the increased cost. 5.69 These strategies are summarised in Figure 9 below and then considered in turn in the following paragraphs. 5.70 The starting position for this analysis is that BT, in its capacity as a TCP, increases its gross termination charge to other OCPs. The responses available to OCPs and the consequent net impact on BT may be contingent on the OCP’s status as a TCP, or BT’s status as an OCP. This means that, for each of the rows in Figure 9 corresponding to the options available to the OCP (which may be a vertically integrated provider operating both in origination and termination), the impact on both BT and the other communications provider (as both an OCP and a TCP) is considered. 85 5.71 86 The arrows in Figure 9 indicate the direction in which the available revenue per minute earned from NTS traffic (net of gross termination payments to BT in the case of the OCP), and volumes, would move, relative to a counterfactual in which BT had not increased its termination prices. A question mark indicates a degree of uncertainty as to the consequences of the response in question. Figure 9: Options available to OCP following a termination price increase by BT Impact OCPs action Altnet (A) (OCP) Altnet (A) (TCP) BT (OCP) Traffic volume Traffic volume Traffic volume ppm margin ppm margin BT (TCP) ppm margin Traffic volume Impact ppm margin (a) absorb increased cost (b) Refuse to carry traffic (?) (c) Pass price increase on to own retail callers (?) (?) (d) Increase own terminatio n rates (i) Altnet (A) (OCP) is worse off (ii) BT (TCP) is better off (i) Altnet (A) (OCP) is worse off (ii) BT (OCP) may gain (iii) BT (TCP) is worse off (i) Altnet (A) (OCP) may lose (ii) BT (OCP) may gain (iii) BT (TCP) may gain or lose This response is not feasible for the reasons given in paragraph 5.64. 5.72 Each of the above options, which are available to other communications providers following an increase in BT’s charges for NTS call termination/hosting, are examined in further detail in the following sections. 5.73 The purpose of this analysis is to identify the constraints faced by BT when it raises its NTS call termination/hosting charges. Given the factors described in paragraphs 5.64 et seq above, the fourth option – TCPs increase their own NTS call termination/hosting charges – is not feasible and so is not discussed in any detail. 5.74 Each of the options available to BT’s competitors is examined to see whether the pursuit of that option would have placed a constraint on BT (i.e. would make its price increase unprofitable), and whether it would be in the commercial interests of the communications provider in question. Ofcom’s conclusions are reported at paragraphs 5.132-5.135 below. Absorbing increased costs 5.75 Under this strategy OCPs do not change their retail prices from the levels that prevailed prior to BT’s NTS call termination price increase. 5.76 This strategy, while it may be optimal for OCPs under certain circumstances (i.e. where increasing prices or refusing to connect calls is not a commercially viable 87 strategy) will not serve to constrain BT’s activities under any circumstances, because it gives neither callers nor service providers an incentive to switch to alternatives and so would not reduce the volume of calls received by BT. Therefore BT’s charge increase will be profitable. Refusing to carry BT-terminated traffic 5.77 Under this strategy, OCPs facing BT’s increased NTS call termination charges, refuse to connect calls from its retail customers to NTS service providers that use BT as a TCP. In other words, if its customers dial an NTS number that is terminated on BT’s network, they will not be able to get through (and will hear an appropriate tone or network message). 5.78 BT, as a TCP, would experience a loss of revenue from terminating traffic by OCPs adopting this strategy. 5.79 OCPs (as pure OCPs, that is, before any changes to revenues which the operator may obtain as a TCP, which are considered below) will lose revenue relative to the scenario under which they absorb BT’s increased NTS call termination charges – assuming that there remains a positive margin available on originating such calls – because they will forego the revenue that they would otherwise have earned from their retail customers calling BT-hosted NTS numbers. 5.80 In addition, the basket of services sold to customers by the OCP becomes less attractive to customers. This is because customers are no longer able to access certain services that may form an important proportion of the overall basket of calls and access services that they demand. Based on a simple calculation (dividing total minutes by total lines) Ofcom has calculated that in 2004/05 the average BT customer (business and residential combined) spent just below £5 per quarter on calls to 0845 and 0870 numbers.176 The market share data set out above suggests that around one-third of these calls would be to BT-hosted numbers. 5.81 Customers can react to this development in one of the following ways: 5.82 176 • stay with the same OCP and no longer make calls to BT-hosted NTS numbers; • stay with the same OCP and switch to calling NTS numbers offered by other NTS service providers that do not use BT as a TCP (some of which may be terminated by the OCP); or • switch to another OCP that allows them to call BT-hosted NTS numbers – note that the more OCPs adopt the strategy of blocking access to BT-hosted NTS numbers, the more likely it is that callers will switch to BT as an OCP, since connectivity to BT-hosted NTS numbers will be increasingly unavailable elsewhere. The attractiveness for OCPs other than BT of the ‘refuse to connect’ strategy therefore depends on how consumers are likely to respond. Under the third of the customer scenarios listed in the previous paragraph, the OCP is clearly in a worse position, and, potentially, BT gains as an OCP. The greater the proportion of customers selecting this strategy, the less likely the ‘refuse to connect’ response would be to constrain BT’s behaviour. The ‘refuse to connect’ response is more likely Source: BT Wholesale; Ofcom estimate based on Communications Report 2006, para 3.4.2. £5 represents 2.5% of residential telecoms users’ quarterly spend (i.e. £183 a quarter). 88 to place a constraint on BT’s pricing in the first two customer scenarios identified in the preceding paragraph. However, from the perspective of the OCP, this response is unlikely to be optimal. BT will lose revenue, but the OCP’s immediate focus will be on its own profitability, which would tend to be reduced as a result of following this strategy unless, in the second scenario, it can expect to make offsetting gains as a TCP. This possibility is discussed further below. 5.83 Unless and until customers have reacted (in one of the three ways listed in paragraph 5.81 above) to the OCP’s refusal to connect calls to BT-hosted NTS numbers, the OCP will always earn higher profits under the “absorb price increase” option. In fact some customers may not react at all, particularly for calls where there are no immediate substitutes or where the cost of dialling NTS numbers is not material to the overall service package (see paragraph 5.85 below). Then, it will be more profitable to absorb the price increase because, as long as the margin earned on calls to BT-hosted NTS numbers remains positive (that is, the retail price remains above cost), it will be more profitable to carry a positive volume of such calls than a zero volume. This will continue to be the case unless BT’s loss of revenue as a TCP resulting from refusal of other OCPs to connect calls, and pressure applied by BT’s NTS service providers as a consequence, causes it to reverse its price increase. 5.84 Ofcom’s view is that OCPs are unlikely to pursue the ‘refuse to connect’ strategy. It may be optimal for them to do so if a refusal to carry traffic were to persuade BT to reverse its gross price increase. This may happen if the OCP does not lose a significant number of customers as a result of refusing to connect to BT’s NTS numbers (and BT does not therefore gain at the origination level). OCPs will lose fewer customers where calls to NTS numbers that BT terminates are not a vital part of their retail offering to customers. This will be the case where the OCP’s customers are willing to switch to calling NTS numbers terminated by TCPs other than BT. 5.85 The extent to which customers will be willing to switch to calling numbers terminated by TCPs other than BT will be determined by a range of factors. In particular, customers might not be willing to switch to other NTS numbers in cases where: 5.86 • calls relate to NTS services for which the customer has no immediate substitute, for example after-sales services where the customer has already purchased the primary product, such as telephone banking; and • the choice of service provider (this might, for example, be the caller’s bank) is based largely on aspects of the overall service package other than those relating to NTS calls.177 Ofcom refers to calls where there is no immediate substitute for dialling a particular number and for which responsiveness to changes in relative prices is likely to be low as ‘locked in’, Ofcom’s survey data suggests that the proportion of callers which are ‘locked in’ to calling a particular number range is as follows: • on 0845 numbers: 7% of minutes; and • on 0870 numbers: 50% of minutes.178 177 Although note that, to the extent that they are visible and form a material part of the lifetime cost associated with the product, customers may take such call prices into account when making initial purchasing decisions. 89 5.87 A service that did not enable an OCP’s customers to call around 30% of NTS numbers (i.e. BT’s share of the NTS call termination/hosting market), including ‘locked in’ numbers, would not be attractive to callers. Any OCP that adopted this strategy would therefore lose market share relative to a counterfactual in which it continued to offer a service that enabled customers to call these numbers. 5.88 Consumer research carried out by Ofcom in 2005 can be seen to give weight to the view that OCPs are unlikely to adopt the strategy of refusing to connect customers following a price increase by BT. Seventy-five percent of those with a landline claimed to use ever use 0845 or 0870 numbers, and some 37% of those who ever used 0845/0870 numbers claimed they thought they would change their landline company if they were no longer able to make calls to some 0845 and 0870 numbers. Given the significant margins that are available on selling other call types and access services (see Ofcom's analysis of the margin squeeze allegation in Section 6), this strategy would clearly not be optimal for a communications provider that wanted to maximise profits in either the short or longer term, given the potential for losing existing customers.179 5.89 Ofcom therefore considers it very unlikely that this strategy would be adopted by any OCP in response to an increase in BT’s NTS call termination/hosting prices. This has been borne out by the actual behaviour of OCPs following NCCN 500, with such a strategy not having been adopted by any of BT’s competitors, as far as Ofcom is aware. 5.90 An example of a situation in which OCPs refused to carry calls to a particular number range (where none of their customers were tied in to calling these ranges) is provided by DQ liberalisation in 2003. In its response to the market review consultation, BT used this case to illustrate its argument that OCPs can refuse to connect calls terminated by a particular TCP.180 5.91 The legacy 192 DQ number, which had been controlled by all of the direct access fixed and mobile originators on their own networks, was replaced by a range of 178 These figures are consistent with the classification of “locked-in” services set out in Annex 15 of the NTS statement. 179 Number Translation Services: 0870 & 0845. Consumer responses to possible changes in pricing or availability of services, August 2005. Unpublished research report (carried out for the NTS call termination market review which as noted at paragraph 2.58 was suspended pending completion of this investigation) based on face-to-face interviews carried out by BMRB as part of an omnibus survey. The report is based on interviews with 921 UK adults (aged 15+) with access to a landline phone that could make and receive calls, weighted to ensure a nationally representative picture (the weighted base is therefore 933 respondents). Respondents who ever used 0845 or 0870 numbers were asked: “Imagine that your landline company told you that you would no longer be able to make calls to some 0845 or 0870 numbers. This would mean you might not be able to call your bank, your credit card company, gas supplier, electricity supplier and some shops from your landline at home. If this happened, realistically, which of these things do you think you would do?: (1) (2) (3) (4) (5) (6) (7) I would change my landline company I would change my bank and/or other suppliers that I couldn't call from home any more I would call these numbers from a mobile phone instead I would call these numbers from another landline (e.g. at work or at a friend's) None of these, I just wouldn't call these numbers any more Other (specify) Don't Know” 180 See BT’s response of 7 January 2005 to the NTS market review consultation. 90 competitively provided 118 XXX numbers. Following liberalisation, some OCPs did not allow their end users to access some of the new DQ numbers, for example BT’s new 118 500 DQ service. 5.92 However, in this situation, all callers had to start dialling new numbers to access DQs, and (to varying extents, depending on which network they were calling from) had a wide range of near-identical alternative products to choose from, so that, in effect, none of the new 118 DQ NTS service providers had any ”locked-in” customers. 5.93 Ofcom’s view is that this unique situation means that the DQ liberalisation experience does not provide a useful guide as to whether OCPs would pursue the “refuse to connect” strategy in the context of NTS calls. Notably, calls to many BT-terminated NTS numbers are received by organisations such as retail banks, the customers of which would not view calls to a competitor as being an acceptable substitute once they have decided to bank with a particular company. 5.94 Ofcom concludes that refusing to connect customers to BT-hosted NTS numbers is not an optimal strategy for OCPs, and that refusal to connect would therefore be very unlikely to provide a constraint on BT’s gross NTS call termination charges. Pass price increase on to own retail callers 5.95 Under this strategy, OCPs other than BT increase their retail prices for calls to NTS numbers to reflect the increased input costs that they incur as a result of BT’s increased charges for NTS call termination. 5.96 Since the marginal cost of inputs, including termination payments to BT, will influence OCPs’ retail pricing decisions, a response of this type might be expected from OCPs, absent any other considerations. Although competing retail OCPs typically charge somewhat higher prices than BT for NTS calls (e.g. a comparison of BT’s rates for 0845 at the time of NCCN 500 being issued with those of the two cable companies plus two of the largest indirect access operators shows that the prices of BT’s competitors were at least as high as those of BT, even before BT’s discounts), they compete with BT on retail price in terms of the entire basket of services sold to end users.181 5.97 Under NCCN 500, BT increased its gross termination charges for termination of some NTS calls by up to about 38% depending on number range, call duration, and time of day (see paragraph 2.42 above) The weighted average NCCN 500 increase was 25%. 5.98 There are two different ways in which OCPs could pass increases in gross termination charges on to end users. First, OCPs could pass on the extra cost across a narrow set of calls – for example, if BT increases its termination charge for 0845 and 0870 calls, passing on the price increase across BT-terminated 0845 and 0870 calls only. Alternatively, they could pass on the price increases across a broader set of services, e.g. • all calls to NTS numbers, regardless of whether or not they are terminated by BT; • all calls (including geographic calls and calls to mobiles); or 181 See, for example, BT’s analysis of retail prices for calls to NTS numbers provided with the letter from [] (BT) to [] (Ofcom) of 20 April 2006. 91 • 5.99 a broader set of services, e.g. a mixture of calls and access. For passing on increases in BT’s gross charge for NTS call termination to act as a constraint on BT, it must be the case that OCPs will have the incentive and are technically able to pass on price increases specifically on the affected NTS calls and that, following such increases being passed on, callers will react in such a way that there is a sufficiently large fall in call volumes to make BT’s initial price increase unprofitable. In the next section Ofcom considers what the effects would be if the price increase were passed through. Effect if price increase were passed on 5.100 Facing an increase in the marginal cost of supplying calls to BT-terminated NTS numbers, Ofcom would expect at least some OCPs, if they were acting rationally, to pass on BT’s price increases, since marginal cost (together with marginal revenue) is an important determinant of the optimal price set by firms. 5.101 However, OCPs are not only customers of BT’s TCP business, but also compete with BT at the retail level (i.e. as retail access providers). They might, as a result, be reluctant to pass on BT’s NTS call termination price increases because of a concern that this would undermine their price competitiveness vis-à-vis BT, which has SMP in certain retail markets and whose input costs were not affected by NCCN 500.182 5.102 To assess the reaction of consumers following retail price increases, Ofcom has carried out an analysis that examines the likely switching behaviour of callers following an increase in BT’s NTS call termination prices being passed on. The aim of the analysis was to estimate, based on BT’s NCCN 500 price increase and assuming that an OCP customer changed its retail prices in such a way as to pass on the entire price increase, how price elastic the demand for the OCP’s calls would need to be in order for BT’s price increase not to be profitable.183 5.103 This analysis, which is set out at Annex 3, does not take account of the tendency of retail customers to switch to BT for the whole bundle of retail services if the price of this bundle, when supplied by competing OCPs, goes up. In other words, Ofcom’s analysis looks at the possible effect on call volumes, assuming no customers switch. It is therefore conservative since, if the magnitude of the call-price elasticities alone suggests that BT’s price increase would be profitable, this result could only be strengthened by consideration of such switching. In addition, the analysis should ideally be carried out based on prices at the competitive level (since market power is measured by the ability to raise price above this level). However, to the extent that actual prices are already above the competitive level, the apparent substitutability of other services is likely to be increased (a phenomenon sometimes known as the “cellophane fallacy”, so again a finding that BT’s price increase was likely to have been profitable may be regarded as conservative.184 5.104 As set out in Annex 3, Ofcom has calculated critical values of the price elasticity (above which the price increase would not be profitable in the absence of switching 182 See for example Retail Price Controls: Explanatory Statement, 19 July 2006, published at: http://www.ofcom.org.uk/consult/condocs/retail/statement/. 183 The own price elasticity of demand for a product is a measure of the responsiveness of the demand for that product to changes in its price. The more elastic the demand, the higher (in absolute terms) its responsiveness to price changes, and vice versa. More formally, the own price elasticity is defined as the ratio of the percentage change in quantity demanded to the percentage change in price. 184 OFT 403, paragraph 5.5. 92 to BT for the retail bundle) based on two assumptions for the proportion of the termination charge increase passed on in the retail price, namely: • a “base case”, under which a value of 30% was assigned to this proportion; and • a conservative case, based on a value of 100%. 5.105 Ofcom considers the former case to be the more likely, based on information from other communications providers. The 30% assumption approximates to BT’s share of termination of non-BT originated NTS calls. The OCP is assumed to increase the price of all 0845 calls, but by only 30% of the amount of the increase. 5.106 The critical elasticity in each case was, respectively (the ranges expressed below being derived from the use of different assumptions regarding the demand curve for retail calls): • for the base case, between -5.6 and -6.8; and • for the conservative case, between -1.6 and -2.1. 5.107 The magnitudes of both of these elasticity estimates are fairly high relevant to likely estimates for the own price market elasticity of demand for retail NTS calls.185 The true value of this parameter is likely to be lower in magnitude than -1.0, given the relatively small proportion of callers’ budget accounted for by retail NTS calls (see paragraph 5.80 above). The market elasticity of NTS calls is most relevant if all nonBT OCPs pass on the increase across NTS calls as a whole since the price of calls to BT relative to those to other TCPs then does not change. The relative price of calls to BT then does not change. Given the elasticity estimates, BT would be likely to benefit from NCCN 500 in this case. Moreover, BT will gain an additional benefit to the extent that the total price of the retail bundle supplied by other OCPs rises and induces switching to BT. 5.108 If OCPs were able to pass the increase on the price of calls terminating on BT’s network alone, then the responsiveness of demand to differences in relative prices between TCPs would be more relevant. But estimates of firm-level elasticities of this sort, which reflect customer switching between providers, are not widely available. This makes it difficult for Ofcom to assess whether such estimates are realistic or not. One indicator that is available is the proportion of locked in calls (see paragraph 5.86 above), for which responsiveness to changes in relative prices is likely to be low. 5.109 In practice, other OCPs have been unwilling to pass on price increases on BTterminated calls alone, for reasons discussed further below. For instance, [] (the only OCP to adjust retail prices at all) advised that increasing retail prices in relation to BT terminated traffic only would not be an attractive option due to difficulty communicating changes to customers, confusion (leading to higher call centre costs) and inability to distinguish BT number ranges from other TCPs. This suggests that both for commercial and technical reasons, OCPs would not pass through increases to BT terminated traffic only (see paragraph 5.114 below). 185 Market elasticity, as opposed to own price elasticity (see footnote183) is the change in total market demand in response to a change in price by all firms in the market. This may be contrasted to firmlevel elasticity, i.e. that faced by a particular firm in response to a change in its price alone, which would also reflect switching to competing providers. 93 5.110 Ofcom has considered BT’s position in the wholesale call origination market as well as its position as a TCP. Another important point to note is that, if an OCP had passed on TCP (e.g. BT’s NCCN 500) price increases to its retail customers, leading to increased prices for retail calls to NTS numbers, the OCP’s overall package would have become less attractive to customers relative to BT, and BT may have gained market share and revenue in retail markets and in the wholesale call origination market. 5.111 This suggests that, following an increase in BT’s NTS call termination charges, if all other OCPs pass the increase on to their customers (and, as is argued above, there is no other way in which BT’s conduct as a TCP can be constrained at the origination level), BT’s price competitiveness as a retail operator will have improved vis-à-vis that of its rivals. This is because, as outlined above, other TCPs are unable to respond with termination price increases of their own that could reduce the competitiveness of BT’s overall package of calls and access. So, even if other OCPs are able to lower BT’s revenues as a TCP by passing on price increases to end users, BT’s revenues at the retail level are likely to increase, compensating it to some extent for any losses that it incurs at the TCP level. This further strengthens the finding outlined above, namely that it is unlikely that BT’s behaviour as a TCP will be constrained. Observed behaviour following termination price increases 5.112 The actual response of BT’s competitors to NCCN 500 may not be a reliable guide as to their optimal responses, absent the threat of regulatory intervention, since, while NCCN 500 was in force, BT’s competitors may have been waiting for a response from Ofcom following its commitment to examine the competitive implications of this issue (both in this investigation and in the NTS call termination market review), and did not therefore pass on BT’s price increases. Notwithstanding considerations of this sort, the extent to which BT’s competitors did pass on price increases is discussed in the following paragraphs. 5.113 Ofcom required thirteen communications providers (see Section 3) to provide details of any changes that they had made to their prices for calls to NTS numbers as a result of the price changes notified in NCCN 500. 5.114 Of the thirteen communications providers from whom Ofcom sought information, only [] said that it had changed its prices for calls to some NTS numbers as a result of NCCN 500. [] told Ofcom that following NCCN 500 it had increased its retail prices for all 0845 and 0870 calls i.e. it had not just increased its prices for BT-terminated calls. [] told Ofcom that increasing its retail prices only in relation to BT-terminated traffic would not have been an attractive option, and that it had rejected such a move. Its internal documentation following NCCN 500 made the following comments on such a change: • “Difficult to communicate to customers; • Confusion will drive calls into the call centres; and • System implications - new charge band for BT number range & maintenance.”186 5.115 Ofcom notes that, as reported in confidential internal [] documents, [] considered a range of possible retail pricing actions in response to NCCN 500.187 Ofcom notes 186 94 []. that [] could have chosen to absorb the NCCN 500 price increase like BT’s other competitors, although this would have resulted in an estimated £[] reduction in []’s margins. Ofcom also notes that []’s action to raise retail 0845 and 0870 call prices for all traffic (terminated by both BT and other TCPs) had the effect of raising []’s 0845 retail prices from []% above BT’s prices to []% above BT’s prices, and in doing so raising []’s margins. This could indicate []’s pricing action was aimed at increasing both revenues and margins, and hence went further than the recovery of NCCN 500 prices. 5.116 None of the other communications providers from whom Ofcom sought information said that they had changed their prices for retail calls to NTS numbers as a result of NCCN 500.188 5.117 A number of them commented on their decisions not to change retail pricing, as set out in the following paragraphs. 5.118 [] stated that: “It was decided… that retail call prices to the affected number ranges would not be amended. This was due to BT’s dominant position in the retail calls market.”189 5.119 [] stated that: “As [] currently charge in excess of BT’s rates it was intuitively felt that there was no scope to increase existing charges so as to protect margins.”190 5.120 [] explained that: “[] does change its retail prices regularly and undertakes a price review process called []. However, this process is essentially market-driven; almost invariably, [] shows that it is necessary to change our prices to meet the market’s requirements. Only very occasionally are cost considerations brought into play. This is because []’s position in the market is essentially that of a price follower.”191 5.121 [] stated that it had not adjusted its retail pricing in response to NCCN 500: “in order to maintain our competitive position.”192 5.122 These accounts show that no OCPs – except for [] – passed on BT’s NTS call termination price increases. In the case of [], it was not able to pass on the increase to customers solely in relation to the 0845 and 0870 numbers hosted by BT, but rather passed on the increase to all NTS end customers. A key motivation behind OCPs not passing through the increase to customers appears to have been a desire to maintain a competitive position vis-à-vis other OCPs including BT. 187 Ibid. [] confirmed in its response that it does not have any retail customers and that the question was not therefore applicable. 189 [] response to section 26 Notice of 28 June 2005. 190 [] response to section 26 Notice of 28 June 2005. 191 [] response to section 26 Notice of 28 June 2005. 192 [] response to section 26 Notice of 28 June 2005. 188 95 Conclusion on pass-through 5.123 The preceding discussion shows that: • only one of BT’s OCP customers passed on BT’s termination price increases to end users; and that • even if all OCPs had passed on BT’s termination price increases, it is unlikely that this would have made BT’s termination price increase unprofitable. 5.124 On this basis, Ofcom’s view is that OCPs passing on BT’s termination price increases would not impose a competitive constraint on BT. Evidence on profit effect of NCCN 500 5.125 Ofcom has not been able to compare what happened to BT’s traffic volumes, (and its revenues and profits) following NCCN 500 with a counterfactual in which BT did not issue NCCN 500. However, given the arguments outlined above, and the fact that BT did not lose market share between 2003/04 and 2004/05, Ofcom does not consider that BT suffered any loss of volumes of ([]-originated) traffic that came close to making NCCN 500 anything other than a profitable strategy for BT. 5.126 Changes in some key operational and financial measures for BT are summarised in the tables below. All figures focus only on the traffic type that was affected by NCCN 500, i.e. calls from other OCPs to BT. Table 6: OCP-to-BT 0845 NTS call volumes (millions of minutes)* 2003/04 2004/05 % change 848 908 7% Voice 1,667 1,598 -4% Data 2,515 2,506 -0% Total *Numbers are not rounded and therefore figures may not total exactly Source: Ofcom calculations using volume information supplied by BT used in Ofcom FAC profitability analysis (see Section 6). Table 7: BT's gross NTS call termination charges* for 0845 OCP-originated calls (average ppm)** 2003/04 2004/05 % change 2.16 2.58 19% Voice 1.12 1.38 23% Data 1.47 1.82 23% Weighted averag * Excludes hosting revenues **Numbers are rounded and therefore figures may not calculate exactly Source: Ofcom calculations using volume information supplied by BT used in Ofcom FAC profitability analysis 96 Table 8: BT's gross NTS call termination revenues for 0845 OCP-originated calls* (£m)** 2003/04 2004/05 % change 18.35 23.40 28% Voice 18.75 22.12 18% Data 37.10 45.52 23% Total * Excludes hosting revenues **Numbers are rounded and therefore figures may not total exactly Source: Ofcom calculations using volume information supplied by BT used in Ofcom FAC profitability analysis Table 9: BT's net 0845 termination charges for 0845 OCP* (average ppm)** 2003/04 2004/05 % change 3.98 3.49 -12% Voice 0.70 0.91 30% Data 1.81 1.85 2% Weighted averag * Excludes hosting revenues **Numbers are rounded and therefore figures may not total exactly Source: Ofcom calculations using volume information supplied by BT used in Ofcom FAC profitability analysis 5.127 The figures above show that, following NCCN 500; • BT’s gross NTS call termination charges increased. BT’s net charge increased overall (although it fell for voice call termination, possibly reflecting reduced payments from NTS service providers); and • while there was some decline in BT’s volumes for data, this was more than offset by the increase in BT’s NTS call termination charges. Moreover, given that no OCP passed the increase in termination charges on in higher retail prices for BTterminated calls (in one case – [] – retail prices were increased, but irrespective of the identity of the terminating network), it seems likely that the reduction in data volumes reflected general market trends. 5.128 In addition, internal BT documents suggest that it believed that NCCN 500 would be profitable for it. It appears to have estimated the likely increase in profit from NCCN 500 as approximately £[] (see paragraph 6.556 below for more details).193 Constraints imposed by NTS service providers 5.129 Ofcom has also considered the possibility that competition to attract NTS service providers might constrain BT’s ability to increase its charges for NTS call termination/hosting by forcing it to pass on a higher share of the (gross) termination revenues earned by BT as a TCP. This is an important consideration in this market, since, if, following an increase in gross charges, BT’s net charges (i.e. the amount retained by BT) were constrained by competition to attract NTS service providers, this might be interpreted as suggesting that BT did not hold a dominant position. 5.130 Ofcom considers that competition to attract NTS service providers is unlikely to constrain BT’s ability to set prices above the competitive level. If BT as a TCP were to increase its termination prices and OCPs were to absorb this price increase, then: 193 Briefing paper about the first complaint, 12 May 2004, provided by BT in response to section 26 Notice of 22 April 2005 numbered 1.11. 97 • NTS service providers would be receiving the same volume of minutes as had previously been the case; and • the inability of other TCPs to respond by increasing their gross NTS call termination charges would mean that NTS service providers would not be able to get a better deal elsewhere, and hence would not be able to constrain BT. 5.131 If BT as a TCP were to increase its gross NTS call termination charges, and this price increase was passed on by other OCPs (causing BT to lose volumes as a TCP), then, if NTS service providers were to receive the same per-minute payment from BT, they would be worse off than was previously the case. In such a case, they might consider establishing relationships with other TCPs that would charge lower termination prices to OCPs, thereby guaranteeing larger volumes and revenues for the NTS service provider. However, provided that BT’s price increase led to higher gross termination revenues, BT would be able to offer the NTS service provider a larger absolute amount of revenue (even if volumes declined) and retain more revenue for itself.194 Conclusion on the constraints faced by BT 5.132 Following a gross price increase by BT, a number of strategies were available to its competitors (i.e. other TCPs providing termination/hosting) and OCP and NTS service provider customers. 5.133 Ofcom’s conclusion is that BT’s ability to increase prices, such as its NCCN 500 charge increases, was not constrained by competition between TCPs for NTS service providers. Ofcom found that BT is in a different position to other TCPs in the market because, as discussed in paragraph 5.46 et seq above, other TCPs cannot increase their gross charges for NTS call termination since: 194 • their charges to BT for BT-originated calls to NTS are effectively fixed because of a combination of BT’s dominance in the wholesale call origination market and the fixed relationship between BT’s retail NTS prices and geographic call prices. In the absence of these factors, a profitable increase in the gross (and net) termination charge by BT would attract entry (or expansion) by other TCPs who could match the increase and pay slightly more to service providers, who would then switch to the new entrant in response. This mechanism would return the net termination charge to the competitive level. However, competing TCPs cannot match BT’s higher gross termination charge in practice, and therefore cannot fund a strategy of offering higher outpayments in a competitive market. Therefore competing TCPs will not be able to attract service providers/customers away from BT without incurring losses; • the lack of direct interconnection between communications providers other than BT and BT’s position as an NTS transit provider prevents other TCPs from applying selective increases in termination charges which depend on the identity of the OCP; and • a price increase by BT at the termination level may, uniquely, lead to benefits to BT at the origination level, mitigating any reduction in volumes incurred at the termination level. There is no similar mechanism in place for other TCPs. Note that the critical loss calculation (Annex 3) does not allow for the reduction in volumes to lead to cost savings for BT, other than savings from reduced revenue share payments. 98 5.134 Ofcom’s view is that possible strategies of OCP and service provider customers are not capable of constraining the pricing behaviour of BT as a TCP, and that BT will therefore be able to set a price for NTS call termination that is above the competitive level. In summary: • absorbing termination price increases will not constrain BT’s pricing behaviour (or that of any other TCP); • a refusal to carry BT-terminated NTS traffic is unlikely to be a viable strategy for other OCPs, and as such did not impose a competitive constraint on BT; and • where OCPs passed the increases to end users, there was no material effect. In any case, given the implications for their position in retail calls and access markets, this would not have been seen by most OCPs as a viable strategy. This response did not therefore constrain BT. 5.135 Additionally, the actions of NTS service providers would not constrain BT’s behaviour, since if BT as a TCP were to increase its termination prices and OCPs were to absorb this price increase, then: • NTS service providers would be receiving the same volume of minutes as had previously been the case; and • the inability of other TCPs to respond by increasing their gross NTS call termination charges would mean that NTS service providers would not be able to get a better deal elsewhere, and hence would not be able to constrain BT. 5.136 As discussed further at paragraph 6.545 et seq below, BT appears to have had two aims in increasing its changes for NTS call termination: to generate additional revenue, and to raise awareness of features of the current NTS regulatory framework that BT considered should be changed: “This proposal will help to add force to BT’s view that the regime needs to be overhauled by demonstrating that BT can use the current structure to its advantage, just as other operators have used it to their advantage in the past. The price change will also emphasise that the regime is highly uncertain and is difficult to plan for.”195 5.137 This suggests that BT did not see itself as facing any competitive constraints on its ability to set charges for NTS call termination. NCCN 651 5.138 As set out at paragraph 2.45 et seq above, on 2 November 2005 BT issued NCCN 651, effective 1 January 2006, in which it reduced its gross NTS call termination/hosting charges, bringing them back into line with the charges of the other TCPs. It is clear that this change did not result from competitive pressures felt by BT, since: • as described above, analytical reasoning indicates that NCCN 500 was a profitable strategy for BT and, to the extent that it has been possible to assess, 195 See paper to BT Wholesale Pricing Authorisation Group of 18 March 2004, supplied by BT in response to Ofcom section 26 Notice of 22 April 2005, volume 1, tab 8. 99 this has been borne out by the profitability analysis constructed by Ofcom from information drawn from BT’s regulatory costing systems; • BT has explained that the decision to issue NCCN 651 was taken in part to improve relations between BT and its customers and between BT and Ofcom, offering the following explanation in a paper of September 2005 to BT’s Operations Committee recommending that the price increases notified in NCCN 500 should in effect be reversed: [];196 • the price increase was maintained for 20 months supporting the case that BT’s prices remained in force for a period of time consistent with an undertaking holding a dominant position in a market.197 Responses to draft decision: Ofcom’s provisional conclusions on dominance 5.139 C&W and Virgin Media both state that they support Ofcom’s provisional conclusion that BT is dominant.198 5.140 BT was the only respondent to provide detailed comments on Ofcom’s provisional conclusion that BT was dominant in the market for NTS call termination/hosting for the period under investigation. 5.141 BT states that: “BT’s share of the NTS hosting market, based on total volumes, or between []% and []% for the relevant period, is significantly below a level that would typically be associated with a position of dominance on the relevant market and despite discussion to the contrary [i.e. Ofcom’s arguments at paragraph 5.24 of the draft decision] would generally not be sufficient to be associated with a position of dominance.”199 5.142 BT further notes that its market share did not increase during the period under investigation: “which whilst not conclusive, is indicative of an absence of dominance particularly when BT’s low market share is taken into account.”200 5.143 BT states that it is one of a number of vertically integrated providers capable of using use retail revenues to fund revenue share outpayments in NTS hosting (although it states that there is no evidence the additional revenues it received when NCCN 500 was in force were put to that purpose), and that: 196 Annex 1, BT response to section 26 Notice of 14 November 2005. Note that the SSNIP test used for market definition purposes considers whether a hypothetical monopolist could sustain a price increase of some 5%-10% above the competitive level for a period typically of 12 months. Arguably pricing behaviour that suggested BT could act independently of its competitors and customers in respect of NTS call termination could, by extension, also be taken to imply BT’s pricing behaviour of 20 months could also support a dominance finding. 198 C&W’s comments on the draft decision, page 1 and Virgin Media’s comments on the draft decision, page 1. 199 BT’s comments on the draft decision, paragraph 3.3. 200 BT’s comments on the draft decision, paragraph 3.7. 197 100 “there is no evidence that BT’s competitors on the market for NTS hosting were not able to match BT’s “prices” indeed during the relevant period BT lost or failed to win a number of significant accounts.”201 5.144 BT states that: “in the absence of the regulatory factors present in this market, competing TCPs would in principle constrain BT from raising its prices for call termination above the competitive level.”202 5.145 BT argues that: “TCPs were unable to match BT’s increase in the gross termination charge because of regulation, specifically the NTS call origination condition, and the removal of this condition would allow competing TCPs to match and therefore constrain BT’s behaviour.”203 5.146 BT argues that, contrary to Ofcom’s assertion (see paragraph 5.52 above), that TCPs are unable to set their own termination charges because of their relatively weak bargaining position vis-à-vis BT: “the reason that TCPs are not able to set their own termination charges is the NTS call origination condition.”204 5.147 BT argues that: “innovation in the provision of value added services has flourished and, for the relevant period, there was (and is) effective competition for the provision of NTS hosting.”205 5.148 BT states that in the absence of the NTS call origination condition BT would not, as Ofcom claims, be able to exploit its position of SMP in wholesale call origination to raise charges for originating NTS calls, in fact it would have been “obliged to negotiate with TCPs and ultimately to purchase [NTS] termination” at prices set by those other TCPs.206 5.149 BT argues that Ofcom is wrong to conclude that the reason most traffic not originated or terminated by BT is transited by BT was that few communications providers other than BT interconnect directly with each other. BT states that the real reason most traffic transits BT’s network is that: “OCPs were able to use the transit of calls over BT’s network combined with the effect of the NTS call origination condition to avoid having to negotiate and agree termination charges directly with TCPs which may have been significantly more than the termination charge charged to BT.”207 201 BT’s comments on the draft decision, paragraph 3.8. BT’s comments on the draft decision, paragraph 3.11. 203 BT’s comments on the draft decision, paragraph 3.14. 204 BT’s comments on the draft decision, paragraph 3.17. 205 BT’s comments on the draft decision, paragraph 3.5. 206 BT’s comments on the draft decision, paragraph 3.20. 207 BT’s comments on the draft decision, paragraph 3.25. 202 101 5.150 BT argues that: “the NTS call origination condition together with the nature of BT’s and other communication providers’ networks, was such as to provide an absolute constraint on the pricing behaviour of all TCPs, including BT during the relevant period.”208 5.151 BT argues that, contrary to Ofcom’s assumption, “An OCP may […] discipline BT and constrain BT’s pricing behaviour for NTS call termination by selectively refusing to connect calls to only those NTS numbers terminated by BT for which there is a ready substitute NTS not terminated by BT.”209 5.152 BT argues that this is precisely what happened following the proliferation of DQ services after the 118 range was made available.210 BT states that Ofcom has relied on OCPs’ comments for its assertion that refusing to connect calls would not be a viable strategy. 5.153 BT argues that Ofcom is incorrect when it concludes that even if OCPs had passed on BT’s termination charge increases, this would have been unlikely to make this an unprofitable strategy for BT. BT argues that: “BT’s pricing for the relevant period was explicitly designed to avoid retail price changes by OCPs…BT deliberately set the relevant charges to avoid OCPs having to pass through (either in part or in full) the increased termination charge by increasing their retail prices.”211 5.154 BT argues that it would not have been in its interests for other OCPs to raise their retail prices in response, as this would have had a negative impact on the business of NTS service providers. BT states that: “The fact that BT carefully and diligently considered and balanced these issues in setting the charge for the relevant period is clear evidence that BT felt and was constrained in fixing the termination charge…”212 5.155 BT states that it set the charges notified in NCCN 500 at a level such that its competitors would not be placed at a competitive disadvantage in absorbing the increase: “BT took explicit account of the commercial drivers of OCPs, setting the charges at a level that would not cause them either to cease buying NTS termination from BT, or that would require OCPs to increase their retail prices…such that there would be a consequent impact on BT’s volumes”.213 208 BT’s comments on the draft decision, paragraph 3.28. BT’s comments on the draft decision, paragraph 3.33. 210 BT’s comments on the draft decision, paragraph 3.34. 211 BT’s comments on the draft decision, paragraph 3.43. 212 BT’s comments on the draft decision, paragraph 3.48. 213 BT’s comments on the draft decision, paragraph 3.53. 209 102 Ofcom’s response 5.156 Ofcom acknowledges that BT’s market share is “significantly below a level that would typically be associated with a position of dominance” (see paragraph 5.24 above). However, market share, for the reasons discussed at paragraphs 5.16 et seq above, is not a reliable indicator of dominance in this case. Ofcom has therefore based its analysis on the strategies available to BT’s competitors following NCCN 500, and concluded on this basis that that suggests that BT has the ability to behave independently of its customers, competitors and ultimately of consumers. 5.157 Ofcom considers that the fact that BT’s market share did not increase during the period under investigation is consistent with Ofcom’s conclusion that BT’s conduct did not constitute an abuse of its dominant position. Rather, BT used its dominant position to set its prices above those of its competitors, which enabled BT to make additional profits, but did not impose a margin squeeze on its competitors. Ofcom notes that the fact that BT's market share did not decline, despite BT pricing significantly above other operators, is consistent with dominance. 5.158 BT suggests that those of its competitors that are vertically integrated could have used their retail revenues to fund revenue share outpayments in NTS hosting. However, not all BT’s competitors are vertically integrated. Moreover, of those that are, it seems likely that most of the NTS calls they originate will terminate on either BT or another TCP and be transited via BT, incurring the higher termination charges that applied under NCCN 500. 5.159 Ofcom agrees that the NTS call origination condition is to some extent relevant to its consideration of competitive constraints in the NTS call termination and hosting market, in that it requires BT to supply NTS call origination to all TCPs. The NTS charge control (as referred to at paragraph 2.18 above) does not, in itself, prevent BT’s competitors from raising their NTS call termination charges, but places a ceiling on BT's charges for call origination – there is nothing to stop other OCPs demanding to pay less, other than BT's market power in call origination. 5.160 However, it is not relevant in this case (where Ofcom is considering BT’s actual past conduct) to speculate about what market conditions might prevail in the absence of the NTS call origination condition or other regulation. The consideration of BT’s position in the market for NTS call termination/hosting, for the purposes of this investigation, has to be assessed in the light of the regulatory conditions prevailing at the time, i.e. the NTS call origination condition. To that extent, speculation as to what market conditions might have been in the absence of such regulation is not relevant in determining the question of dominance in this matter. 5.161 BT submits that there is “effective competition” in the provision of NTS hosting. Ofcom acknowledges that, as shown by the results of its evidence gathering (see paragraphs 6.568 et seq below) there is competition for NTS hosting contracts. However, as set out at paragraphs 5.132-5.137 above, the behaviour of NTS service providers does not act as a constraint on BT’s ability to behave independently of its competitors. 5.162 BT submits that refusal by OCPs to connect NTS calls to BT would be a viable strategy, as evidenced by OCPs’ conduct following the introduction of the 118 range. Ofcom addresses the relevance of the DQ case to this case at paragraphs 5.90-5.94 above. 103 5.163 BT argues that the charges notified in NCCN 500 were set at such a level that BT’s competitors would not have been placed at a competitive disadvantage were they to have absorbed the additional cost imposed by NCCN 500. As set out in Section 6 below, the appropriate way to assess whether there is a margin squeeze is by reference to the retail prices of the firm allegedly conducting the margin squeeze, not those of its rivals. C&W’s allegation of a margin squeeze in call origination was assessed using BT's retail prices, rather than those of its competitors. Ofcom considers that the appropriate scope for the retail margin squeeze test is a broader product bundle than calls to NTS numbers alone. Conclusion on BT’s position in the relevant market 5.164 Ofcom concludes that BT was dominant in the market for NTS call termination/hosting in the UK between 1 May 2004 and 31 December 2005, the period over which the investigation has taken place. 5.165 BT’s ability to act independently of other TCPs reflects BT’s dominance: (a) in the market for call origination which, combined with the NTNP requirements on retail prices, prevents other TCPs from raising their gross termination charges on BT originated calls to NTS numbers; and (b) in the market for single transit which prevents other TCPs from raising their gross termination charges on non-BT originated calls to NTS numbers 5.166 Based on the evidence set out above, Ofcom’s has found that the particular circumstances of this market means that there are minimal constraints on BT’s ability to act independently of its customers and competitors. 104 Section 6 6 The conduct Introduction 6.1 Having concluded that BT was dominant in a relevant market over the period that NCCN 500 was in force, Ofcom assessed whether BT’s conduct in imposing the prices notified in NCCN 500 amounted to an abuse of BT’s dominant position, in breach of the Chapter II prohibition and Article 82. 6.2 Whether a particular type of conduct constitutes an abuse will be a question of fact for every case, as explained by the ECJ: “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.”214 6.3 A dominant firm is subject to a “special responsibility not to allow its conduct to impair genuine undistorted competition”.215 The “special responsibility” will be increased where a firm is super-dominant or near monopoly, although the scope of the special responsibility depends on the specific circumstances of each case.216 Superdominance is not relevant in this case. 6.4 Ofcom’s analysis proceeded according to C&W’s allegations and focused on C&W’s three principal allegations which were: • that NCCN 500 constituted anti-competitive price discrimination by BT; • that NCCN 500 led to a margin squeeze on competing originating providers; and • that the prices notified in NCCN 500 were excessive. 6.5 In addition to the alleged margin squeeze in call origination, Ofcom considered whether NCCN 500 imposed a margin squeeze on BT’s competitors in the provision of NTS hosting, in that the additional cost imposed by NCCN 500 prevented them from competing with BT in the provision of NTS hosting services. 6.6 Ofcom also considered C&W’s further allegations that NCCN 500 constituted an abuse of BT’s dominant position in that it: • increased competing operators’ costs or constrained them to offer an inferior service; 214 Case 85/76 Hoffman-La Roche v Commission [1979] ECR 46, paragraph 91. Case 322/81 Michelin v Commission [1983] ECR 3451, paragraph 57 216 Cases C-395 and 396/96P Compagnie Maritime Belge Transports v Commission [2000] ECR I1365, paragraph 114. 215 105 6.7 • increased BT’s market power in NTS call origination and call termination; and • formed part of a concerted strategy to influence the commercial conduct of BT’s competitors and dilute competition. Ofcom’s analysis of C&W’s allegations is set out in this section. Alleged margin squeeze 6.8 Ofcom’s analysis and assessment of C&W’s margin squeeze allegations are considered in the following section. 6.9 As described at paragraph 6.220 et seq below, Ofcom’s margin squeeze analysis in effect addresses C&W’s allegation of price discrimination, as well as the margin squeeze allegation itself. Both Tests 1 and 2 as set out by C&W (see paragraph 6.222 below) are addressed by Ofcom’s margin squeeze analysis, by the hosting margin squeeze test and the retail origination margin squeeze test respectively. Alleged margin squeeze in retail markets 6.10 This section addresses whether NCCN 500 constituted an abuse of BT’s dominant position in NTS call termination/hosting by imposing a margin squeeze on OCPs other than BT, as C&W has alleged. Such a margin squeeze could have the effect of restricting or distorting competition in the relevant retail calls market(s). Alleged margin squeeze in retail markets: the allegation 6.11 C&W submitted that the price increases notified in NCCN 500 constituted an abuse of BT’s dominant position in that they imposed a margin squeeze on OCPs other than BT. 6.12 C&W’s arguments on margin squeeze are set out at paragraphs 8.6-8.15 of its submission to Ofcom of 15 March 2005 and in a letter to the case team of 16 September 2005. 6.13 C&W argued that: “Since an end-to-end NTS service cannot be provided without call origination, call origination is a key input of the kind described in [Oftel’s] Guidelines.”217 6.14 C&W submitted that the effect of NCCN 500 is to impose a margin squeeze on C&W at the origination level.218 6.15 C&W submitted that: “even with the fewest number of possible network elements deployed, BT would not be able to cover its costs of origination.”219 217 C&W’s submission to Ofcom of 15 March 2005, paragraph 8.6-8.7, referring to The Competition Act 1998, The Application in the Telecommunications Sector (OFT 417), published at: http://www.oft.gov.uk/shared_oft/business_leaflets/ca98_guidelines/oft417.pdf. 218 C&W’s submission to Ofcom of 15 March 2005, paragraph 8.8. 219 C&W’s submission to Ofcom of 15 March 2005, paragraph 8.12. See also paragraph 2.27 et seq above. 106 6.16 C&W provided details of BT’s costs (taken from publicly available sources), submitting that this analysis suggested BT was failing to cover its end-to-end costs of carrying NTS calls.220 6.17 Considering the appropriate scope of the margin squeeze test, C&W submitted that: “the margin squeeze test should be applied to the products in respect of which a margin squeeze is alleged.”221 6.18 C&W argued that: “To take an analogous case, if a dominant undertaking were to engage in margin squeeze selectively against particular competitors, that would clearly still be abusive; and, yet, there would be no margin squeeze across the whole market. By extension, it would be illogical to say that a dominant undertaking may engage in margin squeeze in relation to certain products in a market in which it is dominant. It must be right that the margin squeeze test should be applied only to the products which are the subject of the margin squeeze.”222 6.19 C&W concluded, citing British Sugar, that: “the margin squeeze analysis should necessarily be restricted at most to calls which form the subject of NCCN 500”.223 6.20 Commenting on the relationship between market definition and the appropriate scope of the margin squeeze test, C&W submitted that: “there would be no need to consider the scope of the application of the margin squeeze test if the relevant upstream and downstream markets were defined more tightly.” Although it went on to say that: “Having said that, there are clearly some arguments for a more granular market approach to market definition in this case.”224 6.21 Ofcom considers C&W’s arguments in the following section. Alleged margin squeeze in retail markets: legal and economic framework 6.22 A margin squeeze occurs where an undertaking that is active on two vertically related markets and that enjoys upstream dominance fails to maintain a sufficient margin between its prices in the upstream and downstream markets, thereby restricting competition in the downstream market. 220 Table 1 and Table 2 of C&W’s submission to Ofcom of 15 March 2005. Letter from [] (C&W) to [] (Ofcom) of 16 September 2005. 222 Ibid. 223 Commission decision 88/158 (Case No IV/30.178 Napier Brown-British Sugar), OJ L284 19/10/1998 which, according to C&W “makes clear that the margin squeeze test must be applied to the dominant provider’s products, as opposed to any competing upstream inputs”. Letter from [] (C&W) to [] (Ofcom) of 16 September 2005. 224 Letter from [] (C&W) to [] (Ofcom) of 16 September 2005. 221 107 6.23 The EC and the UK competition authorities have long accepted margin squeeze as conduct that can amount to an abuse of a dominant position abuse in contravention of Article 82/Chapter II. The most recent case law on margin squeeze is the 2008 judgment of the CFI in Deutsche Telekom (see paragraphs 6.26 and 6.31 below) and the 2008 judgment of the Court of Appeal in Albion (see paragraph 6.32 below).225 6.24 In National Carbonising Company, the Commission, while rejecting the complaint that was made, accepted that margin squeeze could amount to an abuse:226 “an undertaking which is in a dominant position as regards the production of a raw material…and therefore able to control its price to independent manufacturers of derivatives… and which is itself producing the same derivatives in competition with these manufacturers, may abuse its dominant position if it acts in such a way as to eliminate the competition from these manufacturers in the market for derivatives. From this general principle the…Commission deduced that [the dominant undertaking] may have an obligation to arrange its prices so as to allow a reasonably efficient manufacturer of the derivatives a margin sufficient to enable it to survive in the long term.”227 6.25 The first Commission decision to find margin squeeze as an abuse was Napier Brown.228 The Commission found that British Sugar was dominant in both the upstream (industrial sugar) and downstream (retail sugar) markets, and that the margin between the prices British Sugar charged for industrial sugar and retail sugar was not enough to enable a reasonably efficient competitor to remain in the downstream retail sugar market, since the margin did not cover British Sugar's own repackaging and selling costs. The Commission found that: “The maintaining, by a dominant company, which is dominant in the markets for both a raw material and a corresponding derived product, of a margin between the price which it charges for a raw material to the companies which compete with the dominant company in the production of the derived product and the price which it charges for the derived product, which is insufficient to reflect that dominant company’s own costs of transformation … with the result that competition in the derived product is restricted, is an abuse of a dominant position.”229 6.26 225 In Deutsche Telekom, the Commission concluded that Deutsche Telekom had abused its dominant position in the fixed telecommunications network market by structuring its pricing so that its wholesale customers were charged higher rates for access to the local loop than its retail customers, thereby leaving no margin for its wholesale customers to compete in the downstream retail market. The Commission Case T-271/03 Deutsche Telekom v Commission Judgment of CFI 10 April 2008 and Judgment of the Court of Appeal 22 May 2008 Dwr Cymru Cyfyngedig and Albion Water Limited and Water Services Regulation Authority [2008] EWCA Civ 536. 226 OJ [1976] L36/6. 227 Ibid, paragraph 14. In this case, the Commission had rejected National Carbonising's original complaint. National Carbonising applied to the ECJ for a review of that decision and applied for interim measures. The courts directed the Commission to adopt interim measures while the case was pending before the court which the Commission did. National Carbonising withdrew its application to the ECJ in 1977. 228 OJ 1988 L284/1. 229 Napier Brown, paragraph 66. 108 said that this was a margin squeeze because competitors could never compete for customers in the downstream retail market. The Commission was of the view that the margin squeeze abuse continued even after Deutsche Telekom took action to ensure that prices for wholesale services were lower than retail prices as the difference was still not sufficient to cover Deutsche Telekom’s own costs for the supply of retail services.230 6.27 The Commission’s Telecommunications Access Notice states that a margin squeeze can be demonstrated by showing that the dominant company's own downstream operations could not trade profitably on the basis of the upstream price charged to its competitors by its upstream operating arm or, in appropriate circumstances, by showing that the margin between the price charged to competitors in the downstream market (including the dominant company's own downstream operations, if any) for access and the price which the network operator charges in the downstream market is insufficient to allow a reasonably efficient service provider in the downstream market to obtain a normal profit (unless the dominant company can show that its downstream operation is exceptionally efficient).231 6.28 In the UK, Oftel’s guideline The application of the Competition Act in the telecommunications sector is consistent with the Commission’s Telecommunications Access Notice, stating that: “Where a vertically integrated undertaking is dominant in an upstream market and supplies a key input to undertakings that compete with it in a downstream market, there is scope for it to abuse its dominance in the upstream market. The vertically integrated undertaking could subject its competitors in the downstream market to a price or a margin squeeze by raising the cost of the key inputs and/or by lowering its prices in the downstream market. The integrated undertaking’s total revenue may remain unchanged. The effect would be to reduce the gross margin available to its competitors, which might well make them unprofitable. In considering whether an undertaking is engaging in price squeezing in breach of the Competition Act, the Director General will consider whether the dominant undertaking would be profitable in the relevant downstream market if it had to pay the same input prices as its competitors.”232 6.29 The OFT's draft competition law guideline Assessment of Conduct states: “A margin squeeze may occur in an industry where a vertically integrated undertaking is dominant in the supply of an important input for a downstream market in which it also operates. The vertically integrated undertaking could then harm competition by setting such a low margin between its input price (e.g. the wholesale price) and the price it sets in the downstream market (e.g. retail price) that an efficient downstream competitor is forced to exit the market or its unable to compete effectively”.233 230 OJ 2003 L 263/9. Commission Notice on the Application of Competition Rules to Access Agreements in the Telecommunications Sector, OJ [1998] C265/2, paragraphs 117 and 118. 232 OFT 417, paragraph 7.26. 233 OFT, Assessment of conduct: draft guideline for consultation (OFT414a), April 2004, published at: http://www.oft.gov.uk/shared_oft/business_leaflets/competition_law/oft414a.pdf, paragraph 6.1. 231 109 6.30 In Genzyme, the OFT investigated Genzyme, which manufactured and supplied Cerezyme, the most effective drug for the treatment of Gaucher’s disease. Genzyme sold Cerezyme to the NHS at a price that included the provision of the separate service of homecare administration of the drug. Genzyme’s price to its own subsidiary was lower.234 The alleged margin squeeze consisted of supplying Cerezyme to third party delivery/homecare services providers of Cerezyme at the NHS list price. Genzyme left no margin for alternative providers to compete with it in the downstream market for homecare services. The OFT concluded that a margin squeeze had occurred in this case. This finding was upheld on appeal by the CAT.235 6.31 The CFI in Deutsche Telekom confirmed that the test for margin squeeze should be based on the dominant undertaking’s own charges and costs or those of an “equally efficient” competitor rather, than a “reasonably efficient” competitor test, as follows: “it is necessary to consider whether [Deutsche Telekom] itself or an undertaking just as efficient as [Deutsche Telekom] would have been in a position to offer retail services otherwise than at a loss if it had first been obliged to pay wholesale access charges as an internal transfer price […]”236 6.32 There are some features that are common to the test for margin squeeze as formulated in the case law. In Albion the Court of Appeal summarised those features: a) vertical integration: the undertaking alleged to be operating the margin squeeze must be vertically integrated, i.e. it must operate in related upstream and downstream markets, its upstream operations supplying an input to its competitors and to its own operations in the downstream market; b) dominance in the upstream market: the undertaking alleged to be operating the margin squeeze must enjoy a position of dominance in the upstream market (with a need for access to an input from the upstream market in order to operate in the downstream market). Such upstream dominance has been a feature of all of the EC and UK cases in which a margin squeeze has been found;237 c) significantly active in the downstream market: the abusive undertaking must be active downstream on a sufficient scale that its pricing decisions are capable of influencing the conditions of competition in the downstream market; d) insufficient margin between downstream and upstream price to cover costs: the undertaking must maintain an insufficient margin between its upstream and its downstream prices, thereby restricting competition in the downstream market; and e) no objective justification for the margin squeeze: it will be a defence for a dominant company to show an objective justification for a margin squeeze established under (a) to (d) above.238 6.33 234 Ofcom considers that criteria (a) – (c) are satisfied in this case as follows: OFT decision 27 March 2003, No CA98/3/03. Case No 1016/1/1/03 [2004] CAT 4 236 Case T-271/03 Deutsche Telekom, paragraph 194. See also paragraphs 188 to 192. 237 See National Carbonising Company, Napier Brown/British Sugar, Deutsche Telekom and Genzyme cases discussed above. 238 Dwr Cymru Cyfyngedig and Albion Water Limited and Water Services Regulation Authority [2008] EWCA Civ 536, paragraphs 87 to 110. 235 110 a) BT is vertically integrated between the ‘upstream’ market of NTS call termination/hosting and the ‘downstream’ market(s) for retail calls. b) BT was dominant in the market for NTS call termination/hosting in the UK over the relevant period. Competitors wishing to be active in NTS call termination/hosting need to purchase termination of calls to BT-hosted NTS numbers from BT (see Section 5 above). c) BT’s activity in the downstream market of retail calls is well documented, e.g. Ofcom’s Retail Price Controls, Explanatory Statement, 19 July 2006.239 6.34 In order to determine whether criterion (d) is met in this case, Ofcom considers that the correct approach, consistent with the case law set out above, is to determine whether an undertaking as efficient as BT’s downstream business could trade profitably in the downstream market when paying the wholesale prices charged by BT upstream; and that this should be tested by reference to BT’s own downstream costs. This is the test applied by the OFT in Genzyme. 6.35 In telecommunications and other network industries, a test based on the costs of a reasonably efficient operator (as opposed to the costs of an operator as efficient as the dominant undertaking’s downstream operations) has been considered to be appropriate either where the incumbent has costs advantages deriving from its status as “first-mover” in the relevant market or from its legacy position as historic incumbent; or where specific sectoral powers are being used in order to promote competition.240 However, as set out above, the CFI confirmed in Deutsche Telekom that the appropriate test for margin squeeze is the “as efficient” competitor test rather than a “reasonably efficient” competitor test. This was also confirmed by the Court of Appeal in Albion.241 6.36 In any event, Ofcom considered whether there were significant cost advantages enjoyed by BT in a relevant market (see paragraph 6.109 et seq below) but found that any such advantages were not material to its conclusions. Alleged margin squeeze in retail markets: analysis 6.37 Ofcom’s analysis followed the approach set out in the previous section, which conducts the margin squeeze test on the basis of the incumbent’s costs, thereby testing whether an equally efficient operator would be able to compete in the relevant downstream markets. 6.38 Ofcom has concluded that the prices notified in NCCN 500 do not impose a margin squeeze on C&W and BT’s other downstream competitors. 6.39 The remainder of this section sets out: • the range of products that should be included within the margin squeeze test (see paragraphs 6.40-6.59 below); • other methodological issues (see paragraphs 6.60-6.86 below); 239 Retail Price Controls: Explanatory Statement, statement of 19 July 2006 (see footnote 70). Commission Notice on the Application of Competition Rules to Access Agreements in the Telecommunications Sector, OJ [1998] C265/2, paragraphs 117 and 118. 241 Dwr Cymru Cyfyngedig and Albion Water Limited and Water Services Regulation Authority [2008] EWCA Civ 536, paragraph 105. 240 111 • details of the data sources and calculations used by Ofcom (see paragraphs 6.87-6.93 below); and • results and conclusions (see paragraphs 6.94-6.123 below). Alleged margin squeeze in retail markets: scope of margin squeeze test 6.40 As discussed in Section 5 above, Ofcom has concluded that BT was dominant in the market for NTS call termination/hosting in the UK over the relevant period. NCCN 500 notified increases to BT’s prices for NTS call termination on certain NTS number ranges. 6.41 The purpose of carrying out a margin squeeze analysis is to assess whether such a price increase is capable of having an impact on competition between firms at a downstream level. This means that the set of services across which the margin squeeze test should be applied depends on the range of services across which firms compete at the downstream level. 6.42 Table 10 summarises the key price changes notified in NCCN 500 (see Annex 1 for further details). Table 10: NTS termination prices pre and post NCCN 500242 0845 – long duration 0845 – short duration 0870 D E W D E W D E W Pre-NCCN 500 2.07 0.36 0.44 5.04 2.00 2.12 5.85 3.11 0.98 Post-NCCN 500 2.57 0.49 0.56 6.95 1.83 1.90 6.05 3.21 1.02 Difference (ppm) 0.49 0.12 0.13 1.91 (0.17) (0.21) 0.19 0.10 0.04 Difference (%) 24% 34% 28% 38% (9%) (10%) 3% 3% 4% Source: NCCN 500. Key: D = daytime, E = evening, W = weekend 6.43 242 As Table 10 shows, the price increases notified in NCCN 500 price differed in both percentage and absolute terms by number range, call duration and time of day, with: • significant increases on 0845 calls (particularly on all long duration, and daytime short duration, calls); • significantly smaller increases on 0870 calls; and • no material increases on other relevant number ranges.243 Prices for the pre-NCCN 500 period are taken from prices derived using the NTS Calculator as at 1 April 2004. Prices for the period during which NCCN 500 was in force are taken from NCCN 500 itself, as reproduced at Annex 1. 243 Ofcom’s analysis has not considered the potential impact of the changes to BT’s NTS call termination charges for 0820 in any depth. BT terminates a very small volume of calls to 0820 numbers on behalf of other OCPs, the total having fallen by an average of 88% per year over the last two years. In 2004/05, BT terminated about []minutes on 0820 on behalf of other OCPs, i.e. about []% of the corresponding total for 0845 calls and about []% of the corresponding total for 0870 calls. This means that even a proportionately very significant charge increase by BT, such as an 112 6.44 6.45 6.46 6.47 6.48 The companies that had their costs increased by BT’s NCCN 500 price increase can be divided into two categories: • communications providers that do not compete with BT in the relevant markets at the downstream level; and • communications providers that compete with BT at the downstream level. Communications providers that do not compete with BT in the relevant markets at the downstream level, and that would not therefore be affected by a margin squeeze, include: • the mobile network operators (O2, Vodafone, T-Mobile, Orange, and H3G), which sell calls services but operate in different product markets to BT (i.e. the market for mobile calls and access); and • Kingston Communications, which sells NTS calls but, as discussed in Section 4, operates in a different geographic market to BT. Communications providers that compete with BT at the downstream level, and which might therefore feel the impact of an anti-competitive margin squeeze, include: • Virgin Media (or, during the period under investigation, ntl Telewest); • other access providers that primarily focus on business customers (e.g. C&W); • indirect access providers, i.e. CPS and WLR (wholesale line rental – see Glossary) service providers such as TalkTalk. None of the types of providers listed in the previous paragraph offer a bundle consisting only of calls to 0845 and 0870 numbers, or calls to all NTS numbers. What they offer to consumers is one of the following: • a full range of services comprising both calls to all number ranges and access (i.e. line rental) and, in some cases, other products such as broadband internet access and digital television; or • a CPS service offering calls to all numbers (CPS Option 3); or • a CPS service offering calls to some numbers (CPS Option 1 and/or Option 2). Customers who are with BT for line rental can choose from a number of CPS options for their calls: • Option 1 (international calls only); • Option 2 (national calls only); • Option 1 and Option 2 together; and average of 1ppm, would have cost OCPs other than BT a combined total of £300 per year. It is clear that such a sum would be very unlikely to have an impact on competition and Ofcom’s analysis therefore generally excludes any consideration of 0820. 113 • 6.49 Option 3 (“all calls“ – international, national, local, mobile, non-geographic (including calls to NTS numbers), personal, paging and directory enquiries.244 Ofcom’s analysis therefore considered the impact of NCCN 500 on the margin earned by BT across: • a combination of all calls and access (relevant to competition with other retail OCPs, i.e. Virgin Media – or, during the period under investigation, ntl Telewest); • all call types but not access (relevant to competition with CPS service providers); and • the “CPS Option 3 incremental product bundle”, i.e. calls that are provided as part of a CPS Option 3 package but not as part of CPS Option 1 and Option 2 packages combined (local calls, calls to mobiles, calls to non-geographic numbers including NTS, personal, paging and director enquiries). If such a test were failed, NCCN 500 could have a negative impact on the level of competition by making the CPS “all calls” option unattractive to both prospective entrants and players currently in the market. 6.50 As noted above, C&W contends that “the margin squeeze test should be applied only to the products which are the subject of the margin squeeze“. Ofcom rejects this formulation as it does not accurately capture the basis on which other providers compete with BT, and cannot therefore enable Ofcom to assess whether BT’s conduct is anti-competitive. Ofcom notes that when observing margins for individual call and access products, (on 0845 and 0870) BT incurs negative margins on NTS call termination, where revenues are insufficient to recover product costs as derived on an FAC basis (see Figure 10 and Figure 11 below). However, as outlined below, a finding of a negative margin in relation to a particular range or across all retail NTS calls would be insufficient to demonstrate a margin squeeze likely to lead to an adverse effect on competition, since competition arises across a far broader set of retail call products. 6.51 Other things being equal, in cases where the upstream product(s) in question form a small proportion of the total costs of supplying the relevant downstream products, it is less likely that upstream price increases will lead to a margin squeeze. Retailers typically sell a bundle consisting of a number of different retail products, not all of which require the input in question. Margins on the relevant bundle of retail products may therefore remain adequate even when the price of the upstream product is increased. Upstream price increases may have an impact on competition if they lead to negative margins across the narrowest bundle of services across which any of the downstream firms competes. 6.52 As noted above, the set of services across which the margin squeeze test should be applied depends on the range of services across which firms compete at the downstream level. The smallest set of services over which it is appropriate to assess the margin is therefore the narrowest set of services sold by competitors in the NTS calls market that includes calls to 0845 and 0870 numbers. If this test indicates that there is no margin squeeze, it may not be necessary to carry out tests on a larger bundle as we would assume that margins on broader groupings of services would be profitable. For completeness, Ofcom also conducted these tests. 244 See Addressing the local call Disadvantage, 30 July 2004, published at: http://www.ofcom.org.uk/consult/condocs/cps_option/cps_statement/cps_stmnt.pdf 114 6.53 Some of BT’s competitors in the retail market focus on a particular group of customers, particularly business customers. It could therefore be the case that NCCN 500 had an impact on competition between suppliers of calls and access services to a particular customer segment. Ofcom therefore analysed services sold to residential and business customers separately. 6.54 We said above that the set of services across which the margin squeeze test should be applied depends on the range of services across which firms compete at the downstream level. As set out in the preceding paragraphs, the set of services across which firms compete, and the consumers they are targeting, vary. Ofcom therefore carried out six separate tests, each of which is based on the range of services offered by a particular type of competitor. These tests are set out in Table 11. Table 11: Parameters of tests for assessing alleged margin squeeze in call origination Customer type Product bundle Test 1 Residential customers all retail calls plus access Test 2 Residential customers all retail calls but not access (CPS Option 3) Test 3 Residential customers CPS Option 3 incremental product bundle (see paragraph 6.49 above) Test 4 Business customers all retail calls plus access Test 5 Business customers all retail calls but not access (CPS Option 3) Test 6 Business customers CPS Option 3 incremental product bundle 6.55 Rather than conducting a separate test for each of BT’s different retail offerings (e.g. the different BT Together options), Ofcom conducted the tests set out in Table 11 at a broader level, aggregated across all of the tariffs that BT offers to residential and business customers. Ofcom took this approach because the conduct under investigation applied equally to all of BT’s retail packages, and although NCCN 500 applied only to calls originated by competing OCPs, the test is constructed on the hypothesis that NCCN 500 had applied to calls made by BT’s retail customers. In any case, as discussed below, the impact of NCCN 500 on BT’s margins is not material whichever of the six test specifications we adopt. 6.56 Ofcom is not aware that the business of any of BT’s competitors in the NTS calls markets is tightly focused solely on the call types that were most affected by NCCN 500 to the extent that varying the proportions in which BT terminates traffic on the different NTS number ranges would affect the result. Ofcom therefore based its analysis on call charges averaged across time of day, calculated using the relevant proportions for BT-to-BT traffic in 2004/05 (“BT’s own time of day weights”).245 6.57 For completeness, Ofcom carried out two further tests focusing on downstream prices: • at a particular time of day; and 245 Source: BT letter to Ofcom dated 9 September 2005 and email response of 11 September 2005 to Ofcom Notice of 25 August 2005. 115 • for a hypothetical competitor supplying only business customers. 6.58 Such tests might represent the relevant scope if any of BT’s competitors in the origination market generated disproportionately high volumes of, for example, 0845 long duration evening and/or 0845 short duration daytime calls, since BT’s prices for these call types attracted the highest price increases. Business customers, for example, would generate higher volumes of daytime calls than residential consumers (in particular, 0845 short duration calls, for which the highest proportionate charge increases were levied). 6.59 Ofcom therefore conducted a sensitivity analysis by repeating its analysis using time of day weights based on the proportions in which one of BT’s competitors at the retail level, which supplies only business customers, originates calls to such call types. The results of this sensitivity analysis are set out in Annex 8. Alleged margin squeeze in retail markets: methodological issues Forward-looking or historical analysis 6.60 An issue to be considered is the methodology by which estimates of the margin will be made. In principle, there are two possible bases that we could use to calculate the margin: • an assessment of historical financial accounting data; and • a forward-looking net present value (“NPV”) analysis over a number of years.246 6.61 Generally speaking, where relevant historical accounting information is available for a meaningful amount of time, a historical accounting approach to profit calculation is likely to provide the more robust measure of profits. This is because it relies primarily on costs and revenues reflecting the actual transactions entered into. Furthermore such information is capable of being subjected to independent audit. 6.62 For the accounting approach, the most reliable source of data is the audited financial statements that cover a 12-month period. More detailed information can usually be obtained from an undertaking’s internal management accounts which are typically produced on a monthly basis and may contain an analysis of the profitability of individual products.247 6.63 BT is required to prepare regulatory accounting information on a current cost basis for certain economic markets in which it is designated as having SMP. This basis seeks to address in particular the use of asset valuations and depreciation charges based on possibly significantly out-of-date historical asset purchase cost. Current cost depreciation charges and asset valuations typically based on current replacement cost should reflect much more closely the opportunity cost of using assets during the period in question. Ofcom has therefore used information drawn from BT’s regulatory costing systems as the source for the financial inputs into its margin squeeze analysis. 246 Ofcom notes that an historical financial accounting approach should be distinguished from historic cost accounting. The former refers to the use of actual data which may include historic cost accounting figures (i.e. HCA) but equally could include current cost accounting (i.e. CCA) figures. 247 Certain regulated businesses, such as BT, are also required to produce detailed, audited accounts to a much greater level of detail than statutory annual accounts. 116 6.64 UK regulators and the Commission concur that the historical accounting approach is most appropriate where it provides a meaningful and reliable measure of the profitability of pricing and investment decisions in the period being considered and the BSkyB and Wanadoo cases provide examples of the explicit preference for a historical accounting approach over a forward-looking NPV approach. 6.65 In BSkyB, the OFT recognised the potential for historical accounting approaches to suffer from accounting distortions, but concluded: “The Director considers that the historical approach is the appropriate way of assessing margin squeeze in this case. The Director does not consider that a positive NPV for DisCo would necessarily imply that BSkyB has not margin squeezed. An historical approach that recognises and treats investment expenditures appropriately is therefore the best option available.”248 6.66 The OFT did not favour the use of a NPV approach where historic data was available in order to avoid the major (and well-known) problem that NPV approaches introduce – the potential that positive NPV results reflect the outcome of the anti-competitive behaviour.249 That is, sacrifices in short-run profits might lead to higher long-run profits, not due to any natural development in the market, but due to the elimination or exclusion of competitors. A margin squeeze, for example, will be profit-enhancing over time (the dominant undertaking incurs initial losses which will be more than offset in the future through higher prices or greater market share in the future), so that if the profitability of the dominant firm is assessed over a sufficiently long period, there would be no losses unless suitable adjustments are made. 6.67 The Commission in Wanadoo concluded that: “the Commission finds that only the adjusted [historical] costs approach allows any valid conclusions to be drawn; the others can help to throw further light on the matter, but no more.”250 6.68 The use of NPV analysis may be appropriate where accounting data do not accurately reflect economic costs, for example, because the margin squeeze is alleged to be occurring in a new market, where it is often reasonable to expect an undertaking to incur losses initially.251 6.69 However, to the extent that the NPV approach relies on forward-looking business plans, it will be based on assumptions regarding future revenues and costs. Assessing the validity of the assumptions in the business plan is not straightforward and inevitably involves a level of judgment. 6.70 In Ofcom’s view, consistent with the cases quoted above, the historical accounting approach is the most appropriate, in the current circumstances, to determine 248 OFT decision CA98/20/2002, BSkyB investigation: alleged infringement of the Chapter II prohibition, 17 December 2002, paragraph 390 http://www.oft.gov.uk/NR/exeres/E5019794-B2A848EF-B420-F96F8A2BAD7F.htm. 249 OFT BSkyB decision (see footnote 248), paragraph 389. 250 Commission decision of 16 July 2003 relating to a proceeding under Article 82 of the EC Treaty (COMP/38.233/Wanadoo Interactive). References in this document are to the English version of the decision available at http://ec.europa.eu/comm/competition/antitrust/cases/decisions/38233/en.pdf. Wanadoo Interactive appealed the Commission decision to the CFI on 2 October 2003 (OJ (2003) C289/46) and the CFI upheld the Commission’s decision on 30 January 2007 (CaseT-340/03). 251 OFT 417, paragraph 7.23. 117 whether or not the prices notified in NCCN 500 led to a margin squeeze during the period that they were in force. 6.71 NCCN 500 was applied in a mature market (the total volume of 0845 calls terminated by all TCPs, for example, declined by an average of 15% p.a. between 2002/03 and 2004/05), and was reversed after a relatively short period of time.252 This means that the drawbacks to the use of historical accounting data identified above are not material to this case. In assessing whether the prices notified in NCCN 500 led to a margin squeeze, the key question to be investigated is whether BT’s retail calls and access business remained profitable while those prices were in force.253 6.72 Ofcom therefore based its analysis in this case on a historical accounting approach. Appropriate basis for estimating costs 6.73 252 253 Another key issue to be considered in carrying out a margin squeeze test is the definition of cost. A number of different concepts of cost are defined in Table 12 below. Source: Responses from TCPs to S26 information request of May 2005 The results given in Figure 16 and Figure 17 answer this question. 118 Table 12: Relevant cost standards for Ofcom’s analysis Incremental cost is the cost of producing a specified additional product, service or increment of output over a specified time period. In many cases, the relevant increment may be the entire output of a particular service or group of services. The incremental costs of a service are then those costs which are directly caused by the provision of that service in addition to the other services which the firm also produces. Another way of expressing this is that the incremental costs of a service are the difference between the total costs in a situation where the service is provided and the costs in another situation where the service is not provided. Incremental cost can be contrasted with the standalone cost (“SAC”) of a service which is the cost of providing that particular service on its own, and with common costs. Common costs are those which arise from the provision of a group of services but which are not incremental to the provision of any individual service. Common costs may be identified in the following way: if the incremental costs of each service are removed from the total cost of providing all services, what are left are the common costs (i.e. those costs which are shared). Where there are no common costs, incremental cost and SAC are the same. Where there are common costs, the SAC of a service is the sum of the incremental cost of the service plus all of the costs which are common between that service and other services. In this case there are said to be economies of scope, that is, total costs are reduced by producing a number of services together, because common costs then have to be incurred only once. Variable costs are costs that vary with an undertaking’s output over a specified time period. By contrast, fixed costs are costs that do not vary with output over the specified period. Total costs are the sum of variable and fixed costs. Fully allocated cost (“FAC”) is an accounting concept by which all the costs of a multiproduct firm are attributed to its various activities, primarily on the basis of cost causation. Average FAC is the nearest equivalent of average total cost (“ATC”) in a multi-product firm. Long-run average incremental cost (“LRIC”) is equivalent to long-run average variable cost for a given output increment (e.g. the service in question) and time period. 6.74 Whether costs can be considered as variable, fixed, incremental or common depends on both the size of the output increment and the time horizon over which the variable or incremental nature of costs is being assessed. Variable and incremental costs are the same if the same output increment and time horizon is used. 6.75 The time horizon may not coincide with the period of investigation – it may be shorter or longer depending on the specific nature of the conduct and the life of the assets involved. 6.76 Ofcom believes that among two relevant factors in the decision on the relevant time horizon are (a) the period over which the alleged margin squeeze prevailed; and (b) whether the margin was temporarily low and expected to rise in the future. Further factors are the asset lives of key investments and the nature of the behaviour under investigation. 6.77 The period over which the charges were in force is significant because it provides a minimum time horizon. For example, if a price was a special offer for a single week or month, it may be appropriate to consider only costs that were variable within that 119 week or month. However, if prices had been sustained (without rising) for a period of over two years, it is reasonable to expect that prices should cover the costs that were variable with output over the two-year period. Otherwise, there would be a risk of concluding that BT could reasonably fail to recover costs that could have been avoided (i.e. were incremental) in that period. 6.78 It could also be argued that a longer timeframe could be appropriate if the conduct had ceased solely due to the commencement of enforcement proceedings. However, this is not a critical issue in this case because, as described below, the findings relating to margin squeeze are not sensitive to the time period over which variability is assessed. 6.79 It is necessary to compare BT’s revenues in the downstream market with its relevant costs in order to determine whether the margin between its upstream and downstream prices was sufficient to enable its relevant downstream operations to cover their costs during the period covered by the investigation. Profitability is used as the measure of whether revenues are sufficient to recover (downstream) costs, and the measure of whether prices are sufficient to recover unit costs. As the OFT states: “To test for margin squeeze, it is usual to determine whether an efficient downstream competitor would earn (at least) a normal profit when paying input prices set by the vertically integrated undertaking.”254 6.80 In assessing a potential margin squeeze, there are a number of cost standards that might be applied. 6.81 In AKZO, the ECJ determined that a price or margin below average variable cost (“AVC”) charged by a dominant firm is presumed to be anticompetitive. This is because, if a price is less than AVC, the firm would be better off ceasing production entirely, and such conduct is therefore (generally) only rational if there is an anticompetitive motive. Thus: “Prices below average variable costs (that is to say, those which vary depending on the quantities produced) by means of which a dominant undertaking seeks to eliminate a competitor must be regarded as abusive. A dominant undertaking has no interest in applying such prices except that of eliminating competitors so as to enable it subsequently to raise its prices by taking advantage of its monopolistic position, since each sale generates a loss, namely the total amount of the fixed costs (that is to say, those which remain constant regardless of the quantities produced) and, at least, part of the variable costs relating to the unit produced.”255 6.82 The ECJ went on to say that a price or margin between the average variable cost of offering a service and the average total cost (“ATC”) may be anticompetitive if it can be shown that there is intent to exclude and that: “Moreover, prices below average total costs, that is to say, fixed costs plus variable costs, but above average variable costs, must be regarded as abusive if they are determined as part of a plan for 254 255 OFT 414a. Case C-62/86 AKZO Chemie v Commission [1991] ECR I-3359, paragraph 71. 120 eliminating a competitor. Such prices can drive from the market undertakings which are perhaps as efficient as the dominant undertaking but which, because of their smaller financial resources, are incapable of withstanding the competition waged against them.”256 6.83 A price or margin above ATC is presumed not to be anticompetitive. In his Opinion ([2000] ECR I-1365) on Cewal’s appeal to the ECJ, Advocate General Fennelly stated: “I would, on the other hand, accept that, normally, non-discriminatory price cuts by a dominant undertaking which do not entail below-cost sales should not be regarded as being anti-competitive. In the first place, even if they are only short lived, they benefit consumers and, secondly, if the dominant undertaking’s competitors are equally or more efficient, they should be able to compete on the same terms.” 257 6.84 Ofcom’s analysis was primarily based on an analysis of FAC, which is the nearest equivalent to ATC in a multi-product firm (see Table 12). A test that is passed based on an analysis of FAC, provided the attribution is reasonable, is usually taken as strong evidence that there is no margin squeeze. It is therefore only in cases where such a test is failed that it is important to undertake an analysis based on variable or incremental costs. 6.85 Ofcom’s tests were based on BT’s retail margin for various products, calculated by subtracting the sum of retail costs and internal transfer charges from retail revenues. Transfer charges between different parts of BT’s business and interconnection charges from other fixed and mobile communications providers form a significant proportion (80% or higher) of the cost of providing the retail calls and access services that form the basis of this margin squeeze test. 6.86 This means that estimates of variable/incremental cost and FAC measured across these services are relatively close together (because the transfer and interconnection charges are a component of both FAC and incremental cost). Because Ofcom’s margin squeeze tests were passed on a FAC basis, it has not been necessary to reach a conclusion to carry out a detailed analysis based on estimates of variable or incremental costs. Data supplied by BT 6.87 Ofcom required BT to provide cost, revenue and volume data from its regulatory costing systems for its various PSTN retail products (see Table 2 above), which include the various retail products considered in the various bundle options. 6.88 Ofcom required BT to provide data for both the 2003/04 and 2004/05 financial years. As NCCN 500 came into effect on 1 May 2004, Ofcom did not consider previous financial years in the context of this piece of its analysis. Nor did Ofcom use forecast data in its analysis for a number of reasons. First, accounting data was not available 256 Case C-62/86 AKZO Chemie v Commission [1991] ECR I-3359, paragraph 72. On the issue of intent in this case, see paragraph 6.545 et seq below. 257 Opinion of Advocate General Fennelly on Cewal’s appeal to the Court of Justice, Cases C-395, paragraph 132, and 396/96P Compagnie Maritime Belge v Commission [2000] ECR I-1365 (“Compagnie Maritime Belge”). This case involved an exception to this general presumption. 121 for the full 2005/06 financial year when Ofcom carried out its analysis. Second, traffic affected by NCCN 500 (notably calls to 0845 numbers) is declining as a proportion of all traffic (from []% to []% from 03/04 to 04/05), which means that it is increasingly unlikely that BT’s conduct in raising its charges could have led to a margin squeeze in future, with the affected calls accounting for a declining proportion of downstream revenues.258 In any case, the prices notified in NCCN 500 ceased to have effect on 1 January 2006, making an analysis of forecast data from beyond this point inappropriate. 6.89 BT was unable to provide origination volumes for 0845 and 0844 separately, or for 0870 and 0871 separately. 6.90 In its analysis Ofcom therefore assumed that NCCN 500 applied to all BT-to-BT 084X and 087X calls. Other things being equal, this would have overstated the impact of NCCN 500. BT-to-BT 0844 and 0871 calls account for a [] proportion of BT-to-BT NTS traffic, with 0844 volumes accounting for less than []% of all such 084X traffic, and 0871 volumes accounting for less than []% of all such 087X volumes in 2004/05. 6.91 The data supplied by BT did not consistently analyse services sold to business and residential customers separately across all its services, and did not separately record information on BT-to-BT traffic. To control for these factors, Ofcom made a number of adjustments to the data supplied by BT before conducting its analysis. These adjustments are set out at Annex 9, which provides a full list of the data sources and calculations used in Ofcom’s analysis. Measure of profitability 6.92 In order for a margin to be considered adequate, it must be sufficient to allow the firm to earn its required minimum rate of return on capital. 6.93 In this case, Ofcom considers that the appropriate measure of the minimum required return on investment is BT’s Return on Sales (“ROS”), for the reasons set out in Annex 5. Results 6.94 On a FAC basis, in 2004/05, [ (some of BT’s services were loss-making)].259 6.95 []. However, Ofcom considers that the additional losses caused by NCCN 500 would have been insufficient to lead to a margin squeeze when considering BT’s margin across all relevant call types, as set out in the following tables. Table 13: BT’s margin (all business products), 2004/05 (£m) Connections From BT regulatory accounting information NCCN 500's impact Line rental Local calls Business FAC Returns Nat It'l calls CTM FF calls 084x 087x [] Numbers are rounded and therefore total figures may not total exactly 258 259 BT response dated 9 September 2005 to Ofcom’s section 26 Notice of 19 August 2005 []. 122 09x Total Table 14: BT’s margin (all residential products), 2004/05 (£m) Connections From BT regulatory accounting information NCCN 500's impact -107 0 Line rental Local calls -256 0 407 0 Residential FAC Returns Nat It'l calls CTM FF calls 80 0 93 0 117 0 084x 087x 09x Total [] Numbers are rounded and therefore total figures may not total exactly 6.96 As Table 13 and Table 14 above show, if BT, as an OCP, had faced NCCN 500, the impact on its profits would have been relatively small. 6.97 As set out at paragraph 6.46 above, at a downstream level, BT competes with companies with a variety of different business models. There are therefore a number of possible constructions for the margin squeeze test, which are listed in Table 11. The following discussion considers BT’s margins on these different service increments, which, to recap, are (for residential and business customers separately): • all retail calls plus access; • all retail calls but not access (CPS all calls); and • CPS Option 3 incremental product bundle. All retail calls plus access (tests 1 and 4) 6.98 The first pair of tests measures BT’s margin across all call types plus access, in order to ascertain whether NCCN 500 created a margin squeeze on ntl Telewest (now Virgin Media) and other direct access providers that compete with BT. 6.99 To the extent that BT’s competitors are able to rely on their own networks (instead of buying wholesale inputs from BT), these tests may overstate the likelihood of a margin squeeze to the extent that BT’s wholesale charges are greater than its competitors’ incremental costs. 6.100 Figure 10 and Figure 11 set out estimates of BT’s margin expressed as ROS for all of BT’s retail services, both before and (assuming that BT faced the price increases notified in NCCN 500) after NCCN 500. The “total” figure in the final column measures BT’s margin across all of the services shown in the Figures, i.e. all call types plus access. In Figure 10 and Figure 11, “CTM” means calls to mobiles and “FF” means calls to freephone (or free to caller, i.e. 0800) numbers. Note that the margin on NTS call types alone is not a relevant test, as set out in paragraphs 6.406.59 above. 123 Figure 10: BT’s margin, all calls and access (business) 2004/05 Figure 11: BT’s margin, all calls and access (residential) 2004/05 6.101 Figure 10 and Figure 11 show that BT earned: • a significant negative margin on access services (with the exception of business line rentals); • a significant positive margin on geographic call types; and • a negative margin on 084X, 087X and 09X calls, which would have been exacerbated by NCCN 500 had it applied to BT-to-BT calls. 6.102 The impact of NCCN 500 on BT’s margin, measured across all of the services listed in Figure 10 and Figure 11, was relatively small. For services provided to business customers, the impact of NCCN 500 would have been to reduce BT’s ROS, measured across all the relevant retail services, from []% to []%. For services provided to residential customers, the impact of NCCN 500 would have been to reduce BT’s ROS, measured across all the relevant retail services, from []% to []%. 6.103 Figure 10 above shows that the impact of NCCN 500 on BT’s profitability, when measured across all business calls and access services, was relatively small. This is because of the relatively small number of calls to the relevant number ranges 124 originated by business customers (primarily because businesses do not, as a rule, use dial-up internet access). 6.104 Figure 11 shows that the impact of NCCN 500 on BT’s margin on residential calls and access would have been somewhat greater, but still small, i.e. it would tend to lower BT’s ROS, measured across all services, from []% to []%. 6.105 Ofcom’s view is that the relative magnitudes of the pre- and post-NCCN 500 profitability levels shown in Figure 10 and Figure 11 provide useful additional evidence in assessing whether NCCN 500 introduced a margin squeeze against BT’s downstream competitors. However, as is apparent from Figure 10 and Figure 11, BT’s ROS for all calls and access combined suggests that BT earned a positive margin on these services. 6.106 Ofcom concludes, on the basis of these results which show that the margin was sufficient to allow competitors to earn an adequate return (as compared to the benchmarks referred to in Annex 6), and the relatively small impact of NCCN 500 on profitability, that NCCN 500 did not lead to a margin squeeze when measured across all call and access services. Therefore Ofcom concludes that NCCN 500 did not distort downstream competition between BT and other direct access providers, i.e. Virgin Media (formerly ntl Telewest). All retail calls but not access (Tests 2 and 5) 6.107 The second pair of tests that Ofcom conducted measured BT’s profitability across all retail calls but not access in order to ascertain the impact of NCCN 500 on competition between BT and CPS providers. 6.108 []. It follows from this that, absent any other adjustments, it is easier for BT to pass a margin squeeze test on the basis of calls only than it is to pass a test based on both calls and access. The results set out in Figure 12 and Figure 13 show that, for both residential and business services: • [] • []. Figure 12: BT’s margin, calls only (residential) Turnover (£m) Residential customers ROS (%) Return (£m) pre-NCCN500 post-NCCN500 pre-NCCN500 post-NCCN500 Test 1 (all calls + access) Test 2 (calls only) [] Figure 13: BT’s margin, calls only (business) Turnover (£m) Test 4 (all calls + access) Test 5 (calls only) Business customers ROS (%) Return (£m) pre-NCCN500 post-NCCN500 pre-NCCN500 post-NCCN500 [] 125 6.109 Ofcom considered whether it was appropriate to make additional adjustments to carry out this version of the test to control for the so-called “local call disadvantage”. 6.110 Even where CPSOs (carrier pre-select operators) interconnect at all of BT’s local exchanges, they still face higher costs than BT. This is because for a local call between two customers that are connected to the PSTN using a BT access line (i.e. customers of BT Retail or CPSOs/WLR service providers), there are often additional switching stages when an interconnecting operator is used to carry the call. 6.111 The calls for which this is an issue are those between customers connected to the same local exchange, and those between customers connected to adjacent local exchanges. BT routes these calls entirely on its own network, so that they do not pass via a transit exchange. However, CPSOs must take these calls off at BT’s DLE (digital local exchange), route them up to their own switch and then send them back to the same DLE, incurring additional costs in the process. This effect is referred to as the “local call disadvantage.”260 6.112 While the impact of the local call disadvantage is difficult to quantify precisely, BT’s margins, measured across all calls, are sufficiently large that, even if returns on local calls earned by BT were reduced or removed from Ofcom’s analysis, BT would still make a positive margin. This is because of the substantial margins earned on national and international calls and on calls to mobiles, as shown in Figure 14 and Figure 15 below. Figure 14: BT’s margin, all calls except local calls (business), 2004/05 260 See Investigation against BT about potential anti-competitive exclusionary behaviour: Decision of the Office of Communications, 12 July 2004, published at: http://www.ofcom.org.uk/bulletins/comp_bull_index/comp_bull_ccases/closed_all/cw007/cw_760_dec/ cw_760_dec.pdf. 126 Figure 15: BT’s margin, all calls except local calls (residential), 2004/05 6.113 BT’s margin (measured as ROS) on sales to business customers across all calls excluding local calls (assuming that BT had faced NCCN 500) was []% (versus []% without NCCN 500). For sales to residential customers, BT’s ROS across all calls excluding local calls was []% (versus []% when BT did not face NCCN 500). 6.114 These results are qualitatively very similar to those shown in Figure 12 and Figure 13 above, showing that BT earned high margins across all calls even discounting all the returns associated with local calls, and that, had NCCN 500 applied to NTS calls originated by BT, the impact on BT’s margin would have been insufficient to lead to a margin squeeze. 6.115 Ofcom therefore concludes that NCCN 500 did not impose a margin squeeze on BT’s downstream competitors that offer calls only, i.e. CPSOs. CPS Option 3 incremental product bundle (Tests 3 and 6) 6.116 Finally, Ofcom conducted a margin squeeze test on the CPS Option 3 incremental product bundle, i.e. calls that are provided as part of a CPS Option 3 package but not as part of CPS Option 1 and Option 2 combined, in order to ascertain whether NCCN 500 created a margin squeeze on CPS operators wishing to provide the “all calls” option. 6.117 The results of this pair of tests are shown in Figure 16 and Figure 17 below. Figure 16: The impact of NCCN 500 on Tests 1, 2 and 3 (services sold to residential customers) Turnover (£m) [a] [b] [c] [d] [e] Test 1 (all calls + access) Test 2 (calls only) CPS all calls CPS national and international Test 3 (margin of [c] over [d]) Residential customers ROS (%) Return (£m) pre-NCCN500 post-NCCN500 pre-NCCN500 post-NCCN500 [] Source: BT data 127 Figure 17: The impact of NCCN 500 on Tests 4, 5 and 6 (services sold to business customers) Turnover (£m) [a] [b] [c] [d] [e] Test 4 (all calls + access) Test 5 (calls only) CPS all calls CPS national and international Test 6 (margin of [c] over [d]) Business customers ROS (%) Return (£m) pre-NCCN500 post-NCCN500 pre-NCCN500 post-NCCN500 [] Source: BT data 6.118 It is clear from Figure 16 and Figure 17 above that there is a significant margin between CPS all calls and the other CPS options and that NCCN 500 had a very limited impact on this margin. 6.119 []. Figure 18: BT’s margin, local calls, calls to mobiles, and NTS calls (residential) Figure 19: BT’s margin, local calls, calls to mobiles, and NTS calls (business) 128 6.120 In order to control for the local call disadvantage (see paragraph 6.111), Ofcom calculated the margin between CPS Option 3 and the other CPS options, but subtracted from this margin the returns associated with local calls. []. 6.121 BT’s margin on CPS Option 3 excluding local calls and the other CPS options, measured as ROS, was []% for business customers ([]% assuming that BT had faced NCCN 500) and []% for residential customers ([]% assuming that BT had faced NCCN 500). 6.122 Based on these factors, and given that removing all of the returns associated with local calls is an extreme assumption, Ofcom concluded that NCCN 500 did not lead to a margin squeeze on competitors offering CPS Option 3 (all calls) as opposed to other CPS options. Conclusion on retail margin squeeze 6.123 Based on the analysis set out above, Ofcom has concluded that NCCN 500 did not lead to an anti-competitive margin squeeze between BT’s upstream wholesale NTS call termination charges and the retail prices that it charged in the downstream retail calls and access markets. Responses to draft decision: retail margin squeeze 6.124 Ofcom received three responses which questioned Ofcom’s retail margin squeeze analysis, from BT, C&W and Thus. BT 6.125 BT states that it would not have conducted the margin squeeze test in the same way as Ofcom, but that it nevertheless supports the conclusions reached: “Generally BT would not necessarily have applied the [retail margin squeeze] test or the analysis in the same [way] that Ofcom has done. However that Ofcom did apply the test in the way that it has done clearly demonstrates that BT’s behaviour in fixing the charges for the relevant period, even if BT were dominant on the relevant markets, was not abusive…” C&W 6.126 C&W notes that there are two possible cost bases for assessing an alleged margin squeeze: “the costs of a reasonably efficient operator on one hand and the dominant operator’s own costs on the other”.261 C&W notes that Ofcom has concluded that in this case it is appropriate to base its tests on BT’s own costs, and submits that Ofcom “provides no conclusive authority to justify this view”, and that “at the very least [Ofcom] has simply chosen to apply the test most favourable to BT”.262 6.127 Having selected this cost base, C&W submits, Ofcom has failed to adjust costs for any scale efficiencies apart from the local call disadvantage (see paragraphs 6.109 et seq above). C&W submits that Ofcom is wrong not to adjust costs and that the implication of this is that the costs used in the test are too low, and the test is therefore too easy to pass. C&W submits that, as a result of the call origination 261 262 C&W’s response to the draft decision, page 31. C&W’s response to the draft decision, page 32-33. 129 condition, “if BT had to pay the NCCN 500 price then […] it may not have been able to cover its cost”.263 6.128 C&W submits that in identifying the narrowest set of products across which companies compete in order to conduct its test, Ofcom has: “…expand[ed] the margin squeeze test across several products that are not directly derived from the upstream product. Ofcom cites no precedent for this expansion, and indeed case law does not support it.”264 6.129 C&W submits that Ofcom has departed from established case law, in which “the analysis of margin squeeze has been confined to the relevant product market”.265 6.130 C&W submits that Ofcom has not examined the impact of NCCN 500 on “those CPS[Os] that only provide calls to some numbers, e.g. CPS Option 1 and/or Option 2”.266 6.131 C&W submits that Ofcom: “has not seriously considered whether end-users buy different kinds of calls from different providers on a “least cost routing” basis, or whether different kinds of calls are bought and sold separately on wholesale markets.”267 6.132 C&W contends that the effect of applying the margin squeeze test to the wrong product set is that the effect of NCCN 500 is “diluted”, such that: “Ofcom fails to take into account the fact that BT was making a loss on retail calls subject to NCCN 500.”268 Thus 6.133 Thus considers believes that: “Ofcom’s margin squeeze conclusions are invalid because it has performed the analysis on the wrong downstream market.”269 6.134 Thus concludes that “Ofcom should re-do its margin squeeze test focusing on the narrower retail market comprising calls to numbers covered by NCCN 500”.270 6.135 Thus argues that “Ofcom has ignored BT’s significant and inherent cost advantages” in constructing its margin squeeze test.271 6.136 Thus is concerned that: 263 C&W’s response to the draft decision, page 34. C&W’s response to the draft decision, page 35. 265 Ibid. 266 C&W’s response to the draft decision, page 36. 267 Ibid. 268 C&W’s response to the draft decision, page 40. 269 Thus’s response to the draft decision, page 8. 270 Thus’s response to the draft decision, page 9. 271 Ibid. 264 130 “Ofcom has ignored the fact that consumers can and do use the ‘1280’ prefix to over-ride routing to their CPS provider and divert calls selectively back to BT….(T)he availability of the CPS prefix was widely recognised in the market, and acted as a mechanism to disaggregate competition to a lower level than has been assumed by Ofcom in its analysis.”272 6.137 Thus submits that the availability of 1280 leads to “competition on a number range by number range basis” between BT and CPSOs, and offers, as evidence of this, online advice for consumers explaining how they can use 1280 on a call-by-call basis, and the wording of the letter sent by BT to customers who had switched from BT Retail to a CPSO at the time that NCCN was in force, which invited them to “check which calls are cheaper with BT”.273 Ofcom comments on responses to the draft decision 6.138 In addition to respondents’ comments on Ofcom’s margin squeeze analysis, which are considered in the following paragraphs, respondents raised a number of issues with Ofcom’s approach to margin squeeze in the context of Ofcom’s proposed market definition and analysis of C&W’s discrimination allegation. We deal with these at paragraphs 4.218-4.231 above and 6.263-6.291 below respectively. 6.139 C&W argues that the margin squeeze test should be based on its own (higher) costs rather than BT’s (i.e. a reasonably efficient operator approach). C&W also argues that in its analysis of the BT Together case (see paragraph 6.141) Ofcom took account of the scale disadvantages of BT’s competitors. 6.140 Ofcom has considered both the equally efficient operator (“EEO”) test and the reasonably efficient operator (“REO”) approaches. As set out above, Ofcom notes the judgment of the CFI in Deutsche Telekom and of the Court of Appeal in Albion in relation to the appropriate test, i.e. the EEO. 6.141 In the BT Together decision, Ofcom conducted an EEO test, an REO test and a hybrid test (combining the EEO and REO approaches) to allow for the local call disadvantage.274 However, Ofcom rejected the full REO approach, on the grounds that it did not accept that the scope of the REO should be extended to take account of matters going beyond unavoidable disadvantages (i.e. unavoidable structural costs). 6.142 In the Gamma Telecom decision, Ofcom applied a hybrid test that allowed for only unavoidable additional costs (as in the BT Together decision).275 6.143 In the Freeserve 2 decision, the primary test was an EEO test based on the dominant undertaking’s costs.276 In this case, BT’s legacy advantages were not relevant, and therefore only the EEO test was conducted. 272 Thus’s response to the draft decision, page 8-9. Ibid. 274 See footnote 260. 275 Complaint from Gamma Telecom against British Telecommunications Group plc (“BT”) about reduced rates for Wholesale Calls from 1 December 2004: Decision of the Office of Communications, 13 June 2005, published at: http://www.ofcom.org.uk/bulletins/comp_bull_index/comp_bull_ccases/closed_all/cw_802/262194/ 276 Investigation by the Director General of Telecommunications (the ‘Director’) into alleged anticompetitive practices by British Telecommunications plc (‘BT’) in relation to BTOpenworld’s 273 131 6.144 As set out at paragraph 6.35 above, where BT has an inherent incumbency advantage Ofcom could consider applying the REO test (either in part or in full, depending on the extent of the advantages). Ofcom notes the tests set out by the Court of Appeal in Albion (see paragraph 6.32 above) and by the CFI in Deutsche Telekom (see paragraph 6.31 above). In any event, in this case Ofcom considered the presence of the so-called local call disadvantage, and has also applied a hybrid or adjusted EEO test. 6.145 As set out at paragraph 6.109 et seq above, the local call disadvantage arises in relation to calls where two customers are connected to the same local exchange via a BT PSTN access line. Whereas BT routes these calls internally, without the need for transit via an exchange, a CPSO (e.g. C&W) must take these calls off BT’s network, route them using its own switch and then carry that traffic back to the same DLE, thereby incurring additional costs. 6.146 Ofcom has, in respect of tests 2,3,5 and 6 (see Table 11 above), acknowledged the higher interconnection circuit costs of a CPSO (i.e. for a reasonably efficient operator) when assessing margin squeeze and interpreting the results. 6.147 Accordingly, scale disadvantages were taken into account by means of the local call disadvantage, an adjustment which Ofcom also made in the BT Together case, as C&W acknowledges. Ofcom states, in the final decision in that case: “Ofcom believes it may be reasonable to take account of structural cost disadvantages in a margin squeeze test and in this investigation Ofcom has considered the effects of structural cost disadvantages faced by CPS Providers at the upstream network input level in terms of the local calls disadvantage. Ofcom does not believe, however, it is reasonable in this case to extend the scope of the ‘reasonably efficient new entrant’ margin squeeze test to take account of matters going beyond such unavoidable disadvantages.”277 6.148 As stated at paragraph 6.52 above: “the set of services across which the margin squeeze test should be applied depends on the range of services across which firms compete at the downstream level.” 6.149 Ofcom considers that retail providers do not compete for the provision of 0845 and/or 0870 calls only. Rather, such calls are offered as part of a broader bundle including various different call types and, in some cases, access. Accordingly, Ofcom does not accept C&W’s argument that the margin squeeze test should be conducted on NTS calls only. Even where NTS calls are not offered as part of call packages by some operators, Ofcom does not consider that this would alter Ofcom’s test. 6.150 C&W suggests that Ofcom’s approach may be inconsistent with case law in that the products included in the relevant product set are “not directly derived from the upstream product” (see paragraph 6.128 above). As discussed at paragraph 4.222 and paragraphs 6.50-6.54 above, however, it is necessary to broaden the product set (beyond products “directly derived from the upstream product”) in order to capture accurately the basis on which other providers compete with BT. Ofcom notes (‘BTOW’) consumer broadband products, 20 November 2003, published at: http://www.ofcom.org.uk/bulletins/comp_bull_index/comp_bull_ccases/closed_all/cw_613/cw_613.pdf 277 Gamma Telecom (see footnote 275), paragraph 120. 132 that C&W appears, indirectly, to support this premise in its comments on excessive pricing, where it suggests that it would be unrealistic for any competitor of BT to provide only NTS services and not to provide any other calling products.278 6.151 Ofcom has carefully considered competition between retail call providers and retail call packages, and has given particular attention to the offerings of CPS providers. In doing so, Ofcom has based its margin squeeze tests on a range of different product groupings, including CPS options available in the retail market over the period of the investigation (including CPS all calls and CPS Option 3 – see paragraph 6.97 et seq above). Ofcom understands that there are no CPS providers that offer calls to NTS numbers only. [] is therefore irrelevant for the purpose of assessing the alleged retail margin squeeze and its impact on other OCPs. 6.152 Ofcom has considered Thus’s claim that CPS customers’ ability to use 1280 to divert some calls to BT demonstrates that the market (and the relevant bundle for conducting a margin squeeze test) is narrower than Ofcom suggests (see paragraphs 4.227-4.231 above). 6.153 As discussed at paragraph 4.229 above, Ofcom sought further evidence for Thus’s claim but has not been able to find any evidence that 1280 was widely used over the period under investigation, to the extent that it would represent the kind of competitive constraint Thus suggests. In any case, it seems unlikely that it would have been possible, by means of NCCN 500, for BT to have established a dominant position in the retail supply of calls to NTS numbers by encouraging use of 1280. Ofcom does not therefore agree that that the availability of the 1280 function affects the relevant market definition or its analysis of margin squeeze. 6.154 As discussed at paragraph 3.35 above Ofcom also required BT to provide details of marketing and advertising undertaken over the period that mentioned the cost of calling NTS numbers. While the material supplied by BT refers to differences between BT’s prices for NTS numbers and those of its competitors, Ofcom does not believe that this material suggests, as Thus claims, that BT “was trying to present the 0845 and 0870 calls market as distinct” – as the intention appears to have been to market BT’s BT Together inclusive packages. Ofcom notes that the advertising campaign mentioned by Thus does not appear to mention the cost of calling NTS numbers.279 6.155 C&W asserts that Ofcom has not considered whether the availability of least cost routing products may affect its proposed retail margin squeeze analysis (see paragraph 6.131 above). 6.156 Ofcom understands least cost routing to be a product offered to business customers, which uses software to route calls made by those customers according to the cheapest route that the provider has identified for a particular call type. Least cost routing providers do not supply calls over their own networks, and buy calls from a number of different providers. Since their customers do not call only NTS numbers, least cost routing products will therefore route calls to providers that provide both NTS and non-NTS calls. Accordingly, Ofcom does not accept that NTS calls (bundled with non-NTS calls) provided through a least cost routing package, are suggestive of a different product bundle upon which to apply Ofcom’s retail margin squeeze test. 278 279 C&W’s response to the draft decision, page 46. See “BT battles against low-cost competitors”, Guardian, 11 June 2004. 133 Possible margin squeeze in NTS hosting Introduction 6.157 In the previous section, Ofcom considered whether BT’s margins on services sold to retail residential and business customers would have been squeezed as a result of the increases in NTS call termination charges notified in NCCN 500. Ofcom found that BT’s retail margins were sufficient for an equally efficient downstream operator to cover its retail costs. Accordingly, Ofcom concluded that there was no retail margin squeeze. 6.158 For completeness, Ofcom also considered whether additional revenues earned by BT from the higher charges that applied under NCCN 500 could have been used to increase payments to NTS service providers to a level that other communication providers could not match without being forced to incur losses. Ofcom has termed this possible abuse a “hosting margin squeeze”. 6.159 This analysis also addresses C&W’s allegation that, as a result of NCCN 500: “BT will be able to leverage its dominance into the “downstream” market of NTS call termination since it will be in a position to use the additional revenues generated relative to its competitors to offer more lucrative fee share arrangements to NTS service providers.”280 Hosting margin squeeze in a two-sided market 6.160 Before NCCN 500, prices for NTS call termination were uniform across the industry (see paragraph 2.26 above), with BT charging the same as other TCPs for terminating calls on its network. Under NCCN 500, however, the gross termination charge payable to BT by other OCPs was more than other TCPs could charge for terminating calls whether originated by BT or other OCPs (unless no transit via BT was involved). 6.161 The TCP can pay out part of the revenues it receives in gross termination charges to its NTS service provider customers. The level of payments to NTS service providers is one of the main ways that TCPs compete against each other. The residual (the net termination charge), represents the TCP’s margin, out of which it must cover the costs of NTS call termination and hosting, including any value added services provided to the extent not charged for separately. 6.162 The higher gross termination charge received by BT under NCCN 500 could have allowed it to impose a hosting margin squeeze as follows. If, as a result of increased revenues from NCCN 500, BT was in a position to increase the revenue share paid out to NTS service providers (reduce its net charges to them for hosting) to a level that other TCPs could not profitably match, this may have given it an unfair advantage over competing TCPs. 6.163 This would constitute a “hosting margin squeeze” because the margin available to competing TCPs would have been insufficient to cover their costs. BT’s ability to impose a hosting margin squeeze would be possible only as a result of its dominance in NTS call termination, which gives it the ability to raise the gross termination charge above the level other TCPs can charge. By raising payments to a 280 C&W’s submission to Ofcom of 15 March 2005, paragraph 8.29. 134 level other TCPs could not match, BT would then be able to increase its share of NTS hosting and possibly force other TCPs to exit the market. Definition of the hosting margin 6.164 The hosting margin is defined as the difference between the gross termination charge received by other TCPs and BT’s revenue share payments to NTS service providers. If this hosting margin is less than BT’s costs of providing NTS call termination and hosting, this would suggest that NCCN 500 led to a hosting margin squeeze. 6.165 Essentially, the test asks whether, given the level of the gross termination charge that competing TCPs receive and the level of BT’s revenue share payments, an equally efficient operator could match BT’s payments and cover its costs in the termination/hosting market. That is, there is a hosting margin squeeze if: Gross termination charge received by other TCPs – BT payment – BT termination/hosting costs < 0 6.166 The use, in the test, of the gross termination charge received by non-BT TCPs (rather than that received by BT) reflects the position that the dominant undertaking’s profitability should be assessed on the assumption that it receives the same gross termination charges as its competitors. Range of services 6.167 Ofcom argues that supply side factors support the aggregation of the set of services across which firms compete to undertake a margin squeeze test (see paragraph 6.52 et seq above). 6.168 Accordingly the test takes into consideration the possibility that NTS service providers could run a business offering 0845 voice or data service only, or a bundle of number ranges (on voice or data), or on all number ranges (on voice or data). 6.169 Ofcom therefore carried out the test on a number of different service increments, to reflect these different possible entry models: • 0845 (voice); • 0845 (data); • 0845 and 0870 (voice); • 0845 and 0870 (data); • 0845, 0870 and 0800 (voice); and • 0845, 0870 and 0800 (data).281 6.170 However, even if some TCPs specialise in a narrow range of services, it would also be relevant to consider whether there were any barriers preventing TCPs from 281 Data obtained from BT operational system for terminating data NTS calls indicate a small volume of traffic minutes were terminated on the 0870 data range. Ofcom includes 0800 to reflect possible supply as part of a broader group of number ranges. 135 offering a wider range of services, if there appeared to be a “margin squeeze” on a narrower set of services (see paragraphs 4.128-4.140 above). Cost of capital 6.171 Ofcom has taken into account the cost of capital at both the network and retail level. At the network level, Ofcom has taken BT’s transfer charges which reflect a return on capital employed estimate for network assets. Ofcom has also considered Mean Capital Employed (“MCE”) in relation to assets at the retail level. Ofcom notes that the retail MCEs were either immaterial or negative (see relevant MCE statements in Annex 9). 6.172 Accordingly, Ofcom has not included these numbers within its estimate of costs (in the first instance where the numbers are immaterial, the numbers are considered de minimis for the purpose of the analysis and where they are negative, they are not considered meaningful for inclusion in the analysis). 6.173 The profit measure Ofcom is using is a current cost return figure for the relevant regulatory product groups’ profit and loss accounts extracted from BT’s regulatory accounting systems. This profit measure can be thought of as having the elements of an EBIT measure. The figure reflects the financial position of BT in respect of these services before interest and tax (the cost of capital of 12.5% is a pre-tax measure).282 In addition, depreciation allowances are included which is more reflective of an EBIT measure (as opposed to an EBITDA measure). Cost standard 6.174 For the reasons discussed at paragraph 6.60 et seq above, Ofcom used historical accounting data, rather than a forward looking DCF (discounted cashflow) analysis, to conduct its assessment. 6.175 The relevant cost standard for this analysis was (as for Ofcom’s analysis of the alleged margin squeeze on OCPs), FAC, for the reasons set out at paragraph 6.84 above. In practice, if it were found that BT failed the hosting margin squeeze on an FAC basis, it would be appropriate to consider tests relative to incremental or variable costs. Relevant time period 6.176 NCCN 500 was in force from 1 May 2004 to 31 December 2005 inclusive (20 months). However, the relevant period in which the hosting margin squeeze may have taken place (and therefore the period over which the test should apply) relates more closely to the period in which additional receipts earned from NCCN 500 could have influenced the level of payments made to NTS service providers. Many of the hosting contracts that were renegotiated subsequent to the introduction of NCCN 500, and which provided for higher payments, were of considerably longer duration than the period during which NCCN 500 was in force.283 6.177 Ofcom has also had regard to the availability of information from BT’s regulatory accounting systems, in particularly regulatory product group information for P346 282 The figure of 12.5% reflects the cost of capital incorporated into BT’s regulatory accounts for the period in question. 283 Some contracts have a minimum period of [], [] the period over which NCCN 500 rates were applied. 136 (data) and P056 (voice).284 Ofcom analysed revenues and costs for financial years 2004/05 and 2005/06 in relation to these product groups. At the time that Ofcom conducted its analysis, actual cost and revenue data were not available for the entirety of all relevant contracts (which in some cases ran until at least March 2007). There is additional complexity given that contract start dates differ, as do the length of the contracts. 6.178 Ofcom’s analysis was conducted for the period April 2004 to March 2005, on the grounds that this period captures the first 12 months in which payment levels would have risen reflecting the application of increased receipts from NCCN 500. Ofcom would not expect costs (with the possible exception of payments) and revenues to differ significantly between March 2005 and December 2005, and therefore considers the test over this 12-month period to be representative of the 20-month period under investigation. 6.179 Clearly, additional contracts are likely to have been renegotiated in the rest of the period. Ofcom has allowed for this in the margin squeeze test by assuming, as a sensitivity, that the NCCN 500 payments applied to all contracts (which represents a more stringent version of the test). 6.180 Ofcom notes that the shorter the period, the less likely costs are to be variable. However, if the hosting margin squeeze test is passed on the basis of FAC, it is not necessary also to carry out the test on the basis of a subset of costs that are variable over the period of one or two years. Data and results 6.181 To apply the hosting margin squeeze test, Ofcom used BT’s cost and revenue data for NTS voice and data call termination/hosting, as reflected in BT’s regulatory product groups P056 (voice) and P346 (data).285 The costs reflected in these regulatory product groups are derived using the principle of cost causation, which should ensure the attribution of all incremental costs and some common costs to these product groups. Using the information supplied by BT, Ofcom sought to establish the profitability of the principal number ranges on an FAC basis. This process included attributing shared costs, weighted by traffic volumes, between number ranges, and in the case of data, attributing shared costs between metered and unmetered narrowband internet access. 6.182 Ofcom attributed shared costs within each product group to number ranges by volume (e.g. shared network costs). Ofcom initially carried out the test based on FAC because, if this test is passed, this is sufficient to establish the absence of a margin squeeze (given that the FAC methodology is reasonable). If the FAC test is failed, it will be necessary to carry out an additional test based on AVC. Revenues 6.183 The gross termination charge received by BT’s competitors for NTS call termination services reflects the equally efficient operator test as discussed at paragraph 6.37 et seq above. The charges have been estimated by taking monthly prices generated by the NTS Calculator (see paragraph 2.27 et seq above) and then averaged to generate average annual termination revenue for the period 2004/05. 284 285 For further explanation of these product groups, see paragraph 6.347. The process is more fully described in paragraph 6.338 et seq. 137 6.184 In addition to NTS call termination and NTS hosting, BT provides a number of value added services to NTS service providers and receives additional revenue from these. Costs 6.185 BT’s NTS call termination and hosting costs have been used to reflect the costs that an equally efficient competitor would need to cover using its hosting margin and other revenue sources referred to above. 6.186 Termination costs are based on those costs reflected by BT, on a current cost accounting (“CCA”) basis, within its regulatory product groups P056 (voice) and P346 (data). 6.187 Ofcom required BT to provide information from its regulatory costing systems to enable it to determine the (cost-based) prices at which BT would potentially have charged a new entrant hosting business seeking to buy in IP network services (instead of having to invest in an end-to-end network which would be uneconomic if only NTS call termination services are offered). 6.188 As discussed below (in relation to Ofcom’s analysis of C&W’s allegation of discrimination) Ofcom considers that internal transfer prices derived from BT’s regulatory costing systems are ‘soft’ charges, in that they do not in themselves represent a cost to BT. Such ‘soft’ charges may approximate to the cost to BT of self-providing network services, to the extent that they are used for internal management accounting purposes and so are intended to provide signals for decision-making within BT. However, this in itself means that they may therefore be influenced by other factors underlying BT’s internal accounting policies and do not therefore necessarily accurately reflect the true cost to BT of self-providing network services. 6.189 However, Ofcom considers that for the purpose of conducting the hosting margin squeeze test, the cost of buying in IP network services is a necessary element of the cost stack and that internal transfer prices represent the best estimate of costs that were available during Ofcom’s investigation. Ofcom understands that use of the IP network is transfer charged into downstream product groups on the basis of that network component’s CCA cost for the corresponding year. This cost base includes a return on the component’s mean capital employed using its regulatory cost of capital. 6.190 Once the number translation has been performed, the call will be routed to a geographic number for termination. Ofcom has estimated the costs of onward conveyance by assuming that the equally efficient hosting business could buy in the necessary network services at the relevant BT tariff rather than being required to invest in its own network. 6.191 Ofcom has estimated hosting costs by considering BT’s hosting costs. As discussed at paragraph 6.183 above, the test applies to an equally efficient operator – in this case, one that is providing the same quality of hosting service – otherwise the test could be either unnecessarily restrictive or potentially too easy to pass. For example, an estimate of hosting costs based on a new entrant offering a basic hosting service could understate the costs of providing BT’s quality of service and hence fail to detect a margin squeeze. Accordingly, the equally efficient operator costs should reflect BT’s hosting costs. 138 6.192 Ofcom estimated payments to NTS service providers in two ways. 6.193 First, Ofcom took actual payments recorded in BT’s profit and loss accounts for the regulatory product groups P056 and P346. Ofcom considers that this approach provides a lower bound estimate for the payments competitors would need to match in the market, since this figure is a weighted average of payment levels negotiated before NCCN 500 and those negotiated after NCCN 500. This variant of the test is referred to as Model A in the subsequent discussion. 6.194 Second, Ofcom considered BT’s contracts with its largest NTS service provider customers which were renegotiated after NCCN 500 came into effect.286 Ofcom took a sample of BT’s largest contracts with NTS service providers (by revenue) and calculated the weighted average level of payments made under these contracts and applied them to all contracts as if they had also been renegotiated at the higher (average) payment level. Ofcom estimated this average for 0844/0845, 0870/0871 and for all number ranges. This estimate could be interpreted as the upper bound of higher payment levels associated with NCCN 500. This variant of the test is referred to as Model B in the subsequent discussion. 6.195 The results of the hosting margin squeeze test are as follows: Table 15: Hosting margin squeeze results for 2004/05 (voice), Model A Model A (outpayments from BT FAC P+Ls) Revenues Gross termination charge Advanced stat packs revenues SP revenues (top up payments) 0845/0844 0845/0844 & 0870/0871 All number ranges £ million [] Costs Termination and hosting costs Retail costs Outpayment Profit/loss Table 16: Hosting margin squeeze results for 2004/05 (voice), Model B Model B (outpayments from BT hosting contracts) Revenues Gross termination charge (BT to OCP) Advanced stat packs revenues SP revenues (top up payments) 0845/0844 0845/0844 & 0870/0871 All number ranges £ million [] Costs Termination and hosting costs Retail costs Outpayment Profit/loss 6.196 Ofcom has applied relevant FAC costs. These include: network transfer charges (at tariff), network transfer charges (at cost) (both including a cost of capital item), and retail costs (although fixed assets are relatively small in this category and hence a 286 Source: BT’s response to section 26 Notice of 24 April 2006. 139 cost of capital is not included in the cost stack), as reported by BT in its profit and loss account for regulatory product group P056. The results of the Model A test indicated that an equally efficient operator, receiving gross termination charges from BT and paying out the same payment levels as BT, receives a hosting margin that exceeds the termination and hosting costs of service provision in respect of key service increments (the most relevant combinations of number ranges). 6.197 Ofcom notes similar results apply when Ofcom uses payment levels taken from BT contracts (Model B). However, as expected, BT’s margins on Model B are lower, reflecting the fact that payments (taken from a sample of hosting contracts renegotiated since NCCN 500) all reflect higher payment rates and therefore represent an upper bound on the level of payments that would need to be met by a hosting competitor in this market. 6.198 Ofcom also applied the test to data NTS call termination. As described above, when it was looking at BT’s voice business, Ofcom applied sensitivities in its analysis of costs by considering an upper and lower bound estimate, which we refer to as Model A and Model B respectively. However, Ofcom could not apply the same approach in data, as the information provided by BT on revenue share payments was insufficient to enable Ofcom to conduct an equivalent Model B analysis, that is to say a lower bound estimate based on a sample of contracts (as opposed to actual revenues derived from financial data provided by BT).287 Table 17: Hosting margin squeeze results for 2004/05 (data) Model A (outpayments from BT FAC P+Ls) Revenues Gross termination charge Port rentals internal external (includes internal) 0845/0844 0845/0844 & 0870/0871 All number ranges £ million [] Costs Termination and hosting costs (transfer charges at tariff and interconnect IP platform (allocated between unmetered and metered Retail costs (allocated between unmetered and metered) Outpayment Profit/loss 6.199 Ofcom has applied relevant FAC costs (including payments, cost of capital and port rental revenues) reported by BT in its profit and loss account for regulatory product group P346, and adjusted to reflect FAC. Ofcom considers that an equally efficient operator would be able to compete and earn a hosting margin that covers the costs of data call termination and hosting for key service increments. 6.200 BT passes the test for 0845, 0845 and 0870 and all number ranges service increments. The profit from 0845 exceeds the profit from 0870, for the joint 287 In its section 26 Notice of 24 April 2006, Ofcom required BT to provide internal documentation relating to the top 10 NTS hosting agreements, by value, which BT entered into during the period that NCCN 500 was in force. Ofcom did not ask BT to provide information separately for voice and data customers. None of the agreements in respect of which BT provided documentation related to the provision of data NTS call termination, e.g. to an ISP. 140 0845/0844 and 0870/0871 service increment due to significant differences in traffic volumes (5.43 billion traffic minutes compared to 8 million traffic minutes). []. Summary of results 6.201 The results of the hosting margin squeeze test indicate that an equally efficient operator providing voice or data call termination and hosting services could earn a hosting margin that covers the costs of termination and hosting for the 0845, 0845 and 0870 and all number range service increments during 2004/05. 6.202 The cost standard applied to the termination and hosting costs is FAC, which, for the multi-product firm, approximates ATC. Accordingly, the fact that the test shows that revenues exceed ATC is good evidence that no margin squeeze has taken place.288 6.203 Ofcom also considered the test at three different levels of service increment to determine whether the test results differed according to scope of entry. 6.204 In conclusion, Ofcom considers that the prices notified in NCCN 500 did not lead to a hosting margin squeeze and concluded that all three levels are profitable. Responses to draft decision: hosting margin squeeze 6.205 BT and C&W commented on this element of Ofcom’s analysis. BT 6.206 BT argues that Ofcom’s hosting margin squeeze analysis fails to take into account: “the ability of other vertically integrated communications providers […] to use the additional revenue earned by them from the higher retail charges that applied to their subscribers during the relevant period to increase revenue share payments to NTS service providers.”289 6.207 BT goes on to say that: “This retail revenue clearly available to those other vertically integrated communications providers was not available to BT during the relevant period.”290 6.208 Other than its comments on the hosting margin squeeze, BT did not comment on Ofcom’s analysis of its conduct as set out in Section 6 of the draft decision. C&W 6.209 Referring back to its comments on market definition, C&W reiterates its view that Ofcom’s analysis of the hosting (side of the) market is insufficiently developed.291 288 This is because as noted at paragraph 6.84, a test that is passed based on an analysis of FAC, provided the attribution is reasonable, is usually taken as strong evidence that there is no margin squeeze. It is therefore only in cases where such a test is failed that it is important to undertake an analysis based on variable or incremental costs. 289 BT’s response to the draft decision, paragraph 4.4. 290 BT’s response to the draft decision, paragraph 4.5. 291 C&W’s response to the draft decision, page 38. 141 6.210 C&W states that it understands the purpose of Ofcom’s “Model A” is “to provide the lower bound estimate that is the very least payment that a competitor would need to make to compete”. C&W submits that there are several problems with this model: • it uses a weighted average price for the 2004/05 financial year, thereby diluting the impact of NCCN 500, which was in force only for 11 months of this period; • it does not capture changes in revenue shares over the year; and • it does not capture changes in individual contracts “to determine where margins altered on particular contracts”.292 6.211 C&W states that it understands the purpose of Ofcom’s Model B is to provide a higher bound estimate, but that “the point of this test is not made clear by Ofcom” and that Ofcom should have used the highest payments that BT made to a service provider, to ensure that “the margin squeeze was not being undertaken on any particular contract”.293 6.212 C&W states that Ofcom’s approach to assessing a possible margin squeeze in data hosting is unclear and that if, as it appears, Ofcom has used only the higher bound estimate, Ofcom “seems to have bent over backwards to be favourable to BT”. C&W further submits that if it did not have the data to construct a full model for data NTS call termination, Ofcom should have required BT to provide it.294 Ofcom comments on responses 6.213 In addition to its specific comments on Ofcom’s hosting margin squeeze analysis, which are considered in the following paragraphs, C&W raised a number of issues with Ofcom’s approach to the hosting margin squeeze in the context of Ofcom’s proposed market definition. We deal with these at paragraphs 4.218-4.231 above. 6.214 Ofcom considers that taking weighted average prices for the 2004/05 financial year (rather than for the 11-month period of May 2004-April 2005) would not materially alter Ofcom’s finding that BT’s conduct did not lead to a margin squeeze in NTS hosting. This is based on the fact that BT passed the hosting margin squeeze test by a sizeable margin over the 12-month period over which the analysis was undertaken. Ofcom notes that operational costs (i.e. efficiencies) are unlikely to have differed materially between April 2004 (when NCCN 500 was not in effect), and the remaining 11 months of the 2004/05 financial year (which lies within the period in which NCCN 500 was in effect). Ofcom considers that on a weighted average basis, over the 12 month period within the financial year 2004/05, the analysis and findings are robust. Even if the weighted average price (as opposed to the weighted average cost) was found to be different from the NCCN 500 price (i.e. the price for the 11 months when NCCN 500 was in force, as opposed to the weighted average for the whole year), the margin by which the test was passed was sufficiently large for the result to be robust. 6.215 Ofcom notes that when the NCCN 500 price increase is applied to all contracts for the entire period (the strict Model B test), BT passes the test. As set out at paragraph 6.194 above, Ofcom took a sample of BT’s largest contracts with NTS service providers that were renegotiated after NCCN 500 came into effect (by revenue) and 292 Ibid. Ibid. 294 C&W’s response to the draft decision, page 39. 293 142 calculated the weighted average level of payments made under these contracts and applied them to all contracts as if all contracts had also been renegotiated at the higher (average) payment level. Accordingly the strict Model B test could be interpreted as the upper bound of higher payment levels associated with NCCN 500. Model B therefore explicitly assumes that a weighted average level of payments made under these contracts was passed on to all contracts in each of the twelve months for the 2004/05 financial year. This provides strong evidence that the measurement of the weighted average price in Model A is not material to the finding of no margin squeeze. 6.216 Ofcom considers that it is correct to assess a hosting margin squeeze over a group of hosting contracts, and not, as suggested by C&W, on individual contracts, because providers of NTS termination and hosting were competing with each other for more than one contract over the period under investigation. In practice, competing NTS termination and hosting businesses do not compete solely for a specific contract, but will be in the market to contract with as many service providers as possible. Ofcom considers that a narrow test based on individual contracts as proposed by C&W would therefore be inappropriate. 6.217 C&W believes that Ofcom has “bent over backwards” to construct a test favourable to BT for data call termination. This is not the case. In order to obtain the information it needed to carry out the hosting margin squeeze tests, Ofcom required BT to provide details of its top 10 hosting contracts, by value. None of these contracts was for the provision of data termination. Given the relative sizes of the voice and data businesses, and the results set out in Tables 15–17 above, Ofcom considers that it is reasonable to conclude that there was no hosting margin squeeze in data call termination. Requiring BT to provide additional information to construct a full model for data would not, therefore, have led to a different outcome. Ofcom does not accept the claim made by C&W that Ofcom has not sought the relevant information to complete its analysis of a hosting margin squeeze on BT. 6.218 Ofcom notes that the strict test would allow for a significantly higher outpayment level. The higher bound test is based on pass-through of a weighted average level of payments in response to NCCN 500 and applied as if the weighted average payment was paid in respect of all contracts with service providers (and not just paid in respect of the sample of contracts from which the weighted average payment was derived – see paragraphs 6.194 and 6.215 above). Ofcom therefore considers that even if some data NTS termination/hosting contracts had fallen within the top 10, it would not have reached a different conclusion. Specifically, Ofcom notes that outpayments would need to have increased by an amount roughly equivalent to a third of BT’s gross termination revenues in order for BT to fail the strict test. Given this order of magnitude by which BT passes the lower bound test, Ofcom is satisfied that BT passes the hosting margin squeeze tests jointly and separately for voice and data. 6.219 BT has claimed that other communications providers could fund higher outpayment levels from higher retail prices. Ofcom does not consider that this is relevant to its hosting margin squeeze test which, as set out at paragraph 6.162 et seq above, is concerned with whether higher NCCN 500 (wholesale) prices have been passed through in the form of higher outpayments in such a way as to force an equally efficient operator seeking to match BT’s higher outpayments to no longer cover the efficient operator’s (i.e. BT’s) cost of NTS termination and hosting. Ofcom notes that, were competing non-BT vertically integrated businesses to raise their retail prices, this would make it more difficult for them to compete with BT, which has SMP in call 143 origination. Accordingly, Ofcom does not consider that this is a relevant consideration for the application of the equally efficient hosting margin squeeze test. Alleged discrimination Alleged discrimination: the allegation 6.220 In its complaint, C&W submitted that the price paid by BT (as an OCP to itself as a TCP) for NTS call termination under NCCN 500 was less than the charge that it levied on C&W when it provided identical services to C&W, and was therefore discriminatory. 6.221 C&W’s arguments on discrimination are set out at paragraphs 8.16-8.59 of its submission to Ofcom of 15 March 2005. 6.222 C&W submitted that there were two possible tests for determining whether NCCN 500 constituted discrimination: “Test 1: […] the charge paid to [C&W] where [C&W] terminates the call must equal the internal transfer charge within BT where BT itself terminates the call”; and “Test 2: […] the charge BT proposes to impose under NCCN 500 for terminating the call must equal the internal transfer charge paid by BT itself for terminating its own calls.”295 6.223 C&W submitted that “if either of these tests reveals a difference in the price charged or paid between BT and [C&W], BT will be engaging in discriminatory pricing”;296 and that: “as a matter of logic, BT must fail one of these tests.”297 6.224 C&W submitted that the nature of the abuse would be determined by whichever of the two tests BT were to fail: “a failure of Test 1 would indicate that BT is abusing its dominance in NTS call origination by paying a competing operator (C&W) less than it pays itself for the equivalent call termination service”;298 since: “price discrimination is only legitimate if it reflects differing costs.”299 6.225 Were BT to fail Test 2, on the other hand, this: 295 C&W submission to Ofcom of 15 March 2005, paragraph 8.21. C&W submission to Ofcom of 15 March 2005, paragraph 8.22. 297 C&W submission to Ofcom of 15 March 2005, paragraph 8.23. 298 C&W submission to Ofcom of 15 March 2005, paragraph 8.25. 299 C&W submission to Ofcom of 15 March 2005, paragraph 8.27. 296 144 “[suggests that BT was] abusing its market power in call termination […] by charging different termination rates to different originating operators (C&W and itself) with no underlying difference in cost.”300 6.226 The result of BT’s conduct would also depend on which of the tests was failed. The implication of a failure of Test 1 would be that: “BT will be able to leverage its dominance into the “downstream” market of NTS call termination since it will be in a position to use the additional revenues generated relative to its competitors to offer more lucrative fee share arrangements to NTS service providers;” whereas if Test 2 were found to be the relevant standard: “The effect of this abuse would be to entrench BT’s market position in NTS call termination at the expense of its competitors.”301 6.227 C&W concluded that: “Regardless of whether BT failed Test 1 or Test 2, the outcome of BT’s discriminatory pricing would be a reduction of competition in NTS call termination.”302 Alleged discrimination: legal framework 6.228 Article 82(c) of the EC Treaty identifies discrimination as conduct that may constitute an abuse of a dominant position in the following terms: “applying dissimilar conditions to equivalent transactions with other trading parties, thereby placing them at a competitive disadvantage.”303 6.229 In its draft guideline on conduct, the OFT states that the most direct form of discrimination occurs through the prices charged to different sets of customers and cites two basic forms of discrimination that might fall within the scope of Article 82 and section 18: (i) “an undertaking might charge different prices to different customers, or categories of customers, for the same product – where the differences in prices do not reflect any differences in relative cost, quantity, quality or any other characteristics of the products supplied;” or (ii) “an undertaking might charge different customers, or categories of customers, the same price even though the costs of supplying the product are in fact very different. A policy of uniform delivered prices throughout the country, for example, could be discriminatory if differences in transport costs were significant.”304 300 C&W submission to Ofcom of 15 March 2005, paragraph 8.31. C&W submission to Ofcom of 15 March 2005, paragraph 8.32. 302 C&W submission to Ofcom of 15 March 2005, paragraph 8.33. 303 Section 18(2) (c) Competition Act uses identical wording. 304 OFT 414a,paragraph 3.1. 301 145 6.230 In Alpha Flight Services/Aeroports de Paris the CFI and the ECJ upheld the Commission’s decision that the operator of the two Paris airports had abused its dominant position by imposing different fees on ground handlers for the provision of the same services, without justification.305 6.231 In British Airways v Commission the CFI found (and its judgement was upheld by the ECJ) that the performance reward scheme operated by British Airways (“BA”) resulted in some cases in different rates of commission being applied to an identical amount of revenue generated by the sale of BA tickets by two travel agents.306 The CFI held that: “In those circumstances, the Commission was right to hold that BA’s performance reward schemes constituted an abuse of BA’s dominant position…….in that they produced discriminatory effects within the network of travel agents established in the United Kingdom, thereby inflicting on some of them a competitive disadvantage within the meaning of subparagraph (c) of the second paragraph of Article 82 EC.”307 6.232 In its draft guideline on conduct, the OFT indicates that discrimination may be an abuse where it is exclusionary, for example where a dominant undertaking uses a discriminatory pricing structure to set predatory prices and/or to set discounts which have the effect (or likely effect) of foreclosing all, or a substantial part, of a market; alternatively where a vertically integrated undertaking is dominant in an upstream market and a competitor in a related downstream market and it uses discriminatory pricing to apply a margin squeeze that distorts competition in the downstream market.308 6.233 The OFT also states that price discrimination may be abusive where it is exploitative, in particular allowing an undertaking to exploit market power by charging excessively high prices to certain customers.309 6.234 The OFT notes that: “when considering whether price discrimination is an abuse, it is often relevant to consider whether the pricing structure in question allows the efficient recovery of fixed costs and expands demand substantially or opens up new market segments.”310 6.235 The OFT guideline The application of the Competition Act in the telecommunications sector suggests a similar effects-based approach to discrimination: “Where discounts are offered to certain customers that do not reflect underlying cost differences they represent a form of price discrimination. There may be legitimate reasons for offering such discounts, even where they do not reflect underlying cost 305 Case T-128/98 [2000]ECR II-3929, Case C-82/01 P [2002] ECR I-9297 Case T-219/99 307 Ibid, paragraph 240. 308 OFT 414a. 309 OFT 414a, paragraphs 3.4 and 3.5 310 OFT 414a, paragraph 3.6 306 146 differences. They may, for example, represent an efficient way of recovering fixed or common costs.”311 6.236 Oftel followed this approach in BT broadband promotion in conjunction with BSkyB.312 In this case, while Oftel found that there was prima facie discrimination by BT in relation to the joint promotion in favour of those of its customers who were also BSkyB subscribers, there was no effect on competition. Oftel cited the OFT’s guidelines on conduct and on competition law in the telecommunications sector, finding that : “It is therefore not an abuse in itself for BT to price discriminate between its customers. However, it would be abusive if it were shown that this discrimination had an effect on competition in the relevant market(s).” 6.237 In the BHB case, the Court of Appeal took the view that while ATR and Phumelela were paying considerably different amounts for similar transactions, it did not necessarily follow that this placed ATR at a competitive disadvantage distorting competition: “Certainly it made ATR’s operation less profitable than it would have otherwise been, but ATR’s ability to compete with Phumelela (or, so far as material, with others) remained intact. Although they impact on each other, profitability and competitiveness are two different things…”313 6.238 In the Court of Appeal’s view, there needs to be proof that a pricing differential actually or potentially distorts competition between the parties subject to that differential.314 Alleged discrimination: analysis 6.239 As set out above, C&W submits that BT engaged in price discrimination in one of two possible ways, both of them constituting anti-competitive price discrimination leading to a reduction of competition in NTS call termination. 6.240 Both of the tests set out by C&W propose a comparison with “BT’s internal transfer charge”. Test 1 compares BT’s internal transfer charge with the charge paid to C&W where C&W terminates the call; Test 2 compares BT’s internal transfer charge with the prices notified in NCCN 500. 6.241 In the analysis set out below Ofcom addresses the following issues: a) establishing the facts about the level of the transfer charges, 311 OFT 417, paragraph 7.30. Decision of the Director General of Telecommunications 15 May 2003 Complaint submitted under the Competition Act 1998 (The Act) by NTL Group Ltd (NTL) alleging an infringement of the prohibitions imposed by Section 2(1) and 18(1) of the Act by British Telecommunications Plc (BT). 313 Judgment of the Court of Appeal 2 February 2007 Case A3/2006/0126 (1) Attheraces Limited (2) Attheraces (UK) Limited v (1) The British Horseracing Board Limited (2) BHB Enterprises Plc EWCA Civ 38, paragraph 270. 314 Case A3/2006/0126, EWCA, Civ 38, paragraphs 277-278. 312 147 b) considering the significance of comparisons between transfer charges and prices in NCCN 500, and whether this constitutes a separate possible abuse of a dominant position under the Act and Article 82. 6.242 In the course of its investigation, Ofcom requested a variety of information which was extracted by BT from its regulatory costing systems (including unaudited downstream financial information).315 In addition, Ofcom required BT to provide copies of its management accounts reporting on NTS call termination.316 6.243 Ofcom found that the relevant BT downstream regulatory product group financial statements did not reflect any internal transfer charge revenues for NTS call termination in respect of BT-to-BT NTS calls, i.e. a specific charge from one part of BT to another in respect of a service equivalent to the services for which price increases were notified in NCCN 500. This resulted from BT’s accounting policy of reflecting the full value of retail turnover for BT-to-BT NTS calls in the relevant downstream regulatory product groups. 6.244 BT is not required to report transfer charges given that Ofcom has not previously made an ex ante finding that BT has SMP in the NTS call termination/hosting markets. The following discussion explains why the transfer charge, or absence of one, in any event, is not determinative in the context of this investigation. 6.245 BT’s management accounts appeared to reflect internal transfer charge revenues in respect of some of the services for which price increases were notified in NCCN 500. BT argued in its response to Ofcom’s section 26 Notice of 24 August 2005 that these internal transfer charges were not the relevant basis for Ofcom’s analysis. In response to the question: “Explain what prices BT charges to itself for each of the call types notified in NCCN 500, i.e. the internal transfer charge for NTS call termination where BT is both the terminating and originating provider of the call. Provide information on the relevant charges for each of the relevant call types for the period: • before 1 May 2004; • immediately after 1 May 2004; • following any further changes to BT’s prices since NCCN 500 became effective”. BT explained that: “The relevant internal transfer charges in answer to this question are the internal charges set out in BT's regulatory (”AS”) accounts. We have provided this information below under the heading ‘AS accounts – internal transfer charge.’ BT does not consider the internal transfer charges in the management accounts to be the relevant transfer charge for any purpose other than internal accounting. BT has, however, also provided information on the internal charges in BT's management 315 316 See Table 2 above. Ofcom section 26 Notice of 28 June 2005. 148 accounts to address questions raised by Ofcom on the 18 August conference call on management account information. For the avoidance of doubt, the only internal charge that we believe is relevant to the Competition Act investigation is the internal charge set out in the AS accounts.”317 6.246 Ofcom has considered the sense in which BT may be said to have discriminated in the charges it set for NTS call termination while NCCN 500 was in force. 6.247 The transfer charge recorded in BT’s management accounts for the provision of NTS call termination to itself is different from the charge levied on other OCPs for the same service. 6.248 It is of concern to Ofcom that BT itself may have believed that it has “discriminated” in this sense: []318 6.249 This was also indicated in a statement made by BT shortly after NCCN 500 was issued: “Colin Annette said that BT had issued NCCN 500, which increased termination payments for access to its 0845 and 0870 services from the 1st of May 2004. These rates are only applicable to non-BT originated minutes and BT Retail would continue to pay pre-May '04 rates.”319 6.250 Internal documentation supplied by BT confirms that BT’s understanding was that the charges notified in NCCN 500 would apply only to other OCPs: “We are only changing OLO [“Other Licensed Operator”, i.e. OCPs other than BT] originated 0845/0870/0820 call. The manner in which BT accounts for BT-BT calls remains unchanged.” “This does not include traffic from BT’s own customers, as this is strictly controlled by the NTS regulatory pricing structure.”320 6.251 Ofcom therefore investigated carefully BT’s regulatory costing systems to assess the actual internal transfer charge paid by BT as explained above. Ofcom notes BT’s response in relation to the actual internal transfer charge as set out at paragraph 6.245 above. Ofcom agrees that any transfer charge recorded in BT’s management accounts for the provision of NTS call termination is inevitably a ‘soft’ charge, and does not represent the true cost to BT of purchasing NTS call termination from itself. This is because it is simply an amount recorded in the internal accounts of a vertically integrated firm (BT). 317 Annex A to BT response of 26 August 2005 to Ofcom Notice of 24 August 2005. []. 319 NTS Focus Group meeting notes and actions, 20 April 2004, published at: http://www.ofcom.org.uk/telecoms/groups/nts_focus/notes/nts20040420. 320 email from [] to [], 5 April 2004, numbered DK450 and briefing paper dated 10 June 2004, volume 1 tab 13, BT response of 13 May 2005 to Ofcom section 26 Notice of 22 April 2005 – see Table 2 above. 318 149 6.252 The recording of such a charge might be likened to the transfer of coins from one pocket to another – it cannot affect the overall value of the company unless the level of the transfer charge affects pricing or other commercial decisions by BT in its role as an OCP, TCP or NTS service provider. It may be contrasted with a ‘hard’ charge, such as that paid by other OCPs to BT for NTS call termination, which is a real cost to those businesses (see also paragraph 6.188 et seq above), and so affects their commercial decisions. 6.253 Therefore, and crucially, the fact that such a charge may exist in BT’s management accounts does not tell us whether or not other OCPs could have been placed at a competitive disadvantage by paying a charge different from that ‘soft’ charge. There was not robust evidence of a “hard” charge to BT that was different from the charge that applied to BT’s competitors. 6.254 Ofcom considered whether other OCPs would have been placed at a competitive disadvantage by NCCN 500, i.e. whether there is discrimination leading to a competitive disadvantage which is of an exclusionary nature. 6.255 Cases involving internal transfers have traditionally been looked at from the point of determining whether a margin squeeze is occurring which is foreclosing the relevant market to competitors. 6.256 In Genzyme, the CAT upheld the OFT’s finding of an abuse of Genzyme’s dominant position in the upstream market for “effective treatments for Gaucher’s disease” based on a margin squeeze abuse of pricing its Cerezyme drug to its competitors at a price higher than it priced to its downstream subsidiary, thereby reserving the downstream market to itself.321 6.257 On the facts of this case, in order to address the question of competitive disadvantage that could arise from a difference in charges that applied between BT and other OCPs, Ofcom has first sought to identify the maximum charge for NTS call termination that BT could have charged itself while covering the costs of originating, conveying and terminating an NTS call on its network, given the retail price of that call. 6.258 If this maximum is below the charge which BT made to other OCPs for NTS call termination then, if BT had faced the same ‘hard’ charge for NTS call termination as other OCPs, it would have made a loss on these calls. It can then be inferred that the implicit charge that BT made to itself for NTS call termination, consistent with covering the end-to-end costs of the call on this basis, must have been below the charge made to other OCPs for the same service. 6.259 However, in considering whether there has been discrimination leading to a competitive disadvantage or, equivalently, an anti-competitive margin squeeze, it is necessary to consider carefully the range of services over which the test should be conducted. This may be broader than just the single service in which there is a difference in pricing. 6.260 Therefore, in determining whether BT’s charges for NTS call termination, as notified in NCCN 500, might amount to “applying dissimilar conditions to equivalent 321 [2004] CAT 4, 11/3/2004. The OFT’s decision was also based on a finding of anti-competitive bundling (Genzyme’s practice of including Homecare Services in the NHS List price for Cerezyme) but this aspect of the OFT’s decision was not upheld by the CAT. However, the CAT found that Genzyme’s bundling practice facilitated the margin squeeze abuse. See for example paragraph 654 of the judgment. 150 transactions with other trading parties, thereby placing them at a competitive disadvantage”, Ofcom considers that the relevant test is whether BT would have been able to make a profit had it paid the charges notified in NCCN 500, taking into account profits earned on all the relevant services i.e. whether BT’s conduct led to the operation of a margin squeeze on other OCPs. As set out above, Ofcom has assessed whether the prices notified in NCCN 500 led to a margin squeeze on OCPs, and has concluded that they did not. 6.261 Notwithstanding BT’s statements on NCCN 500, given the results of the margin squeeze tests, Ofcom did not find compelling evidence to demonstrate that BT had discriminated between itself and other OCPs putting them at a competitive disadvantage. 6.262 Ofcom has also considered whether the application of NCCN amounted to excessive pricing (see paragraphs 6.292-6.477 below). Responses to draft decision: Ofcom’s analysis of alleged discrimination 6.263 Two respondents, C&W and Thus, commented on Ofcom’s analysis of alleged anticompetitive price discrimination. BT did not comment on Ofcom’s assessment of C&W’s allegation of anti-competitive price discrimination. C&W 6.264 C&W submits that, despite evidence that BT was not paying the same rates as other providers for “exactly the same transactions”, Ofcom “fail[s] to carry out a robust analysis” to establish precisely what BT was charging itself.322 6.265 C&W submits that Ofcom could have used information from BT’s regulatory accounts and/or from BT’s management accounts to establish a transfer charge as a relevant point of comparison for an analysis of discrimination.323 6.266 C&W submits that Ofcom gives no consideration to whether the charge increases imposed by BT were “objectively justified”, and submits that they could not have been so if their stated purpose was to increase revenue.324 6.267 C&W states that Ofcom: “seems to have taken the view that the dissimilar conditions [i.e. alleged price discrimination] must have an exclusionary effect in order to constitute abuse of a dominant position.”325 6.268 C&W submits that Ofcom has erroneously based its opinion on the OFT’s draft guidelines. C&W believes that: “the OFT says that discrimination which is exclusionary, and it gives margin squeeze as an example, is likely to breach the prohibition. But this is not a maximum test, it is merely an example.”326 322 C&W’s response to the draft decision, page 22. C&W’s response to the draft decision, page 25. 324 C&W’s response to the draft decision, page 26. 325 C&W’s response to the draft decision, page 27. 326 Ibid. 323 151 6.269 C&W states that the relevant test is not the approach suggested by the OFT, but the wording of Article 82(c), which refers to “competitive disadvantage” rather than exclusionary effect. C&W notes that the courts have found discrimination to be an abuse without demonstrating any exclusionary effect.327 6.270 In any case, C&W submits that, had Ofcom properly conducted an analysis of the alleged discrimination, “it would have proceeded to demonstrate the resulting impact on competition”, and submits that the revenue lost by BT’s competitors as a result of NCCN 500 would have led, “over time”, to exclusion, as: “we would not have been able to sustain that level of loss in the long run without having to make commercial decisions about our continued ability to stay in the market.”328 6.271 C&W concludes that: “Ofcom errs as a matter of law by failing to apply the discrimination test mandated in the Ch 2 prohibition and Article 82, together with relevant case law, properly or at all; Ofcom errs in its finding of fact when it says that it is unclear whether BT has as a matter of fact discriminated; Ofcom errs as a matter of law by relying solely on a margin squeeze test in attempting to assess the effect of discrimination.”329 Thus 6.272 Thus has concerns that: “Ofcom has been unable to make a clear statement about the discriminatory nature of the prices offered for access to BT’s 0845 and 0870 services to BT downstream retail business and altnets.”330 6.273 Thus states that Ofcom places “too much emphasis on what is and isn’t included within BT’s management accounts”, which Thus believes are irrelevant as “BT’s call origination, discount deductions and NTS retailing cost are well-understood regulated elements”.331 6.274 Thus states that: “Neither BT’s management nor regulatory accounts were designed to record this type of detailed internal charging, since no cost accounting or accounting separation obligations applied in respect of NTS termination. Hence it is unsurprising that Ofcom have failed to find what was never there in the first place; the problem is that Ofcom looked in the wrong place.”332 327 Ibid. C&W’s response to the draft decision, page 29. 329 C&W’s response to the draft decision, page 30. 330 Thus’s response to the draft decision, page 5. 331 Ibid. 332 Ibid. 328 152 6.275 Thus considers that it would have been possible for Ofcom to have based its analysis on “recognised cost elements on a per call basis”.333 6.276 In Thus’s view: “it would be unsafe…to reach a decision…. without first calculating the BT Group revenue and termination costs for on-net 0845 and 0870 calls using known cost elements from regulatory accounts.”334 6.277 Thus states that: “No objective justification has been put forward for the price rise and BT’s competitors were undoubtedly placed at a disadvantage.”335 Ofcom comments on responses 6.278 C&W argues that, because NCCN 500 only applied to non-BT OCPs, BT was paying the same amount to itself, as TCP, as it was before NCCN 500, which was less than it was charging other OCPs “for exactly the same transactions”. 6.279 Ofcom has considered the sense in which BT may be said to have discriminated in its charges to other TCPs for NTS call termination while NCCN 500 was in force. 6.280 Ofcom notes that there was what might be termed “apparent discrimination”, insofar as the transfer charge recorded in BT’s management accounts for the provision of NTS call termination services to itself is different from the charge made to other OCPs for the same service. Ofcom is concerned that BT internally appeared to consider that it had discriminated in this sense (see paragraph 6.248 above). 6.281 However, more detailed investigation by Ofcom, as set out at paragraphs 6.2516.252 above, found that any transfer charge recorded in BT’s management accounts for the provision of NTS call termination services is inevitably a ‘soft’ charge, not a “hard” charge, such as that paid by other OCPs to BT for NTS call termination, which is a real cost to those businesses. Ofcom notes BT’s response to Ofcom’s section 26 Notice of 24 August 2005 (see paragraph 6.245 above). 6.282 Moreover, Ofcom notes that the reference to the EC Access Notice quoted by C&W in fact describes the weaknesses of transfer charges for use in competition cases.336 It is precisely because transfer charges “could be used as a mechanism for disguising excessive pricing, predatory pricing or a price squeeze” that they should not be relied on in competition cases. A test for margin squeeze, for example, should therefore be based on the charges levied by the dominant firm to its downstream competitors rather than the transfer charges shown in its accounts. 6.283 C&W and Thus contend that Ofcom could have calculated an exact charge for comparison purposes on the basis of “BT’s regulatory accounts and/or from BT’s management accounts” (C&W) or “recognised cost elements” (Thus). 6.284 Ofcom notes that C&W’s alternative proposal, of inferring the transfer charge from the retail price of an NTS call less charges for call origination, relies on an 333 Ibid. Ibid. 335 Ibid. 336 C&W’s response to the draft decision, page 23, referring to paragraphs 120-127 of the Commission’s Access Notice (see footnote 231). 334 153 assumption that the retail price should be equal to cost. The existence of discrimination in the sense that the retail price is below cost including the charge that BT made to other OCPs for NTS terminations is not, however, sufficient to establish that such “discrimination” would impact on competition. In this regard, it may be noted that this test for discrimination is equivalent to the test for the existence of a positive margin on the service in question. However, in considering whether there has been a margin squeeze, it is necessary to consider carefully the range of services over which the test should be conducted. 6.285 Therefore, Ofcom considers that the relevant test is whether BT would have been able to make a profit had it paid the charges notified in NCCN 500, taking into account profits earned on all the relevant services i.e. whether BT’s conduct led to the operation of a margin squeeze on other OCPs. 6.286 The test that Ofcom outlined for discrimination was to assess whether the conduct lead to a competitive disadvantage that was of an exclusionary nature (see paragraph 6.255 above). Ofcom further explained that this meant whether the conduct distorted competition in the relevant downstream market. 6.287 Therefore, Ofcom’s test did not state that discrimination must have an exclusionary effect, but rather included a requirement that there should be an actual or likely effect on, or distortion of, competition. 6.288 Ofcom’s test is not inconsistent with the Chapter II prohibition or Article 82 and is consistent with case law, most notably the decision of the Court of Appeal in BHB (see paragraphs 6.237 et seq above).337 6.289 Ofcom did take into account existing case law when assessing the allegation of discrimination in this case and formed the view that as C&W’s allegation is based on BT’s internal transfer charge for NTS call termination, Ofcom would need to make an accurate determination of the actual hard internal charge (rather than soft inferred charge) in order to progress the analysis of discrimination in a traditional sense. 6.290 As set out in paragraph 6.252 et seq above, Ofcom has not been able to determine a “hard” charge representing the price actually paid by BT to itself. Ofcom has also rejected the proposal of inferring an internal transfer charge from the retail price of an NTS call less charges for call origination, relying on the assumption that the retail price should be equal to cost. In the absence of a ’hard’ charge, Ofcom applied the retail margin squeeze test as set out above to determine the actual or likely effect on competition had BT faced the NCCN 500 increased charge itself. 6.291 C&W and Thus suggest that Ofcom should have given greater consideration to whether BT’s charges were objectively justified. Since Ofcom did not find that BT’s charges constituted abusive discrimination (or margin squeeze), Ofcom did not need to go on to consider objective justification which could have been a defence for BT in an allegation of abuse. Ofcom is not of the view that any increase in prices without an objective justification, in itself, demonstrates that conduct would lead to a competitive disadvantage/distortion of competition. 337 Case A3/2006/0126, EWCA Civ 38, paragraphs 265-278. 154 Alleged excessive pricing Alleged excessive pricing: the allegation 6.292 In its complaint, C&W submitted that the prices notified in NCCN 500 were excessive. 6.293 C&W’s detailed arguments on excessive pricing are set out at paragraphs 8.34-8.39 of its submission to Ofcom of 15 March 2005. 6.294 C&W submitted that, “as a matter of logic”, BT must either be pricing call origination below cost, or over-recovering its call origination costs, and that the prices notified in NCCN 500 must therefore be excessive on the basis of the tests set out in Napp Pharmaceutical Holdings Ltd v Director General of Fair Trading (“Napp”).338 6.295 C&W further submitted that: “the prices set out in NCCN 500 are likely also to be excessive on other relevant measures, since, inter alia: • the price charged by BT to itself is less when terminating calls which originate on its own network. In the absence of objective justification for this difference in treatment, the clear inference must be that the higher price (i.e. the price charged to C&W) is excessive; • the price charged by C&W to BT (under the NTS Formula) for call termination services is less than that charged by BT.”339 Alleged excessive pricing: legal and economic framework 6.296 All firms that try to make profits have an incentive to raise price above cost if they can. In competitive markets, the ability to do so is constrained by the actions of rival firms attempting to undercut each other to gain market share. 6.297 A dominant undertaking may be able to raise prices above the competitive level, increasing its profits at the expense of its customers, who are worse off as a result. However, the dominant undertaking’s gain will be smaller than customers’ losses, except in the case of completely inelastic demand, as demand from some customers who would be willing to pay more than the cost of producing the service (and would do so if prices were at the competitive level) is choked off by the high prices. This means that excessive prices in retail markets would make society as a whole worse off, even if the firms’ profits and benefits to consumers were to be given equal weight. This is one reason giving rise to concern about excessive prices. 6.298 Excessive prices in wholesale markets may raise other concerns. In particular, an excessive wholesale price charged by a vertically integrated dominant undertaking to customers which are also its competitors in retail markets may have an exclusionary effect and so allow leverage of dominance from the upstream to the downstream market. Ofcom has examined possible distortion of competition in retail calls or NTS hosting in previous sections. 338 339 C&W complaint of 15 March 2005, paragraphs 8.37-8.38. C&W complaint of 15 March 2005, paragraph 8.39. 155 6.299 The ultimate concern about excessive pricing is an exploitative or exclusionary effect (see paragraph 6.297-6.298 above). As set out above, Ofcom has concluded that there is an absence of compelling evidence of adverse effects on competition. As regards any adverse effect on consumers, Ofcom notes that most of BT’s competitors did not increase NTS call prices following the introduction of NCCN 500 – although Ofcom notes that BT’s competitors may have been waiting for the outcome of Ofcom’s investigation and market review before taking any decision on changing their prices. Of the 13 communications providers who were required to provide Ofcom with details of retail prices over the period of NCCN 500, only one ([]) indicated that it had increased retail NTS call prices (see paragraph 5.113 et seq above).340 6.300 Ofcom notes that in response to NCCN 500, [] could have chosen to absorb the NCCN 500 price increase (as did BT’s other competitors) which would have resulted in a reduced margin.341 Ofcom notes that [] chose to raise its retail 0845 and 0870 call prices (which were already above BT’s retail prices – [] in the case of 0845) and in doing so raise its margins. Accordingly, Ofcom considers that although consumers were impacted by []’s decision to raise retail 0845 and 0870 call prices (by []ppm and []ppm respectively), the price increases implemented by [] appear to have been at least partly related to a commercial strategy by [] to raise its own margins and retail revenues (see paragraph 5.115 above). Most communications providers, however, did not raise their retail call prices for reasons including: maintaining their competitive position in the market (compared with BT and other OCPs), their stated position in the market as a price follower, and their own retail charges exceeding those of BT prior to NCCN 500 (paragraphs 5.118-5.121 above). 6.301 Ofcom has conducted the United Brands/Napp/Albion tests, as set out in paragraphs 6.305 et seq below. But, in light of the absence of compelling evidence of adverse effects on competition and limited evidence of adverse effect on consumers, Ofcom considers that compelling evidence of excessive pricing would be needed under these tests to support an infringement finding. 6.302 A price apparently above cost is not necessarily excessive. Even in competitive markets, prices may be high relative to cost at times, for example due to short-term fluctuations in demand or supply, or because the firm has developed an innovative product or process (and may for example be protected by intellectual property rights). 6.303 Excessive pricing by a dominant undertaking may be an abuse of a dominant position contrary to Article 82 of the EC Treaty and section 18(2) of the Act. 6.304 Section 18(2) of the Act provides that conduct by a dominant undertaking may constitute an abuse, in breach of section 18(1), as follows: “Conduct may, in particular, constitute such an abuse if it consists in • directly or indirectly imposing unfair…selling prices…” 6.305 In United Brands, the ECJ stated that: 340 Ofcom notes that the 13 communications providers from whom Ofcom requested pricing information accounted for a large proportion of the non-BT share of retail customers. 341 []. 156 “…charging a price which is excessive because it has no reasonable relation to the economic value of the product supplied would be…an abuse.”342 6.306 The ECJ then set out a two-step process for identifying excessive pricing, stating that the questions to be determined were: “whether the difference between the costs actually incurred and the price actually charged is excessive, and, if the answer to this question is in the affirmative, whether a price has been imposed which is either unfair in itself or when compared with competing products.”343 6.307 In Scandlines Sverige AB v Port of Helsingborg (“Scandlines”), the Commission applied the tests established in United Brands, assessing the costs actually incurred in providing the services supplied and making a comparison with the price actually charged for those services. 344 6.308 The Commission noted that even if it were assumed that the profit margin of the Port of Helsingborg was high or “excessive”, this would not be sufficient to conclude that the price charged bore no reasonable relation to the economic value of the services provided: “the mere fact that revenues may exceed costs actually incurred is not sufficient to conclude that the difference is ”excessive“ in the meaning of the first question posed by the Court in paragraph 252 of the United Brands judgment. In any event, the Commission considered that even if it were to be assumed that the difference is ”excessive“, the Commission would have to proceed to the next question as laid down in United Brands in the same paragraph, in order to determine whether the prices charged to the ferry-operators are unfair, either in themselves or when compared to other ports.”345 6.309 To determine whether the prices were unfair, the Commission first looked at a comparison of the port fees charged to ferry operators and the port fees charged to cargo vessels at the port and found that as the overall services provided to ferry operators and cargo vessels were not equivalent, it was not possible to draw a comparison between these prices. When the Commission looked at the ship fees component of the port fees charged, it found that Scandlines paid less per unit than other vessels. 6.310 The Commission then looked at the port charges paid by Scandlines at another port in an attempt to draw a comparison of the contested price to prices of comparable products offered by other companies on different markets. However the Commission: • could not find evidence that the services provided at the two ports were equivalent; • could not find evidence that the conditions of supply would be equivalent; and 342 Case 27/76 [1978] ECR 207 paragraph 250. Paragraph 252 (the ECJ went on to point out in the following paragraph that there were other ways “of selecting the rules for determining whether the price of a product is unfair”). 344 Case COMP/A.36.568/D3, 23 July 2004. 345 Case COMP/A.36.568/D3, 23 July 2004, paragraph 142, reiterated at paragraph 216. 343 157 • was not convinced that the regulated prices at the comparator port would be applicable to Helsingborg as the principles of the regulation were unknown. 6.311 The Commission then looked at the port fees charged by other ports even though it was of the view that it could not be considered that all ports provide and charge for the same basic services. There was no evidence that the prices charged to ferry operators at Helsingborg stood out in comparison to the other ports the Commission looked at. 6.312 In Corinne Bodson v Pompes Funèbres des regions libérées (“Bodson v Pompes funèbres”), the ECJ referred to a comparison between prices charged where there was competition, and those charged where there was a local monopoly, as one way of determining whether the price for a product is excessive.346 6.313 In the UK, the OFT has suggested a number of possible comparisons for assessing whether prices are excessive. Evidence of excessive pricing could be based on comparisons of prices with those of the same products in other markets or at other times, and with underlying costs, as well as evidence of excessive profits. Profits may be excessive when they are significantly and persistently in excess of the level which would be expected in a competitive market.347 6.314 This was the test applied by the OFT in its investigation into excessive pricing allegations in Napp. In its decision, the OFT was of the view that: “a price is excessive and an abuse if it is above that which would exist in a competitive market and where it is clear that high profits will not stimulate successful new entry within a reasonable period. Therefore, to show that prices are excessive, it must be demonstrated that (i) prices are higher than would be expected in a competitive market, and (ii) there is no effective competitive pressure to bring them down to competitive levels, nor is there likely to be.”348 6.315 On appeal, the CAT endorsed the OFT’s approach, stating that: “while there may well be other ways of approaching the issue of unfair prices under section 18(2)(a) of the Act, the Director’s starting point…..seems to us to be soundly based in the circumstances of the present case.”349 6.316 The CAT also endorsed the approach taken by the OFT in using the following comparators to determine whether a price is above the level that would exist in a competitive market: a) Napp’s prices with Napp’s costs; b) Napp’s prices with the costs of its next most profitable competitor; c) Napp’s prices with those of its competitors; and 346 Case 30/87, [1988] ECR 2479, paragraph 31. OFT 414a, paragraph 2.7. 348 CA98/2/2001, 30 March 2001. 349 Napp Pharmaceutical Holdings Ltd v Director General of Fair Trading [2002] CAT 1, paragraph 391. 347 158 d) Napp’s prices with prices charged by Napp in other markets.350 6.317 In Albion Water Limited v Water Service Regulation Authority (formerly Director General of Water Services) (“Albion”), the CAT stated: “…..where the issue is one of abuse, it is desirable that the question whether the price is “unfair” is to be assessed on the basis of a fully informed calculation of costs. In particular, the extent to which a price is unrelated to costs is relevant to the question whether that price is “unfair”.”351 6.318 The CAT also noted that to treat a class of customers the same as another class of customers where the costs are significantly different, would, in the absence of objective justification, be discriminatory and that, in those circumstances, the issue of whether the price in question was unfair should continue to focus primarily on the comparative situation of customers on the facts of that case (in that case customers for potable and non-potable water respectively).352 6.319 The United Brands two-stage test was recently applied by the Court of Appeal in the BHB case, where the Court of Appeal made a number of points on the test: a) First, the Court of Appeal noted that the ECJ judgment poses two questions: “The first is whether the difference between the costs actually incurred and the price actually charged is excessive. The second question is whether, if the first question is answered affirmatively, a price has been imposed which is either unfair in itself or when compared to competing products.”353 b) Second, the Court of Appeal noted that: “the central concept in abuse of dominant position by excessive and unfair pricing is not identified as the cost of producing the product or the profit made in selling it, but as the “economic value of the product supplied”. The selling price of a product is excessive and an abuse “if it has no reasonable relation to its economic value.” c) Third: “the court did not say that the economic value of a product is always ascertained by reference to the cost of producing it plus a reasonable profit cost (cost +), or that a higher price than cost + is necessarily an excessive price and an abuse of a dominant position. The court as indicating that one possible way (inter alia) of objectively determining whether the price is excessive and an abuse is to determine, if the calculation was possible, the profit margin by reference to the selling price and the cost of production.”354 350 Napp Pharmaceutical Holdings Ltd v Director General of Fair Trading [2002] CAT 1, paragraph 392. 351 [2006] CAT 36, paragraph 247. 352 Albion Water Limited v Water Services Regulation Authority (formerly Director General of Water Services) [2006] CAT 36, paragraphs 251 and 252. 353 Case No: A3/2006/0126 [2007] EWCA Civ 38, paragraph 116. 354 Case No: A3/2006/0126 [2007] EWCA Civ 38, paragraphs 117-118. 159 d) Finally the Court of Appeal noted that: “the law on abuse of dominant position is about distortion of competition and safeguarding the interests of consumers in the relevant market. It is not a law against suppliers making “excessive profits” by selling their products to other producers at prices yielding more than a reasonable return on the cost of production, i.e. at more than what the judge described as the “competitive price level”. Still less is it a law under which the courts can regulate prices by fixing the fair price for a product on the application of the producer who complains that he is being overcharged […]”355 6.320 In summary, the Court of Appeal noted that “exceeding cost + is a necessary, but in no way sufficient, test of abuse of a dominant position.”356 In that case, the Court of Appeal found that the judgment at first instance had based the economic value of the product supplied by the British Horseracing Board (UK pre-race data) on ‘cost +’ which did not take sufficient account of the value of that pre-race data to Attheraces (the complainant) and tied the costs allowable in ‘cost +’ too closely to the costs of producing that data. 6.321 Ofcom began its analysis of the excessive pricing allegation with an assessment of BT’s prices compared to measures of cost. Alleged excessive pricing: assessing BT’s prices compared to costs 6.322 Ofcom notes the Court of Appeal’s comments in BHB and the ECJ’s test in United Brands. In particular, the ECJ set out two tests as highlighted by the Court of Appeal: • whether the difference between the costs actually incurred and the price actually charged is excessive. This test is analogous to comparator a) identified in Napp as set out in paragraph 6.316 above. • whether, if the difference is excessive, whether the price under investigation is unfair in itself or when compared with competing products. This test is analogous to comparators b) to d) in Napp. 6.323 The aim of Ofcom’s initial analysis in line with the first test was to determine whether the difference between the costs incurred by BT in supplying NTS call termination to other OCPs and the price it charged could be considered excessive. 6.324 In this analysis, Ofcom considered two different measures of profitability (return on capital employed, or ROCE, and return on sales, ROS), two different measures of cost (FAC and SAC), and made comparisons with rates of return earned by comparable undertakings. 6.325 As discussed at Annex 5, there are various reasons for using either ROCE and ROS. Ofcom calculated BT’s profitability on both a ROCE and ROS basis. 6.326 Ofcom’s initial analysis of BT’s profitability used FAC data from BT’s regulatory accounting systems. As noted in Table 12 above, FAC is an accounting concept which ensures that all BT’s costs are attributed to its activities. In a multi-product firm like BT, it is analogous to the concept of average total cost in a single product firm. 355 356 Case No: A3/2006/0126 [2007] EWCA Civ 38, paragraph 119. Case No: A3/2006/0126 [2007] EWCA Civ 38, paragraph 209. 160 Ofcom’s analysis, as set out below, concluded that BT’s returns, calculated on a FAC basis, were high in relation to its costs. 6.327 However, this does not of itself demonstrate that the prices notified in NCCN 500 were excessive. This is because of the existence of significant common costs between BT’s activities. BT’s FAC is one method of attributing these common costs, but it is not unique. Other attribution methodologies may also be reasonable, and could produce quite different views of BT’s rate of return. In the absence of a definitive view on the maximum proportion of these common costs that it is reasonable for BT to recover from NTS termination it is informative to consider the further cost measure of SAC, which includes all (relevant) common costs. 6.328 The OFT states that: “Where a dominant undertaking does not generate supra-normal profits overall, this does not rule out the possibility that it charges excessive prices on a particular line of business”; and that: “To assess the profitability of a line of business it may be necessary to allocate common costs to the particular activities identified…In some circumstances the standalone cost of the line of business may be relevant…Where an activity generates revenues that persistently and significantly exceed its standalone costs…this would be good evidence of excessive profits being earned on that activity.357 6.329 Ofcom has therefore estimated the SAC of NTS call termination and compared the prices notified in NCCN 500 with those costs. 6.330 The SAC of a product is defined as the cost that an efficient firm producing only the service in question would incur. Such a firm would not benefit from economies of scope and common costs which might, in a multi-product firm, be recovered from a number of services, but, in the case of a single-product firm, will need to be incurred for and recovered from the single product. 6.331 The concept of SAC has its origins in the theory of contestable markets.358 A contestable market is one in which the complete absence of barriers to entry means that incumbent firms, even monopolists, are constrained to price no higher than average costs by the threat of entry. In the multi-product case, the incumbent also makes only normal profits overall. The price of each product must be above its incremental costs (otherwise the firm could increase profits by ceasing to produce that product, or would be vulnerable to an entrant producing all products except that one). 6.332 The highest price that a multi-product firm could charge for any individual good or service in a contestable market is given by the efficient SAC of that good or service. This is because a price above this level would attract entry by a single product firm which would compete the price down to this level. It is therefore the maximum price consistent with equilibrium in a contestable market and the absence of entry barriers. 357 OFT 414a, paragraphs 2.12-2.13. See Baumol, W., Panzar, J. and Willig, R. Contestable Markets and the Theory of Industry Structure, (1982), New York, Harcourt Brace Jovanovich. 358 161 In the multi-product context, a price (significantly and persistently) above SAC might therefore be regarded as excessive. 6.333 Ofcom considered two methods of measuring the SAC of NTS call termination. One method would be to use data derived from BT’s regulatory costing systems. These allow the production of measures of the incremental cost and SAC of BT network services for use as ‘first-order’ tests of anti-competitive (i.e. exclusionary or predatory) and excessive pricing respectively. However, there are only two increments in the BT incremental cost model which is used to calculate them – the whole of BT’s network conveyance services and the whole of its local access network.359 6.334 This means that intra-network common costs (costs that are common to more than one core network service but incremental to the core network as a whole) are included in the incremental costs of the various network services. The resulting measures of incremental cost and SAC, known as “distributed” incremental cost and “distributed” SAC, will therefore normally be respectively above and below the “true” incremental cost and SAC measures based on defining each service as the relevant increment.360 A finding that a price was above distributed SAC (“DSAC”) would not therefore be sufficient for a finding that a price is excessive, as such a price could still be below (true) SAC. 6.335 In Albion the CAT rejected estimates of SAC produced by Dwr Cymru and the regulatory authority which had been produced “on a ’new build’ basis even though it was accepted, quite rightly, in evidence that those calculations did not, and could not, form any basis for charging”. The CAT also noted that: “the proper approach to ‘stand-alone’ costs depends on the context and circumstances. None of the above [OFT] definitions were given in the context of an access charge for common carriage, which is the present context”.361 6.336 Ofcom's view is that the OFT definition of SAC, which is similar to that given at paragraph 6.330 above, is appropriate to the present case, which is not concerned with setting charges for common carriage.362 Moreover, Ofcom's estimates of SAC do not rely on an assumption of new build financed on a venture capital basis. 6.337 Ofcom therefore constructed a model to estimate, on a “bottom-up” basis, the costs that a hypothetical single product firm would incur in providing NTS call termination/hosting on a standalone basis, that is (true) SAC. This analysis is set out at paragraph 6.363 et seq. 359 For further explanation of the method used to calculate incremental and standalone costs for BT’s network services, see Guidelines on the Operation of the Network Charge Controls from October 2001, published by Oftel, December 2001, http://www.ofcom.org.uk/static/archive/oftel/publications/ind_guidelines/pcr1201.htm, especially at Annex B. 360 Ibid. 361 [2006] CAT 23, paragraph 575. 362 OFT 414a refers to SAC as “those costs that would be incurred if the company undertook only the line of business in question”. A discussion paper by Oxera, an economic consultancy, published as OFT 657 in July 2003, defines “stand-alone” cost as: “The costs of an activity or line of business that would be incurred if the company undertook that activity only. All common costs are attributed to the activity in question.” See http://www.oft.gov.uk/advice_and_resources/publications/reports/Economicresearch/oft657. 162 FAC profitability analysis: introduction 6.338 NCCN 500 came into effect on 1 May 2004, one month after the beginning of BT’s 2004/05 financial period. To assess the impact of NCCN 500, Ofcom therefore compared BT’s profitability in 2004/05 with its profitability in the previous financial year. 6.339 Changes in reported profitability between financial periods could be due to a number of factors, for example, changes in network costs or revenue share payments. To isolate the impact of NCCN 500 from other factors, Ofcom sought to distinguish the profitability of calls originated by other OCPs and terminated by BT (which were subject to the price increases notified in NCCN 500), and the profitability of BT-to-BT calls (which were not). 6.340 The only item which was treated in the analysis differently depending on whether it related to calls originated by other OCPs or calls originated by BT was NTS call termination revenues. Otherwise Ofcom’s approach to the attribution of costs and revenues did not differ according to whether calls were OCP-originated or BToriginated, meaning that any differences in profitability between the two call types would be solely attributable to NCCN 500. 6.341 As discussed in Section 4 above, NTS call termination/hosting is in a two-sided market. NTS call termination is always provided in conjunction with NTS hosting (whether it is provided by a TCP or a reseller). It was therefore only relevant to Ofcom’s analysis to determine the profitability of NTS call termination/hosting combined. 6.342 Ofcom also compared BT’s profitability for different number ranges and for different types of NTS call termination/hosting (voice and data). 6.343 The two number ranges most impacted by NCCN 500 were 0845 and 0870 (see Annex 1). Ofcom analysed the profitability for these ranges separately for two reasons: • the price increases notified in NCCN 500 were significantly higher on 0845 than on 0870 (~23% versus ~4% on a weighted average basis); and • BT’s competitors may supply services predominantly or exclusively using either 0845 or 0870. 6.344 While the prices notified in NCCN 500 applied equally to all NTS calls to the relevant number ranges, BT offers two different retail services (voice calls to NTS numbers and metered dial-up internet access) for which NTS call termination is a source of revenue. These services are based on different underlying costs, and receive additional revenue from different sources. Reflecting this situation BT prepares separate management and regulatory financial information for its voice and data termination/hosting businesses. Ofcom therefore analysed BT’s profitability for voice and data separately. 6.345 The profitability analysis for 0845 was split between voice and data, because this number range was material in both, but 0870 profitability analysis only related to voice, since 0870 data calls were negligible. 163 FAC profitability analysis: source of information 6.346 The information Ofcom used in its analysis was initially sourced from BT’s regulatory reporting systems. For the services Ofcom was considering, BT can prepare regulatory product group financial statements as a by-product of the processes it undertakes to prepare its audited published regulatory financial statements.363 These particular regulatory product group statements are themselves not audited. 6.347 Two regulatory product groups were relevant. These include all the activities under investigation as well some others which were not: • P056: BT Linkline: captures all NTS voice calls terminated by BT, both those originated by other OCPs, which generate NTS call termination revenue for BT, and those originated by BT, which are reflected at their retail value to BT.364 P056 includes call termination as well as the hosting activities and related account management support required to serve BT’s voice NTS service provider customers. It also includes the retail and call origination activities in respect of BT-to-BT calls. • P346: Dial IP: captures all non-geographic data calls terminated by BT, both those originated by other OCPs, which generate NTS call termination revenue for BT, and those originated by BT, which are reflected at their retail value to BT. P346 also captures unmetered data calls terminated by BT at their wholesale values. P346 includes call termination and conveyance to the internet provided over BT’s IP network for both metered and unmetered calls, and related account management support for BT’s external ISP customers. P346 also reflects the revenue generated from “port rentals” (the basis on which BT charges for wholesale metered and unmetered data call termination) both in relation to ISPs and BT’s own downstream retail ISP operations. It also includes the retail and call origination activities in respect of BT-to-BT calls. 6.348 The regulatory product group financial statements did not provide the necessary degree of granularity (e.g. revenue analyses split by number range and by OCP) for Ofcom’s analysis. Ofcom therefore required BT to provide further information (see Table 2) consisting of more granular analysis of the costs, revenues and volumes used to prepare the relevant regulatory product group financial statements, and additional information consistent with these statements. FAC profitability analysis: basis of preparation 6.349 The basis of preparation of the profitability financial analysis including the way in which Ofcom processed the information provided by BT is set out in Annex 9. 363 BT is required to produce annual Current Cost Financial Statements, the purpose of which is for BT to demonstrate compliance with its ex ante obligations of cost orientation and non-discrimination for certain products and services. These statements, which separate out BT’s operations between wholesale and retail activities, are prepared under various SMP conditions imposed following market reviews. BT’s regulatory reporting obligations for 2004/05 were set out in The regulatory financial reporting obligations on BT and Kingston Communications: Final Statement, published on 22 July 2004 (www.ofcom.org.uk/consult/condocs/fin_reporting/fin_report_statement/finance_report.pdf). 364 With the exception of calls to Premium Rate (09) numbers, which are captured in a different regulatory product group 164 FAC profitability analysis: results 6.350 The stages of analysis followed in Ofcom’s profitability analysis are set out in Annex 9. In the tables below a distinction is drawn between profitability arising from BT-toBT traffic and OCP-to-BT traffic. In practice when negotiating hosting contracts with service providers BT will take into account the NTS termination revenues it will receive both from its own customers and from OCP customers (subject to NCCN 500) in determining the revenue it proposes to share with the provider. 6.351 It is therefore arguable that it is only meaningful to look at NTS termination/hosting profitability combining both types of calls. This could still be the case even if (as Ofcom understands and as reflected in the following profitability analysis) BT has no way of knowing once the contract has been negotiated whether a call terminated in respect of an individual service provider originated on an operator’s network subject to NCCN 500. The comparison drawn between BT-to-BT and OCP-to-BT traffic may however still be helpful in understanding the incremental impact on profitability arising from the introduction of NCCN 500. 6.352 The results of the analysis are set out (on an end-to-end basis) in Table 18-Table 20 below. 165 Table 18: BT’s profitability: 0845 voice 0845 Voice 2003/04 2004/05 BT to BT OCP to BT £m a Revenue NTS call termination Other Total % point premium £m b BT to BT OCP to BT £m a b-a £m b % point premium b-a x Costs (profit & loss) Revenue share outpayments Network costs Retail costs Total [] Profit y Mean capital employed (MCE) z Volumes (millions of minutes) Performance measures Return on capital employed (ROCE) y/z% Return on sales (ROS) y/x% Note: numbers are not rounded and therefore figures may not total exactly. Table 19: BT’s profitability: 0870 voice 0870 voice 2003/04 BT to BT OCP to BT £m £m a Revenue NTS call termination Other revenues Total % point premium b-a x Costs (profit & loss) Revenue share outpayments Network costs Retail costs Total [] Profit y Mean capital employed (MCE) z Volumes (millions of minutes) Performance measures Return on capital employed (ROCE) y/z% Return on sales (ROS) y/x% Note: numbers are not rounded and therefore figures may not total exactly. 166 b 2004/05 BT to BT OCP to BT £m £m a b % point premium b-a Table 20: BT’s profitability: 0845 data 2003/04 0845 data Revenue NTS call termination Other revenues Total 2004/05 BT-to-BT £m OCP-to-BT £m % point premium BT-to-BT £m OCP-to-BT £m % point premium a b b-a a b b-a x Costs (profit & loss) Revenue share outpayments Network costs Retail costs Total [] Profit y Mean capital employed (MCE) z Volumes (millions of minutes) Performance measures Return on capital employed (ROCE) y/z% Return on sales (ROS) y/x% Note: numbers are not rounded and therefore figures may not total exactly. 6.353 Several factors, some of which are listed below, mean that great care needs to be taken when interpreting the financial information in Table 18-Table 20 above. • The costs reflected in the regulatory product groups (i.e. P056 and P346) are themselves the result of an extensive and multi-level cost attribution process undertaken by BT to prepare its regulatory financial statements. In the context of a multi-product firm like BT this means that the source data for Ofcom’s profitability analysis has already been highly processed by BT. • In order to ascertain the impact of NCCN 500 on BT’s profitability Ofcom needed to focus on relatively narrowly defined services (e.g. 0845 voice termination hosting for OCP-to-BT calls). The absolute cost and revenue figures for these services finally obtained are small, both in absolute terms and in relation to the entirety of BT’s activities. Profitability measures based on such small figures will be inherently very sensitive to the accounting policies and attribution methodologies adopted, both by BT and Ofcom, in the preparation of the information shown in the tables. • Ofcom selected volumes (millions of minutes across all times of day) as the cost driver to apportion shared costs to individual number ranges. This single cost driver is necessarily an overall proxy for the underlying cost driver for each individual activity supporting NTS termination/hosting services. • It was necessary to request volume information from BT above and beyond that required by BT to prepare financial statements for P056 and P346. This additional information was sourced from BT’s operational systems supporting its NTS voice and data termination activities and was not always transparently consistent with information extracted from BT’s regulatory accounting systems. 167 6.354 Collectively these factors mean that the figures in these tables are not completely precise, but should be regarded as subject to a margin of error. In addition, as part of its consideration of the other tests for excessive pricing set out at paragraph 6.316 et seq above, Ofcom also took account of the premium of the returns obtained by BT from OCP-to-BT calls (subject to NCCN 500) over and above the returns on BT-toBT calls (not subject to NCCN 500), and on trend information obtainable by comparing 2004/05 with 2003/04. These may also be less subject to some of the factors set out above. 6.355 From BT’s voice activities (Table 18 and Table 19 above) it is evident that BT’s revenue share payments to NTS service providers have become increasingly significant over time against a background of an overall modest increase in call volumes. Consistent with this theme, payments from NTS service providers to BT (reflected in “other revenues”) for 0845 voice have fallen over time. These are the principal drivers behind reduced profitability in BT’s voice activities before taking into account the impact of NCCN 500 on OCP-to-BT calls. 6.356 Within BT’s data activities the dominant theme was the significantly reduced volumes between 2003/04 and 2004/05. BT’s revenue share payments to NTS service providers fell slightly on a unit basis. 6.357 The results of Ofcom’s profitability analysis have been brought together in the following summary table. FAC profitability analysis: summary of results Table 21: BT’s profitability: summary 2003/04 BT-to-BTOCP-to-BT a b 2004/05 % point premium b-a BT-to-BTOCP-to-BT a b % point premium b-a Return on capital employed (ROCE) Voice 0845 0870 Data 0845 [] Return on sales (ROS) Voice 0845 0870 Data 0845 6.358 For both 0845 voice and 0845 data the significant positive impact on BT’s profitability resulting from NCCN 500 is reflected in OCP-to-BT calls. The incremental impact on Ofcom’s profitability measures is shown in the “% point premium” column for 2004/05. 6.359 This premium is wholly attributable to the difference between the charges notified in NCCN 500 and NTS Calculator rates, as all other costs and revenues have been apportioned between BT-to-BT and OCP-to-BT calls on an equivalent basis. This premium is much more modest for 0870 voice, which is in keeping with the smaller price increase on this range. 168 6.360 A number of points can be drawn from this financial analysis: • NCCN 500 increased BT’s profitability on NTS voice and data calls originated by other OCPs, significantly so in the case of the 0845 number range; • []; • voice profitability (as measured by ROCE and ROS) is higher than data profitability; and • profitability (again measured by ROCE and ROS) is falling for voice and data before the impact of NCCN 500 is taken into account. 6.361 Ofcom also calculated the combined returns earned by BT from its voice and data NTS call termination/hosting businesses. Ofcom calculated separate values for 0845 voice and data and for all 0845 and 0870 activities combined. The OCP-to-BT returns for 2004/05 reflect the profits that BT earned as a result of NCCN 500. Table 22: BT’s profitability: weighted average returns 2003/04 2004/05 BT-to-BT OCP-to-BT BT-to-BT OCP-to-BT Return on capital employed (ROCE) 0845 voice and data All [] Return on sales (ROS) 0845 voice and data All 6.362 The proportion of total calls represented by OCP-to-BT calls differs by range and by call type. The weighted average returns in Table 22 are not therefore the same for OCP-to-BT calls as they are for BT-to-BT calls in 2003/04. With this caveat in mind, these combined figures demonstrate that profitability fell between the years 2003/4 and 2004/5, before the impact of NCCN 500 is taken into account. Alleged excessive pricing: profitability analysis based on SAC SAC profitability analysis: methodology 6.363 As part of its assessment of BT’s profitability (see paragraph 6.322 et seq above), Ofcom modelled the SAC faced by a hypothetical single-product firm providing NTS call termination/hosting and compared it with the revenues that BT received under NCCN 500 for the provision of the same service. If, at the prices charged by BT, it could earn a positive return compared to SAC significantly above its cost of capital, this might suggest that the prices charged by BT for NTS call termination when NCCN 500 was in force were excessive. 6.364 Ofcom first analysed BT’s profitability on the 0845 range, which was subject to the highest price increases under NCCN 500 (see Annex 1) before expanding its analysis to incorporate 0870. 169 6.365 Ofcom adopted a hybrid approach to modelling the SAC of the hypothetical single product firm, using a detailed bottom-up modelling exercise to establish the hypothetical single product firm’s likely network costs and estimating all the other costs of its business on the basis of BT’s costs. 6.366 This was because the retail costs of other NTS providers from whom Ofcom obtained cost information (see paragraph 3.27 above, final bullet) would not have reflected the retail costs faced by the hypothetical single product firm. 6.367 Ofcom has assumed that the hypothetical single product firm operates on the same scale as BT. It would therefore face retail costs associated with managing large contracts (for example those provided by BT in response to Ofcom’s section 26 Notice of 24 April 2006). By contrast, the smaller providers from whom Ofcom obtained information faced minimal retail costs, and in some cases none at all. For example, one way of doing business at a small scale is to enable customers to manage their own accounts online, so that there is no face-to-face customer relationship management and therefore no significant ongoing retail costs. 6.368 The purpose of Ofcom’s analysis was to assess whether BT’s charges notified in NCCN 500 were excessive in relation to the underlying costs. It followed that the firm should be able to provide the same services at the same scale as BT, and thus be capable of theoretically satisfying all of BT’s present demand. Ofcom therefore assumed that the hypothetical single product firm would receive the same volumes of NTS traffic as BT and would follow the same time of day profile. Modelling the hypothetical single product firm’s network costs 6.369 The hypothetical single product firm’s network costs include both capital expenditure and operational expenditure, notably: • switching; • other hardware and site costs; • staff; • software and other resources required to run the network; and • payments to BT for geographic call termination. 6.370 To estimate these costs, Ofcom first determined the size of the network that the hypothetical single product firm would need to be able to provide a volume and quality of service equivalent to that of BT. The scale of the hypothetical single product firm’s network was determined with reference to the volume of minutes terminated by BT in 2004/05. 6.371 The hypothetical single product firm’s network as modelled is shown in Figure 20 below: 170 Figure 20: Network model for hypothetical single product firm Load balanced 1/3 traffic Site 3 Network monitoring centre 1,886 E1’s Switch 1,886 E1’s BT DMSU 1,886 E1’s Switch Site 1 Site 2 SCP SC SCP 1,886 E1’s Switch 1,886 E1’s 1,886 E1’s Switch Load balanced- 1/3 Load trafficbalanced + 17% 1/3 traffic redundanc OSS OSS BSS BSS 1,886 E1’s Switch 1,886 E1’s 1,886 E1’s Switch Load balanced- 1/3 Load balanced traffic + 17% 1/3 traffic redundanc Network costs: key assumptions 6.372 The hypothetical single product firm’s network infrastructure, as modelled, operates under one principal constraint, which is that it is not modelled as having a national network fully interconnected with that of BT. The rationale for this is that such a level of infrastructure is possessed only by communications providers offering a broad range of services over their networks (e.g. BT and C&W). It is highly unlikely that a firm offering only NTS call termination/hosting, even at BT’s scale, would find it economic to build and maintain its own national network. 6.373 Ofcom’s analysis therefore assumes that the hypothetical single product firm has a minimal network and buys from BT such network conveyance services as it requires (for example geographic call termination). 6.374 A hypothetical single product firm interconnecting with BT at a single point would be vulnerable to service outages resulting from localised problems on BT’s network. BT’s larger NTS service provider customers, for whom continuity of service is likely to be a crucial requirement, would be unlikely to consider an alternative provider whose network was less robust. 6.375 To partially compensate for this, the hypothetical single product firm is modelled as having three interconnections to one single BT Digital Main Switching Unit (“DMSU”), of which any two are of a sufficient size to carry all traffic. If one of the three routes into BT’s network was to become unavailable (for example, because of problems in 171 the ducting which contains the data cables) the hypothetical single product firm would be able to carry the same total volume of traffic via the remaining two routes. 6.376 Ofcom believes that voice NTS providers with a single point of interconnection with BT are relatively common (although they may not have the same provision for continuity of service in the event of network outages, discussed in the preceding paragraph, as our hypothetical single product firm). 6.377 Ofcom does not, however, believe that there is an equivalent business model for providers of data NTS call termination. It appears to be much larger multi-product firms such as BT and C&W that provide data NTS call termination/hosting. These firms have achieved economies of scale and scope essentially by taking on the functions of ISPs themselves. Both BT and C&W possess large modem banks and thus convert their traffic into IP at the earliest opportunity, which represents a more cost-effective means of conveying data calls than the PSTN. 6.378 Ofcom has not however modelled an IP network provider because firms such as C&W and others with large IP networks are not single product firms, and so their costs would not be appropriate as an estimate of our hypothetical firm’s SAC. The fact that single product providers of data NTS call termination/hosting may not be observed in practice does not vitiate the use of this model for the SAC test. Indeed, in most markets where there are significant economies of scope, single product firms would not generally be expected to be able to operate profitably. 6.379 However, there is no theoretical reason why a hypothetical single product firm could not offer data NTS call termination/hosting without investing in its own IP network. For a standalone operator it would not be economical to invest in an IP network – all the operators that provide an IP network are multi-product firms. A standalone provider could receive NTS data calls and send them over the PSTN to a geographic number hosted by an ISP with its own modems. In this way, the hypothetical single product firm would not be differentiating between voice and data calls. Ofcom has therefore considered such a business model in its analysis. 6.380 Even a hypothetical single product firm with a single point of interconnect with BT would still require significant network infrastructure to process the large volume of traffic implied by this test. The principal determinant of the scale of a single product firm’s network, and thus the appropriate equipment to use, is the peak volume of traffic that this network would have to process. 6.381 The proportion of calls in the busy hour was assumed by Ofcom for the purposes of modelling to be 20%. This is significantly larger than the figure BT uses for its own planning purposes, which we understand to be around 10-15%. BT may not pass the test with a lower figure. However, Ofcom considers that the hypothetical single product firm is likely to need more capacity than this for two reasons. Firstly, it is handling NTS calls exclusively, and NTS service providers tend to generate significant peaks in the demand for access to their numbers. Secondly, the hypothetical single product firm has to offer a level of service equivalent to BT, and would need to reassure large customers that it had enough network capacity to satisfactorily process all the incoming calls received. Therefore, even Ofcom’s assumption of a 20% ratio may be seen as conservative. 6.382 Based on these assumptions, Ofcom estimated the likely capacity that the hypothetical single product firm would need to provide an equivalent service to that offered by BT at the busiest time of the day. 172 6.383 In order to handle such volumes, Ofcom’s model is based on “carrier class” switches, i.e. the type of switch that would be used by large communications providers such as BT or C&W. Although smaller providers of NTS call termination/hosting may use inexpensive PC-based switches, these components would not reliably be able to provide services at the scale being considered, both because of the volumes handled and the quality of service provided by large dedicated switches (which would be demanded by large NTS service provider customers for whom, as discussed above, continuity of service is likely to be business critical). 6.384 Ofcom determined that four suitably configured Ericsson AXE 10 switches would be required to service the volume of traffic expected at the busiest periods. To provide a degree of resilience, the three-site configuration detailed above was modelled with two such switches at each site. This configuration provides a capacity of 50% more than the busy hour requirement, enabling the network to cope with the failure of a single switch or of an entire site (i.e. six switches in total). 6.385 Ofcom has derived total switch costs for the single product hypothetical entrant by multiplying the number of required switches (as set out above in paragraphs 6.3836.384 above) by the relevant running cost (i.e. operational expenditure (Opex)) and up-front cost (i.e. capital expenditure (Capex)) estimates for the switches (as set out below in Table 23). Note Ofcom annuitises the Capex estimate (over five years at a weighted average cost of capital (“WACC”) of 11.4%) to arrive at its annuity Capex estimate. Table 23: Switch costs and capacity parameters Carrier class switch Ericsson AXE 10 Initial cost (Capex) £2,000,000 Running cost (Opex) Expected lifetime (years) £200,000 5 Source: Ofcom estimates 6.386 The interconnection cost for the single product hypothetical entrant from its three sites to BT’s network is set out in Table 24 below. Connection costs were annuitised (over five years at a WACC of 11.4%) to arrive at an annuity Capex estimate. Fixed rental costs are multiplied by the estimated number of interconnection circuits. Per kilometre rental costs were multiplied by the estimated number of interconnection circuits and the distance between the entrant’s site premises and BT’s network (21 kilometres). Total interconnection costs (including an annuitised connection, fixed rental and per kilometre rental costs) were then added together. The total estimated interconnection cost was then expressed in pence per minute (“ppm”) by dividing through by traffic minutes in the 0845 voice and data product group and for the 0845 and 0870 voice and data product group – see Table 29 and Table 30 below). 173 Table 24: Interconnection costs Average cost over period of NCCN 500 Connection £1,292 Rental – fixed (notes 2 & 5)365 £1,777 Rental – per km (notes 1, 2, 4 & 5)366 Period of recovery £24 5 Source: Ofcom, BT carrier price list 6.387 Ofcom’s model also included the control software and hardware elements needed to provide NTS call termination/hosting. These include a service control platform (“SCP”), which directs the translation of NTS calls into geographic calls and provides some hosting services, such as automated voice response and intelligent routing. Ofcom has estimated these costs in Table 25 below. Table 25: Interactive Service Platform IN SCP cost per hundred367 IN SCP operational costs per hundred Period of recovery £2,000,000 £200,000 5 Source: Ofcom estimates 6.388 Ofcom derived total annual SCP platform costs by summing SCP Opex and an annuitised SCP Capex estimate (over five years at a WACC of 11.4%). These estimates were then expressed in ppm by dividing though by traffic minutes in the 0845 voice and data product group and for the 0845 and 0870 voice and data product group – see Table 29 and Table 30 below). 6.389 The model also included an interconnection billing system which records details about the ingress and egress of all calls and is used to verify payments from BT to the hypothetical single product firm for NTS call termination/hosting and the payments made by the hypothetical single product firm to BT for geographic call termination. These primary systems are located at one site with backups at another in case of site failure. 365 Refers to notes to BT’s carrier price list (see btwholesale.co.uk). Refer to notes to BT’s carrier price list (see btwholesale.co.uk). 367 Refers to capacity of 100 transactions per second (“TPS”). Note Ofcom converts this number into a “call arrival per second” metric ([] for all NTS calls) and multiplied by two to obtain a “transactions per second” metric ([] for all NTS calls). The “number of hundreds” required by the hypothetical entrant can be derived by dividing this estimate by [] (approximately [] “hundreds”, hence total SCP costs for hypothetical entrant is £[]or £[] annuitised over 5 years at a AWACC of []%).). 366 174 Table 26: Operations Systems Software (OSS) & Billing Systems Software (BSS) costs System CAPEX (£) OPEX (£) Replacement period BSS/OSS Hardware (Active) 1,000,000 100,000 5 273,298 BSS/OSS Hardware (Standby) 1,000,000 100,000 5 273,298 WIA/billing & mediation (Software and licensing) 1,500,000 150,000 5 409,947 CRM (software & licensing) 2,000,000 200,000 5 546,596 Source: Annuity required Ofcom estimates 6.390 Ofcom added Opex and annuitised Capex (annuitised over five years at a WACC of 11.4%) estimates to derive a total OSS and BSS cost. These estimates were then expressed in ppm by dividing though by traffic minutes in the 0845 voice and data product group and for the 0845 and 0870 voice and data product group – see Table 29 and Table 30 below. 6.391 Another element of the costs included in the model was the staff that would be needed to maintain, operate and develop the hypothetical single product firm’s business. Provision was made for a dedicated team at each site, a central management team responsible for monitoring the hypothetical single product firm’s network 24 hours a day, a team responsible for implementing technical projects and maintaining the core systems, and developers responsible for bespoke software projects (either to maintain and improve the SCP and associated software or to develop bespoke systems which customers may require). 6.392 Table 27 below sets out these costs in detail. 175 Table 27: Network staff costs Per site employees FTE required Salary (£) Data build engineers 2 50,000 O&M engineers 3 50,000 1.3 50,000 Switch manager 1 60,000 Cabling resource 1 45,000 Data centre manager 1 45,000 IN support 1,400,000368 Total salary cost Network management centre employees NMC manager 1 60,000 Network surveillance 12 30,000 Technical support managers 1 60,000 Technical support 6 45,000 Database administrators 6 50,000 Service delivery 12 50,000 Reporting 2 45,000 Total salary cost 1,740,000 Technical project management team Business analyst 4 40,000 Technical analyst 4 50,000 Release manager 1 50,000 Technical project manager 2 60,000 Testers 4 40,000 Architecture and design 1 60,000 Total salary cost 750,000 Development team 368 Note that per site salary costs are multiplied by 3 to obtain a sub-total cost of £1,400,000 which reflects total per site employee costs across the three sites. 176 Per site employees FTE required Salary (£) Developers 8 70,000 Development manager 1 75,000 Total salary cost 635,000 Senior network management Head of development 1 80,000 Head of service delivery 1 80,000 Head of databases 1 80,000 Head of project management 1 80,000 Network Director 1 150,000 Total salary cost 470,000 Source: Ofcom estimates 6.393 The staff costs reported in Table 27 above represent salary costs only. To adjust to reflect the additional non-salary costs of employing and retaining staff (such as National Insurance, training etc) Ofcom applies a costing factor of 2.5 to reflect the total salary and non-salary related costs for the entrant.369 These total costs were then divided through by traffic minutes to derive a ppm estimate of the staff costs for the 0845 voice and data product group and for the 0845 and 0870 voice and data product group – see Table 29 and Table 30 below. 6.394 Ofcom also considered site costs for the single product hypothetical firm (see Table 28 below). These total costs were also apportioned on the basis of traffic minutes to derive a ppm estimate of the site costs for the 0845 voice and data product group and for 0845 and 0870 voice and data product group – see Table 29 and Table 30 below. 369 Ofcom estimate. 177 Table 28: Site costs (operations and resource) Equipment Up-front costs (£) Switch 100,000 Up-front Annuity required (£) 27,330 Replacement Period 5 OPEX (£) UPS 150,000 40,995 5 15,000 Air conditioning 300,000 81,989 5 50,000 Fire suppression (e.g. FM200)370 60,000 16,398 5 20,000 Cabling, DDF, jumpers 40,000 10,932 5 Generator 300,000 81,989 5 10,000 20,000 Electricity 25,000 Property rental 100,000 Support and maintenance costs per site 100,000 Sum of OPEX and annuitised CAPEX per site per annum 599,633 Source: Ofcom estimates 6.395 In order to determine the payments the hypothetical single product firm would make to BT for geographic call termination, Ofcom undertook a geographic analysis of the size of the population that would be covered by each of BT’s DMSUs at singletandem rates. It then assumed that the hypothetical single product firm was interconnected to the DMSU that provides the highest population coverage, 13%. For the remaining double tandem termination payments to BT Ofcom used data provided by [] to estimate the proportions of short, medium and long double tandem termination. Ofcom used BT’s carrier price list to identify for relevant conveyance services and prices (in ppm). Total geographic call termination costs were apportioned on the basis of traffic minutes to derive a ppm estimate of the costs for the 0845 voice and data product group and for the 0845 and 0870 voice and data product group – see Table 29 and Table 30 below. 6.396 A cost of capital of 11.4% in pre-tax nominal terms was used to annuitise all the capital expenditure. This is the cost of capital that BT faces for regulatory purposes outside its copper access network (for which the rate is 10%), which is used as a reasonable assumption in the absence of evidence of the cost of capital reflecting the 370 Brand of fire suppression system. 178 specific characteristics of the NTS termination/hosting business.371 However, BT is a large multi-product business, with many income streams, and such favourable access to capital may not therefore be available to a smaller single-product firm.372 Modelling the hypothetical single product firm’s non-network costs 6.397 The non-network costs faced by the hypothetical single product firm cover all the other functions of the hypothetical single product firm’s business, including : • sales and marketing; • customer support; • business support (finance, HR);and • general administration. 6.398 For the reasons set out above, Ofcom did not perform bottom-up modelling to derive costs for these functions, but instead derived its own estimate of the SAC derived from the total retail cost attributed by BT to the regulatory product groups analysed by Ofcom for the purposes of its FAC profitability analysis.373 6.399 BT classifies non-network costs related to the provision of NTS call termination/hosting within ”retail costs” within its regulatory financial information. The relevant regulatory product groups for this analysis are (as set out at paragraph 6.347 above) P056 and P346. 6.400 The total retail costs within these groups (after elimination of the costs which relate to the retailing by BT of NTS calls to its own subscribers) include costs for number ranges other than 0845 and 0870 and so the first step to estimating the hypothetical single product firm’s costs was to apportion these costs between the relevant number ranges. 6.401 Ofcom did not consider that apportioning these costs to the services in question based on call volumes (which was the approach used in Ofcom’s FAC profitability analysis) was the most appropriate approach when deriving a measure of the relevant SAC. This was most clearly the case in P346, where an apportionment on the basis of call volumes would lead to unmetered data calls picking up []% of retailing costs but generating []% of revenues. In P056, an allocation based on call volumes would assign most of the costs of retailing to 0845 calls, as these account for a greater proportion of total volumes. However, this overlooks the fact that 0870 calls contribute twice the revenue of 0845 calls, as they provide more revenue per minute. 6.402 To account for the relative value of the call types in question, Ofcom decided on an approach which would allocate the retail costs based on net revenue (excluding revenue share payments to NTS service providers). This is consistent with the 371 See Ofcom’s approach to risk in the assessment of the cost of capital: Final Statement, 18 August 2005, published at: http://www.ofcom.org.uk/consult/condocs/cost_capital2/statement/final.pdf. 372 Had Ofcom’s analysis indicated that the charges notified in NCCN 500 were excessive based on BT’s cost of capital, it might have been appropriate to conduct a sensitivity analysis using different rates. Since the analysis did not lead Ofcom to this conclusion, it was not necessary to carry out this sensitivity. 373 In this context retail costs excludes revenue share outpayments, which are treated separately in the standalone cost analysis. 179 approach Ofcom adopted in respect of marketing expenditure in setting the NTS retail uplift.374 6.403 This methodology enabled Ofcom to estimate BT’s non-network costs for the products under consideration. 6.404 However, the retail cost information drawn from BT’s regulatory accounting system is prepared on an FAC basis, which means that further adjustments are required to prepare an estimate of SAC for the purposes of this modelling exercise. 6.405 One approach to estimating the SAC would be to estimate a ratio of SAC to FAC based on an appropriate comparator. One that is readily available is the ratio of DSAC to FAC ratios based on BT network services in the relevant regulatory product groups, P056 and P346, which yields ratios in the region of 2.4 to 2.8. These comparators are not ideal in the present context, for two reasons; first, they relate to network rather than non-network costs, and second, they are based on DSAC not SAC. 6.406 In relation to the first of these points, economies of scale and scope may be greater for network services than for retail services, and the derived DSAC to FAC ratios could therefore overestimate the true DSAC to FAC ratio for non-network costs. Having said that, economies of scale and scope in non-network functions are not negligible. For example, in the modelling used to inform the setting of BT's retail price controls in 2002 and in the NTS Retail Uplift in 2005 Ofcom used a cost-volume elasticity (the ratio of the percentage change in costs to the percentage change in output causing it) of 0.25 for retail costs (apart from bad debt which varies directly with revenue) while in other cases Ofcom has used a cost volume elasticity (“CVE”) of around 0.45 for indirect costs.375 These CVEs reflect significant economies of scale in these activities (or, equivalently economies of scope in the services which the functions support). 6.407 In relation to the second point, on the other hand, it is likely that DSAC is below the true SAC for any given service, for the reasons set out above (see paragraph 6.334 et seq above), namely that intra-business common costs are distributed among all the services to which they are common (rather than added to the incremental costs of just the service in question, as required for a true SAC). As a result, the DSAC:FAC ratios referred to above could lead to an underestimate of SAC for non-network expenditure. 6.408 Taking these factors into consideration, Ofcom derived SAC estimates for nonnetwork expenditure by applying a factor of 2 to the FAC estimates. Ofcom regards 374 See paragraphs A2.22 - A2.29 of the September 2005 statement Charges between Communications Providers: Number Translation Services Retail Uplift charge control and Premium Rate Services bad debt surcharge. The analysis on which this decision was based deals mainly with marketing expenditure. The underlying logic is that "it is the profit earned supplying products to its customers that provide the motivation to advertise". It would be consistent that sales and marketing effort, and thereby cost, would be concentrated upon those areas which earn the highest margins. 375 CVE is defined as the percentage by which costs increase for a 1% increase in volume. See Protecting consumers by promoting competition: Oftel's conclusions, 20 June 2002, published at: http://www.ofcom.org.uk/static/archive/Oftel/publications/whole_line/2002/pcr0602.htm and, for example, Determination of fixed portability costs and charges and statutory consultation on proposed modifications to BT's Licence to give effect to charge controls for portability, 31 May 2002, published at http://www.ofcom.org.uk/static/archive/oftel/ind_info/licensing/mods/2002/galvin31502.htm or Wholesale Line Rental: Reviewing and setting charge ceilings for WLR services, 9 November 2005, published at: http://www.ofcom.org.uk/consult/condocs/wlrcharge/wlr.pdf 180 this as a reasonably conservative assumption, in the sense that it may tend to understate the costs of the hypothetical single product firm. 6.409 In addition, there is some corroboration for the level of non-network costs implied by the ratio of 2 being conservative, which in the case of the modelling for the hypothetical single product firm supplying both 0845 and 0870 NTS call termination/hosting comes to an SAC of £63m. Ofcom’s analysis shows that BT's total retail costs for the two regulatory product groups, on an FAC basis, are around £[]. Ofcom has adjusted this figure downwards to exclude BT’s costs of retailing calls to its retail customers (about £[]), leaving a total £63 million, which would need to be increased by an appropriate factor to give an SAC figure. 6.410 The model does not include hardware elements such as leased lines and any other software that may also be required to provide such services since as discussed, these are bespoke services in addition to NTS call termination/hosting. Modelling these on a bottom-up basis could not be achieved without detailed information on the types and specifications of the services and to whom and in what quantity they are offered. However, as the results do not suggest that the charges notified in NCCN 500 were excessive (see below), Ofcom does not consider that it is necessary to factor in this additional cost. 6.411 The model also includes revenue share payments to NTS service providers. These are taken from a breakdown provided by BT for the purposes of the FAC profitability analysis of its 2004/05 retail costs which Ofcom has expressed on a pence per minute basis based on the total volumes terminated by BT on the number ranges in question. BT’s revenues 6.412 BT’s gross revenues in the NTS termination/hosting market come from two sources: revenues for the provision of NTS call termination/hosting and payments received from NTS service providers. The net revenue is the sum of these two elements less any revenue share payments to NTS service providers. 6.413 The main source of is payments from OCPs for NTS call termination/hosting. These revenues have been calculated as the product of the charges notified in NCCN 500 (see Annex 1) and the volume of calls terminated (i.e. the actual volume of calls terminated by BT, as set out at paragraph 6.370 above). 6.414 BT also derives revenue from payments from NTS service providers for some NTS hosting services. BT does not (and is not required to) publish separate charges for these services. Ofcom therefore derived a ppm figure for the average income from these services for each number range from the regulatory product group information provided by BT (i.e. from P056). Note that Ofcom has excluded port rental revenues received from data service providers in this analysis. In addition, the network cost element of the model includes a cost for software developers responsible for creating and maintaining bespoke services. Results 0845 only 6.415 The results for the combined SAC analysis of 0845 voice and data calls are set out in Table 29 below. 181 Table 29: BT’s loss on ppm basis based on the SAC of the hypothetical single product firm (0845 voice and data combined) Item 0845 voice and data combined Termination revenue Other revenues [] [] Total revenues [] Payments to NTS service providers [] Net revenue [] Switch Site costs OSS & BSS: Interconnection Interactive services platform Operations and resources Geographic termination payments Total network costs [] [] [] [] [] [] [] [] Non-network costs [] Total costs [] BT’s loss -0.368 6.416 These results indicate that BT’s prices for termination of 0845 calls when NCCN 500 was in force did not exceed the estimated SAC of the hypothetical single product firm providing NTS call termination/hosting at BT’s scale for the whole of the 0845 number range and there was a loss of 0.368 ppm. 6.417 This evidence does not support the view that BT’s prices for the termination of calls to 0845 numbers, as notified in NCCN 500, were excessive. 0845 and 0870 6.418 Ofcom also considered a hypothetical single product firm providing NTS call termination/hosting on both 0845 and 0870. 6.419 Call volumes for 0870 are smaller than for 0845 ([] terminated minutes as opposed to [] for 0845 in 2004/05). However, the higher revenues associated with 0870 mean that it is as significant to BT as 0845 in revenue terms. 6.420 The hypothetical single product firm providing NTS call termination/hosting on both 0845 and 0870 is modelled by Ofcom in much the same way as the 0845-only single product firm. 6.421 However, there are some important differences in costs and some significant economies of scale. The area where the costs for 0870 differ significantly from those for 0845 is in the level of payments made to NTS service providers. The average payment made by BT for an 0870 call is [] ppm, compared to [] ppm for an 0845 call. This higher revenue share payment reflects the higher ppm NTS call termination/hosting revenues BT receives from 0870. 182 6.422 The other major difference is that, in this model, payments from NTS service providers to BT tend to be higher for 0870 calls, as 0870 calls tend to be voice calls and 0845 data calls do not attract any of these additional payments. 6.423 The hypothetical single product firm providing NTS call termination/hosting on both 0845 and 0870 is able to benefit from some significant economies of scale. The network configuration outlined above is still judged to be the most suitable setup for the combined operator. Some of the modelled costs increase in proportion to the increase in traffic (notably the number of interconnects with BT and the quality of SCP required). 6.424 However, all of the network costs decrease on a ppm basis due to economies of scale, apart from payments to BT for geographic call termination. This is due to the different time of day profile of 0870 calls as a whole as a whole compared to 0845 calls. The volume of data NTS calls terminated on 0870 is negligible, meaning that 0870 calls are generally speaking voice calls, which are more likely to be made during the day and therefore attract higher geographic termination charges. 6.425 Non-network costs for the hypothetical single product firm are also higher than when it is terminating 0845 calls This is due mainly to the methodology Ofcom has used to derive the ppm non-network costs whereby 0870 calls, due to their higher net revenues, attract more of the total “retail costs” attributed to the relevant regulatory product groups . 6.426 The results for the single product firm proving NTS call termination/hosting for both 0845 and 0870 is given at Table 30 below. 183 Table 30: BT’s loss on a ppm basis based on the SAC of the hypothetical single product firm (0845 and 0870) Item Value (ppm) Termination revenue Other revenues [] [] Total revenues [] Payments to NTS service providers [] Net revenue [] Switch Site costs Operating and billing systems: Interconnection Intelligent networking platform Operations and resources Geographic termination payments Total network costs [] [] [] [] [] [] [] [] Non-network costs [] Total costs [] BT’s loss -0.013 6.427 These results indicate that BT’s prices for 0845 and 0870, taken together, when NCCN 500 was in force, did not exceed the SAC of a hypothetical single product firm providing NTS call termination/hosting at BT’s scale. This evidence does not support the view that BT’s prices for the termination of calls to 0845 and 0870 numbers, as notified in NCCN 500, were excessive. 6.428 This result is more marginal than the 0845-only results presented in Table 29 above, which appears to be largely because of the higher net revenues associated with 0870 (and economies of scale from the volumes of 0845 and 0870 taken together). Model input analysis 6.429 Ofcom undertook sensitivity analysis to determine which inputs in the modelling process had the greatest bearing on the outcomes as reported in the tables above. 6.430 The two factors which had the largest impact on the hypothetical single product firm’s costs were the uplift that was applied to BT’s FAC costs of retailing to arrive at a SAC estimate, and the proportion of calls that were deemed to fall within the ‘busy hour’. 6.431 Since the network used by the hypothetical single product firm was modelled in detail, network costs can be estimated to a relatively high degree of accuracy so that likely variations in individual constituent costs such as the cost of switching equipment and the pay rates of employees had an immaterial effect on the outcomes reported. 6.432 As already discussed above Ofcom applied a factor of 2 to BT’s attributed FAC nonnetwork costs for 0845 and 0870 calls in order to derive an estimate of these costs 184 on an SAC basis. Ofcom believes that this is a relatively conservative assumption and that there might be a justification for using a higher ratio. However, as Ofcom’s analysis does not indicate that the charges notified in NCCN 500 were in excess of SAC, it is not necessary to subject this assumption to a sensitivity analysis of the effect of using a higher ratio. 6.433 As regards the busy hour proportion, as discussed above there are features of NTS traffic that may generate spikes in demand, that is very high short-term peaks in traffic volumes. For an operator attempting to supply NTS at BT’s quality of service, including coping with these peaks in demand, the assumption made in Ofcom’s cost analysis may be something of an underestimate. If this is a significant factor, the true SAC may be somewhat above Ofcom’s estimate. However, as Ofcom is able to reject a finding of excessive pricing in any case it is not necessary to further allow for this factor. 6.434 A further important variable is the level of payments made by BT to NTS service providers. For voice calls, Ofcom considers that using the actual value of payments made by BT is appropriate, as a hypothetical single product firm would need to offer equivalent payments to win business from BT. However, in the data business the payments reflected within regulatory product group P346 also take account of additional services that BT provides, such as converting PSTN traffic to IP which are not provided by Ofcom’s modelled operator. Therefore, it may be the case that the hypothetical single product firm using the configuration detailed above would have to offer significantly higher payments than BT to win BT’s business (or invest in its network to offer the same services as BT). Either scenario would result in reduced profitability for the hypothetical single product firm and would therefore make a finding of excessive pricing against BT less likely. Ofcom has not therefore conducted a sensitivity analysis on this cost, as it is able to reject a finding of excessive pricing without doing so. Alleged excessive pricing: conclusion 6.435 Before finding an infringement of Article 82 and/or the Chapter II prohibition, Ofcom must satisfy itself on the evidence that an infringement is more likely than not to have occurred.376 The undertaking being investigated is entitled to the presumption of innocence and to any reasonable doubt that there may be. Ofcom must therefore proceed on the basis of an analysis that is robust and soundly based.377 6.436 As discussed at paragraphs 6.305-6.321,above, the case law presents a number of possible tests and comparators that can assist in determining whether an undertaking’s prices are “excessive”, i.e. whether they are higher than would be expected in a competitive market. 6.437 In United Brands, the ECJ set out a two-step process for identifying excessive pricing, stating that the questions to be determined were: “whether the difference between the costs actually incurred and the price actually charged is excessive, and, if the answer to this question is in the affirmative, whether a price has been imposed 376 Napp Pharmaceutical Holdings Ltd v Director General of Fair Trading [2002] CAT1, paragraph 109. 377 JJB Sports plc and Allsports Limited v Office of Fair Trading [2004] CAT 17, paragraphs 191-199. 185 which is either unfair in itself or when compared with competing products.”378 6.438 There are therefore two broad areas for consideration in determining whether the prices notified in NCCN 500 were excessive. First, whether BT’s prices were excessive compared to costs (i.e. “the difference between the costs actually incurred and the price actually charged”); and second, if this were found to be the case, whether the prices notified in NCCN 500 were “unfair in [themselves] or when compared with competing products.” It is Ofcom’s view that, in the absence of compelling evidence of adverse effects on competition and limited evidence of an adverse effect on consumers, Ofcom would need cogent evidence that BT’s pricing was excessive under such tests. 6.439 The absence of likely or actual harm to competition appears to have been a key consideration in BHB (see paragraph 6.601 et seq below) and Ofcom looked at whether any actual harm occurred to competition in this case, or whether it was likely that any harm to competition would have occurred as a result of the price changes notified in NCCN 500. 6.440 Ofcom considers that, had the prices notified in NCCN 500 been excessive, this might have led to two possible types of harm. First, BT’s competitors would have been harmed by having to pay the higher charges, reducing their ability to compete with BT in the origination segment and in NTS hosting. This type of harm was assessed by the retail and hosting margin squeeze analysis, which concluded that BT’s competitors did not suffer a margin squeeze either in origination or in NTS hosting as a result of BT’s conduct. Analysis of market shares over the period that NCCN 500 was in force (and since) suggest no relevant change in market share for BT in the relevant market during this period. 6.441 Second, had BT’s competitors passed the higher charges through to their retail customers, this might have led to consumer detriment in the form of higher retail prices. As set out at paragraphs 5.114-5.115 above, Ofcom is only aware of one instance where a competitor to BT ([]) passed the higher prices through to their customers, and notes that this price rise appeared to go beyond what was required to cover the additional cost imposed by NCCN 500. However, Ofcom notes that BT’s competitors may have been waiting for the outcome of Ofcom’s investigation and market review before taking any decision on changing their prices (see paragraph 5.112 above). Were BT’s prices excessive compared to costs? 6.442 To consider whether BT’s prices were excessive compared to its costs, Ofcom undertook an analysis of BT’s profitability on two different measures: FAC and SAC. 6.443 The results set out at paragraph 6.346 et seq above showed that BT’s profitability on an FAC basis increased as a direct result of NCCN 500. For the reasons set out above (at paragraph 6.353 above), the resultant profitability indicators are subject to some possible margin of error. In addition, there seemed to be some differences between the profitability of BT’s voice and data businesses. Nevertheless, overall, 378 Case 27/76 United Brands v Commission [1978] ECR 207, paragraph 252 (the ECJ went on to point out in the following paragraph that there were other ways “of selecting the rules for determining whether the price of a product is unfair”). 186 the results of Ofcom’s FAC analysis suggest that BT’s NCCN 500 prices were comfortably above costs on an FAC basis. 6.444 However, the FAC measure of costs depends on the particular allocation of costs reflected in BT's accounts, which is not necessarily the maximum which could reasonably be recovered from NTS call termination. While the FAC analysis suggested that BT’s prices were comfortably above cost, in the circumstances of this case, Ofcom’s view is that this in itself was not sufficient to conclude that the prices notified in NCCN 500 were excessive compared to BT’s costs, in relation to the first test set out in United Brands. 6.445 Ofcom therefore conducted a further analysis that compared BT’s prices to the SAC faced by a hypothetical single product firm. As noted above (at paragraph 6.331 above), in principle SAC represents the highest price that could be sustained in a contestable market. 6.446 As one would expect, the SAC of a hypothetical single product firm was higher than the costs generated by the FAC analysis, because the latter only reflect a proportion of the (potentially significant) common costs between the services under investigation (NTS call termination/hosting, primarily on the 0845 and 0870 number ranges) and the wide range of other services that BT supplies in conjunction with NTS termination/hosting. 6.447 The results of Ofcom’s SAC analysis (as set out at paragraph 6.415 et seq above) suggest that the prices notified in NCCN 500 were below SAC. 6.448 Ofcom considers that the comparison of BT’s prices and costs offered by each of the FAC and SAC analyses provides relevant but conflicting evidence. The inference to be drawn from the comparison of prices and costs therefore depends on the strength of the evidence overall. In the circumstances of this case, Ofcom’s judgement is that the evidence of excessive pricing provided by the FAC analysis is insufficient on its own, because it reflects one particular way in which common costs could be recovered, but it is not clear that this is more reasonable than many other ways of recovering common costs. The different results in the analysis of SAC, which represents an upper bound on prices sustainable in a contestable market, demonstrate the importance of the common cost recovery issue in this case. 6.449 The United Brands test is a two-step test. Even if Ofcom had concluded, on the basis of its FAC analysis alone, that the prices notified in NCCN 500 were excessive compared to BT’s costs, it would then have needed to conduct the second test set out by the ECJ in United Brands, as applied by the Court of Appeal in BHB, i.e. whether the prices notified in NCCN 500 were “unfair in [themselves] or when compared with competing products.” This analysis is set out in the next section. Were the prices notified in NCCN 500 unfair in themselves or when compared with competing products? 6.450 As discussed above, the case law offers a number of possible comparators that may help Ofcom scope a test for determining whether, in terms of the United Brands test, the prices notified in NCCN 500 were unfair when compared with competing products. 6.451 In Napp, for example, in considering whether the second test set out in United Brands was passed, the OFT considered the following possible comparators (in addition to a comparison of the target’s prices and its underlying costs): 187 • Napp’s prices with the costs of its most profitable competitor; • Napp’s prices with those of its competitors; and • Napp’s prices with prices charged by Napp in other markets. 6.452 In the main Albion judgment, the CAT did not employ all of the tests suggested in Napp. In relation to the first comparison (between the dominant undertaking’s prices and its costs – i.e. the profitability analysis discussed in the preceding section) the CAT referred to the fact that the existence of supply contracts to other non-potable water users was not a valid comparator for determining whether excessive pricing had occurred in that case as there was no evidence that such contracts were cost related.379 6.453 In its December 2006 Albion decision, the CAT cited paragraph 252 of United Brands, where the ECJ refers to a price being either unfair in itself or when compared to competing products and the CAT noted that the latter was difficult to apply in Albion as “there is no substitute for the service of the transportation and partial treatment of water here in question”.380 In those circumstances, the CAT decided to focus on the comparison of supply costs to prices to customers for potable and non-potable water.381 6.454 In this case, as in Albion, not all of the comparisons suggested in Napp are relevant. The nature of the NTS call termination/hosting market and the services supplied by BT mean that there is limited scope for replicating the tests applied in Napp in order to determine whether BT’s prices were “unfair in themselves or when compared to competing products.” 6.455 Ofcom considered the possibility of comparing BT’s prices with the costs of its next most profitable competitor. In order to undertake such a comparison, Ofcom would have had to assess the profitability of a number of other TCPs to identify BT’s most profitable competitor. Other TCPs do not produce regulatory financial statements (the accounting system for which was primary source of data for Ofcom’s assessment of BT’s profitability), and discussions with C&W suggested to Ofcom that other TCPs do not produce financial information at a level of granularity that would have enabled Ofcom to conduct an equivalent assessment of their profitability in relation to the relevant services. 6.456 In section 26 Notices dated 28 June 2005, Ofcom asked a number of providers to: “Provide an estimate of your costs (other than revenue share outpayments) of NTS call termination for 2004/05, at as granular a level as possible, setting out in detail your method of preparation of assumptions (particularly in relation to allocation of common costs) and clear descriptions of each cost category you have identified.”382 6.457 None of the respondents were able to provide any level of detail regarding these costs that would have assisted Ofcom. 379 [2006] CAT 23, paragraph 753-757. [2006] CAT 36, paragraph 250. 381 December 2006 decision of the CAT in Albion (see footnote 380), paragraph 252. 382 This question was included in section 26 Notices sent to C&W, Centrica, Energis, Gamma, ntl, Telewest, Thus and Tiscali. 380 188 6.458 C&W stated in its response of 10 July 2005 that “network costs are not attributed at this level”. Energis, in a subsequent exchange of emails about the feasibility of providing more granular cost data, stated that it would welcome clarification of how, if it were able to provide this data, Ofcom would adjust it to ensure a fair comparison with BT’s costs, and went on to say that: “We don't want to lessen the importance of Ofcom's analysis of excessive pricing but given the limitations that we discussed we would be concerned about attempting a meaningful assessment of BT's profitability using Energis’ costs as we do not have a high degree of confidence that even if all the relevant costs were readily obtainable that they are directly comparable with BT’s.”383 6.459 Since Ofcom had concluded that the first test set out in United Brands was not clearly passed, it concluded that the complexity and scale of such an exercise would have outweighed the value that it would have brought to Ofcom’s conclusion, in particular given the comments made by C&W about the reliability of their own data as a suitable basis for comparison. Ofcom has not therefore compared BT’s prices with the costs of its most profitable competitor. Ofcom has, however, compared BT’s prices with the costs of a hypothetical single product firm, as set out above. That analysis concluded that BT’s prices were not in excess of the costs of the hypothetical competitor. 6.460 As discussed at paragraph 2.26, prior to NCCN 500, prices for NTS call termination were uniform, with BT charging the same rate as other TCPs. NCCN 500 increased BT’s charges for NTS call termination price by up to 37.8% on some ranges, with a weighted average price increase on 0845 of approximately 26% and on 0870 of approximately 3% (on a weighted average basis). 6.461 Moving to a comparison between the prices notified in NCCN 500 and the prices charged by BT in other markets (following the final test adopted in Napp), Ofcom noted that NTS call termination is supplied only in the NTS call termination/hosting market. Ofcom does not consider that BT offers any comparable products in other markets that would offer a relevant benchmark for assessing whether BT’s prices for NTS call termination, as notified in NCCN 500, were unfair when compared to competing products. 6.462 Ofcom notes that there is little guidance in case law to assist in determining how a price may be unfair in itself. For example, in United Brands, while stating that “charging a price which is excessive because it has no reasonable relation to the economic value of the product supplied” would amount to excessive pricing, the ECJ offered little guidance on how to determine the “economic” value of the product or service provided.384 6.463 In dealing with this question of whether the price was “unfair in itself” in Albion, the CAT noted that in its view there was a certain ambiguity in the test set out in United Brands as to whether there is a distinction between a price which is excessive and a price which is unfair and stated: “At present, we are not persuaded that, in a case such as the present, there is necessarily a distinction between a price that is 383 384 email from [] (Energis) to [] (Ofcom), 19 September 2005. Case 27/76, [1978] ECR 207, paragraph 250. 189 “excessive”, as distinct from a price which is “unfair”, or that “economic value” is intrinsically difficult to ascertain.”385 6.464 The CAT went on to say: “…this case does not concern the establishment of some justum pretium (just price) in the abstract sense, but the much more prosaic question of whether the costs of the “distribution” of non-potable water are the same as the costs for the “distribution” of potable water.”386 6.465 In Scandlines, the Commission was of the view that, in determining whether a price might be unfair in itself, the economic value of the product/service should also take account of other non-cost related factors, especially as regards the demand-side aspects of the product/service concerned.387 6.466 In view of the recent case law in excessive pricing that questions whether there is any real distinction between whether a price is excessive when analysed in relation to costs or unfair in itself or compared with competing products and Ofcom’s assessment that there was insufficient evidence to conclude that the NCCN500 prices were excessive compared to BT’s costs, Ofcom does not believe there is sufficient evidence to conclude that these prices were unfair in relation to costs. Conclusion on excessive pricing 6.467 As set out in the previous two sections, Ofcom has considered whether, based on the balance of probabilities, the two-step test to determine whether excessive pricing has occurred – as set out in United Brands and followed by the CAT in Napp and Albion, the Court of Appeal in BHB, and the Commission in Scandlines – is passed in this case.388 6.468 The United Brands test is sequential, relying first on a determination of whether a price can be considered excessive and, if it can, then assessing whether that price is unfair in itself or when compared with relevant benchmarks. 6.469 As set out above, the Court of Appeal made it clear in BHB that it is also important to assess in relation to allegations of excessive pricing whether there is evidence of competition in the market being distorted by the pricing.389 6.470 Ofcom has considered whether the available evidence, using comparisons with BT’s FAC and the SAC of a hypothetical single product firm, would support a finding of infringement. 6.471 Ofcom has considered whether it has sufficient evidence to satisfy the standard of proof.390 As stated by the CAT in Napp: “…..the standard of proof in proceedings under the Act involving penalties is the civil standard of proof, but that standard is to be applied bearing in mind that infringements of the Act are serious 385 [2006] CAT 36, paragraph 250. [2006] CAT 36, paragraph 251. 387 Case COMP/A.36.568/D3, 23 July 2004, paragraph 226. 388 JJB Sports plc & Allsports Limited v Office of Fair Trading [2004] CAT 17, paragraph 191-196. 389 Case No A3/2006/0126 EWCA Civ 38, paragraphs 215-218. 390 The Racecourse Association v Office of Fair Trading [2005] CAT 29, paragraph 131. 386 190 matters attracting severe financial penalties. It is for the Director to satisfy us in each case, on the basis of strong and compelling evidence, taking account of the seriousness of what is alleged, that the infringement is duly proved, the undertaking being entitled to the presumption of innocence, and to any reasonable doubt there may be.”391 6.472 The CAT has subsequently endorsed this approach in JJB Sports plc & Allsports Limited v Office of Fair Trading.392 6.473 Ofcom also notes the view expressed by the Commission in Scandlines that it bears the burden of proof based on cogent evidence to establish the existence of the abuse.393 6.474 In the circumstances of this case, Ofcom does not believe that the evidence meets the standard of proof set out in the case law to establish that the prices notified in NCCN 500 were excessive. 6.475 While there is no doubt that NCCN 500 did have the effect of increasing BT’s revenue at the expense of its competitors (see paragraph 6.555-6.556 below), without a justification based on BT’s underlying costs, Ofcom does not consider that this in itself is sufficient to draw a conclusion that BT’s prices were excessive. 6.476 As noted above, Ofcom is mindful of the Court of Appeal’s decision in the BHB case that abuse of a dominant position is about distortion of competition and safeguarding the interests of consumers, and is not a law against suppliers making excessive profits by selling their products to other suppliers at prices yielding more than reasonable returns on the cost of production.394 Ofcom did not find any evidence that BT’s conduct in this case led to a distortion of competition or that it led to an adverse effect on consumers generally, as only one of BT’s competitors raised its prices in response to NCCN 500. 6.477 Therefore Ofcom concludes that there is insufficient evidence to support C&W’s allegation that the prices notified in NCCN 500 were excessive and that BT’s conduct thereby constituted an abuse of BT’s dominant position within the meaning of the Chapter II prohibition and Article 82. Responses to draft decision: excessive pricing 6.478 C&W, CPW and Thus comment on Ofcom’s excessive pricing analysis. BT does not comment on Ofcom’s assessment of C&W’s allegation of excessive pricing. C&W 6.479 C&W submits that: “[Ofcom’s] approach [to its analysis of alleged excessive pricing] is simplistic and based on an inappropriate approach to cost allocation. Ofcom is overly-reliant on a simple analysis using the United Brands approach and thus looks strictly at cost issues; in fact the case law 391 [2002] CAT 1, paragraph 109. [2005] CAT 29, paragraph 197-200. 393 Case COMP/A.36.568/D3, 23 July 2004, Scandlines Sverige AB v Port of Helsingborg, paragraph 244 394 Case No A3/2006/0126 [2007] EWCA Civ 38, paragraph 119. 392 191 since United Brands makes it clear that excessive pricing cannot be analysed simply on the basis of cost considerations.”395 6.480 C&W states that Ofcom has “rejected” its own analysis of BT’s profitability on an FAC basis: “because it argues there are different ways in which common costs could be allocated, and the methodology used in BT’s FAC is only one way.”396 6.481 C&W submits that: “if Ofcom had doubts about FAC then it could have considered a range of allocation methodologies rather than disregarding altogether the appropriateness of taking into account that in a multiproduct firm with common costs like BT, the “competitive” price level is below the standalone cost.”397 6.482 C&W proposes a number of alternative approaches to cost allocation which it believes Ofcom should have considered, namely: • equi-proportional mark-ups (EPMU); • Ramsey pricing; and • Efficient Component Pricing Rule (ECPR).398 6.483 In C&W’s view, “Ofcom should have investigated use of these cost allocation methodologies rather than relying just on a SAC model to assess excessive pricing.”399 C&W submits that Ofcom could, and should have used LRIC estimates, rather than SAC, as the basis of its excessive pricing comparison. 6.484 Turning to Ofcom’s SAC analysis, C&W has a number of concerns with Ofcom’s approach. 6.485 C&W contends that the use of SAC as a cost base is inappropriate as: “It is unrealistic to expect any competitor to provide only NTS services and not to provide any other calling products…..Ofcom has taken account of this to a limited extent by assuming that a network built out for NTS termination would not include conveyance assets, however this is unlikely to fully capture this issue. […] new entry in the NTS termination market is likely to occur at a price which is lower than SAC and therefore this lower price should be considered in the test.”400 6.486 C&W submits that Ofcom is wrong to rely on Albion as a relevant precedent, and to rely on the OFT’s guidelines.401 395 C&W’s response to the draft decision, page 41. C&W’s response to the draft decision, page 42. 397 Ibid. 398 C&W’s response to the draft decision, page 43. 399 Ibid. 400 C&W’s response to the draft decision, page 46. 396 192 6.487 C&W submits that: “Precedent shows that a finding of excessive pricing not only requires consideration of costs, but should include other tests so that the evidence can be considered in the round.”402 6.488 C&W submits that “the fact that prices cannot be shown to be above SAC is not sufficient to rule out that prices might be excessive” and suggests, in relation to the four tests set out in Napp, that: • “Comparison of BT price with BT’s cost: [...] SAC may be inappropriate for this market and thus more emphasis should have been placed on other cost standards; • Comparison of BT’s costs with cost of the next most profitable competitor: Ofcom did not test this on the basis that it could not identify the next most profitable competitor and did not believe that cost data was available. These arguments are weak given that Ofcom could have compared price against all competitors and could have requested additional information [...] • Comparison of BT’s price with those of its competitors: This was not tested by Ofcom. However the price charged by BT was above those of its competitors. Moreover, the prices charged by BT’s competitors were set by the NTS origination retention.”403 6.489 C&W acknowledges that the fourth test in Napp, comparing prices with those charged in other markets, may not be appropriate here.404 6.490 C&W cites the more recent decision in BHB which suggests that assessment of a customer’s willingness to pay, rather than an undertaking’s costs, may be an appropriate way to assess alleged excessive pricing.405 6.491 C&W submits that Ofcom has not properly considered the second strand of the test set out in United Brands, i.e. whether the price charged is unfair in itself or when compared to competing products.406 6.492 C&W suggests that Ofcom has “fail[ed] to complete its analysis [i.e. to carry out additional tests] because of lack of evidence”.407 CPW 6.493 CPW does not believe that Ofcom has carried out a “sufficiently sophisticated” analysis of the excessive pricing allegation.408 6.494 CPW submits that: 401 C&W’s response to the draft decision, page 44. Ibid. 403 C&W’s response to the draft decision, page 45. 404 Ibid. 405 C&W’s response to the draft decision, page 46. 406 C&W’s response to the draft decision, page 47. 407 C&W’s response to the draft decision, page 48. 408 CPW’s response to the draft decision, page 3. 402 193 “…the fact that prices cannot be shown to exceed SAC is not sufficient to rule out that prices might be excessive. In other words, if Ofcom would have found that prices were above SAC, this would have been sufficient to conclude that they were excessive.”409 6.495 CPW states that Ofcom should have carried out a variety of tests for excessive pricing, in line with the general approach in case law. C&W notes that Ofcom “strangely dismisses” three of the four tests identified by the OFT in Napp, and notes that in BHB “willingness to pay” is used as a further benchmark.410 6.496 Further, CPW notes, citing Naloo, that: “more recent case law even suggests that one need to consider whether the excessive pricing is applied in a discriminatory fashion.”411 6.497 When looking at Ofcom’s SAC analysis, CPW has concerns that Ofcom’s model contains unrealistic assumptions, since “no new entrant would decide to enter the UK communications market with a view to supply only NTS services.” Ofcom’s removal of conveyance assets from the model does not, in CPW’s view, fully address this issue.412 6.498 CPW is also: “concerned that Ofcom appears to dismiss the findings using FAC because it argues that there are different ways in which common costs could be allocated, and the methodology used in BT’s FAC model is only one way….Ofcom should consider a range of allocation methodologies rather than disregarding altogether the appropriateness of taking into account that in a multi-product firm with common costs like BT, the “competitive” price level will be below SAC.”413 6.499 CPW believes that: “If Ofcom believes it is not possible to employ different allocation methodologies in BT’s FAC model Ofcom should take BT’s top-down LRIC cost and add the fixed and common cost mark-up which can be done using a range of alternative approaches.”414 Thus 6.500 Thus states that it: “has grave concerns that Ofcom has made the decision that BT’s pricing was not excessive based on the output from a very simplistic stand alone cost model….(t)he reality is that BT is a complex business supplying multiple products over a shared platform. We have serious concerns for the future of our sector if Ofcom make an 409 Ibid. CPW’s response to the draft decision, page 5. 411 Ibid. 412 Ibid. 413 Ibid. 414 Ibid. 410 194 excessive pricing finding dependant on exceeding Stand Alone Costs. In our sector we can think of very few examples where SAC modelling would accurately reflect market conditions.”415 6.501 Thus believes that Ofcom should have developed a model based on a LRIC + EPMU approach.416 6.502 It also states that: “in considering the issue of excessive pricing it is our belief that Ofcom should have conducted a wider enquiry into the allegation and not one (that) was focused solely on cost modelling alone.”417 Ofcom comments on responses Ofcom’s approach and legal precedents 6.503 The respondents raise a number of issues about Ofcom’s approach to the excessive pricing allegation in the context of the case law. 6.504 C&W’s assertion that Ofcom “relies on Albion Water as authority for its approach” is inaccurate. The use of SAC was rejected as a basis for setting prices in Albion. However, the different circumstances of this case make it the appropriate basis for Ofcom’s analysis here. 6.505 C&W submits that Ofcom has placed too much emphasis on a footnote in the OFT’s draft guideline Assessment of Conduct as support for applying SAC in this matter. 418 6.506 Ofcom notes that the OFT’s discussion of allocation of costs in this section of the guideline is of a general nature: “To assess the profitability of a line of business it may be necessary to allocate common costs to the particular activities identified. Whether and how this should be carried out will depend on the circumstances of the case.”419 6.507 The footnote qualifies the previous statement by stating that “in some circumstances the standalone cost of a line of business may be relevant”. Ofcom has provided its reasons for believing that it is appropriate to consider SAC in this case (see paragraph 6.330 et seq above). 6.508 The respondents submit that Ofcom has focused only on BT’s profitability and failed to consider the allegation “in the round” or to take into account other tests suggested by the case law. This is clearly not the case. As discussed above, the other comparisons specified in Napp (see paragraph 6.316 above) were taken into account to the extent that they were meaningful. 6.509 The respondents suggest that Ofcom should also have considered willingness to pay as a relevant benchmark, as discussed in the recent BHB judgement. 415 Thus’s response to the draft decision, page 10. Ibid. 417 Ibid. 418 OFT 414a. 419 OFT 414a, paragraph 2.13. 416 195 6.510 Ofcom notes that BHB related to an intellectual property service, for which sole focus on the cost of production might not be appropriate. It is unclear to Ofcom the extent to which an analysis of willingness to pay is relevant to the very different type of services that are covered by NCCN 500. However, even if relevant, Ofcom notes that a standard for excessiveness based on willingness to pay is a tougher threshold to cross even than SAC, since customers may well be willing to pay in excess of this level, particularly if in practice, customers have shown themselves willing to pay the prices charged, as in this case. Alternatives to SAC as basis of comparison 6.511 C&W proposes three alternatives to the use of FAC and SAC as a benchmark for excessive pricing. These are LRIC+EPMU (long run incremental cost plus equiproportional mark-up), Ramsey pricing, and the ECPR (efficient component pricing rule). Ofcom did not consider any of these alternatives in its analysis, as none is relevant to this case. 6.512 LRIC+EPMU is an approach in which the share of all relevant common costs allocated to each service is equal to the ratio of the incremental costs of providing that service to the sum of all incremental costs for the multi-product firm. If all charges are set at LRIC+EPMU, full recovery of common costs will be achieved, with no supernormal profit. It is likely that LRIC+EPMU will be similar to FAC, which also involves allocating all relevant common costs among all of the services. Therefore, an analysis of LRIC+EPMU may add little to the evidence already provided by the FAC analysis. In addition, there are other methods of allocating common costs and there is nothing intrinsically correct in a proportionate allocation of common costs in assessing excessive pricing. Accordingly, the evidence provided by such an approach is likely vulnerable to the same limitations as the FAC approach discussed above at paragraph 6.327 above and there are potentially a number of other reasonable allocation methods that could be used. Some services would then receive an allocation of common costs greater than that implied by strict proportionality to the incremental costs of that service. The range of reasonable allocations could include some in which all common costs are recovered more than proportionately from one or a small number of services. 6.513 In Ofcom’s view it would not necessarily be unreasonable for BT to set the charge for the services covered by NCCN 500 to recover a share of common costs greater than that implied by an allocation in proportion to the incremental costs of that activity, particularly given that these are a few of many services across which competition occurs. First, other charges may be set at less than LRIC+EPMU and thus be making a smaller contribution to common cost recovery. It would then be necessary for some charges to be above LRIC+EPMU to ensure full cost recovery. Second, even in competitive markets it is not the case that overall revenues, taking all products together, are at all times just equal to cost including a rate of return equal to the cost of capital.420 Third, as discussed as paragraph 6.301 above, cogent evidence of excessive pricing would be needed under these tests to support an infringement finding. 6.514 Ramsey pricing is a means of calculating the price which optimally recovers fixed costs subject to an overall zero (supernormal) profit constraint. Specifically, it requires the price of each service to be equal to the marginal cost of that service plus a mark-up which is inversely proportional to the (market) elasticity of demand for that service. It therefore relies on elasticity information which, in practice, is not likely to 420 OFT 414a, paragraphs 2.16-2.20. 196 be available. Moreover, a rational firm will set prices taking account of firm level elasticities which may differ from the market elasticities required for efficient Ramsey prices to be set. 6.515 Even if demand elasticity information were available, Ofcom notes that, while Ramsey pricing represents an efficient cost recovery rule for fixed or common costs (and hence represents one interpretation of FAC for a multi-product firm), it is not necessarily unreasonable for the charge for a service to recover a share of fixed or common costs in excess of that implied by the Ramsey pricing rule for that service. First, as with LRIC+EPMU charges, the Ramsey rule cannot be applied to one service in isolation. If other charges are set at less than the level implied by the Ramsey pricing rule and are thus making a smaller contribution to common cost recovery, it would be necessary for some charges to be above the Ramsey price to ensure full cost recovery. Second, even in competitive markets it is not the case that overall revenues, taking all products together, are at all times just equal to cost including a rate of return equal to the cost of capital.421 6.516 In its recent judgment in the BHB case the Court of Appeal came to the view that, in deciding whether a price is excessive, in the sense of bearing “no reasonable relation to economic value”, customer willingness to pay should be taken into account.422 A Ramsey price could be said to be consistent with this principle since, by being based on elasticities of demand, it is related to willingness to pay. 6.517 However, a Ramsey price is not the only price that might be said to reflect willingness to pay, and there is no suggestion in the BHB judgment that the Court of Appeal intended to suggest this. Indeed, it is clear that originators of NTS traffic continued in practice to (be willing to) pay NCCN 500 prices. In any case, as discussed further below, the BHB case is different in crucial respects to the present case. 6.518 The ECPR is an access pricing rule which is intended to prevent statically inefficient entry by pricing access at retail price minus costs saved by supplying a wholesale competitor rather than the retail market. The ECPR price is therefore derived from and dependent on the retail price and does not bear any necessary relation to the “costs actually incurred” in the provision of the wholesale service. It is therefore unlikely to be appropriate as a test for excessive pricing. It is rather a rule which specifies the margin between a retail and a wholesale price and so is, in effect a form of the margin squeeze test on a narrowly defined market, which Ofcom has in any case rejected as inappropriate (see paragraphs 6.148 et seq above). 6.519 C&W has suggested that LRIC may be the appropriate cost standard in this case. To suggest that any price above LRIC should be considered excessive is incorrect as, if all prices were set at or below LRIC, recovery of common costs would be impossible. This would be unsustainable. Ofcom’s SAC model 6.520 C&W objects to SAC on the basis that it does not allow the firm to “benefit from economies of scale and common costs”. C&W is incorrect on the first point, since Ofcom allows for economies of scale by modelling a firm supplying a volume of calls equivalent to BT’s. While it is true that the single product firm modelled by Ofcom does not benefit from economies of scope, this is by design and is appropriate to the test, since SAC by definition assumes a single-product firm. 421 422 Ibid. Case No A3/2006/0126 [2007] EWCA Civ 38, paragraphs 212-218. 197 6.521 The validity of the SAC test does not depend on the actual existence of singleproduct firms. Indeed, in competitive markets where economies of scope are significant, competition would place pressure on such companies to diversify in order to lower costs, or leave the market. 6.522 C&W appears to understate the ability of small firms without significant network to operate in NTS termination/hosting, though of course this may to some extent be dependent on the availability of some network services at regulated charges, which mitigates “the natural monopoly characteristics of a telecoms access network [and] the extent of shared infrastructure between calling products” described by C&W.423 6.523 Ofcom notes in passing that the actual existence of some single-product operators in NTS call termination/hosting (e.g. operators providing voice NTS call termination services) does not imply BT’s NTS call termination/hosting prices are excessive. The evidence suggests that these providers are based on a different business model to that set out in the SAC model. In the SAC model, the single-product firm was modelled to provide the same quality of service to a similar size of customer base to BT, but to have only a few points of interconnection with BT’s network. Accordingly, actual single-product providers of NTS call termination/hosting are likely to provide a lower level of service than the firm modelled for the purpose of the SAC model, and therefore have lower unit costs (e.g. customers managing their accounts themselves online for an actual single-product firm, compared to dedicated customer account managers for our modelled firm). Other comments 6.524 CPW argues that Ofcom should have investigated a range of FAC approaches. Ofcom did undertake a range of FAC based tests (see paragraphs 6.338 et seq above) which provide evidence relevant to assessing excessive pricing. While, in these tests, Ofcom found BT was pricing above the FAC for relevant NTS number ranges while NCCN 500 was in force, the evidence provided by the analysis of FAC was not conclusive evidence in itself, given that BT passed the test based on SAC and in the absence of compelling evidence of adverse effects on competition and limited evidence of an adverse effect on consumers (see paragraph 6.299). 6.525 CPW has concerns that Ofcom’s model contains unrealistic assumptions, since “no new entrant would decide to enter the UK communications market with a view to supply only NTS services.” Ofcom’s removal of conveyance assets from the model does not fully capture this issue, in CPW’s view. 6.526 Ofcom has already addressed respondents’ concerns that the “unrealistic” assumptions invalidate the application of a SAC approach (see paragraph 6.521 et seq above). In response to CPW’s second point, the reason for modelling a “network-less” provider buying in geographic termination services from BT is to ensure a model which is both tractable and transparent, and where relevant information was available to populate the model. Moreover, such an approach provides for a conservative assessment of SAC, since by not requiring significant investments in fixed network assets, a pass of this test by BT at this SAC level will necessarily imply a pass in an SAC approach which seeks to model significant network costs. 6.527 In considering prices relative to costs (i.e. assessing BT’s profitability) on both an FAC and SAC basis and, in respect of each approach, assessing profitability for 423 C&W’s response to the draft decision, page 46. 198 various number ranges and product groupings, as well as assessing the fairness of the prices (including whether there existed relevant comparisons), Ofcom has considered the allegation of excessive pricing in the round. Other allegations made by C&W 6.528 In addition to the three principal allegations considered above, C&W’s submission argued that BT’s conduct constituted an abuse of BT’s dominant position in that it: • increased C&W’s costs of providing a competing service (or forced it to offer a service inferior to that provided by BT, e.g. a call origination service which does not include calls to particular numbers); • increased BT’s market power in NTS call origination and NTS call termination, which for a dominant undertaking such as BT constitutes an abuse in itself; and • formed part of a concerted strategy to dilute competition. 6.529 These further allegations are addressed below. NCCN 500 increased C&W’s costs of providing a competing service 6.530 C&W submitted, at paragraphs 8.40-8.46 of its submission to Ofcom of 15 March 2005, that NCCN 500 constituted an abuse of BT’s dominant position in that it increased C&W’s costs of providing a competing service, or forced them to offer an inferior service. 6.531 C&W set out three possible courses of action that BT’s competitors might take in response to the price increase notified in NCCN 500: “absorb the price increase and become less profitable in providing retail access services and potentially less profitable compared with service providers whose calls originate with BT; increase prices to their end user customers; or offer a more limited service that does not provide access to BT’s hosted numbers.”424 6.532 Ofcom considers that were a price increase to make BT’s competitors “less profitable”, this would not constitute an abuse of BT’s dominant position unless it could be demonstrated that the impact on BT’s competitors’ profits was sufficient to lead to a margin squeeze. 6.533 As set out above, Ofcom has considered whether the prices notified in NCCN 500 led to a margin squeeze in either retail calls and access provision or NTS hosting. Ofcom concluded that NCCN 500 did not lead to a margin squeeze in either market. 6.534 It follows from this that BT’s competitors, if at least as efficient as BT, could remain profitable (even if they were less profitable than previously) without having recourse to either of the other two possible courses of action suggested by C&W. 424 C&W submission of 15 March 2005, paragraph 8.41. 199 6.535 Ofcom therefore concludes that the charges notified in NCCN 500 do not constitute an abuse of BT’s dominant position on the basis that they increased C&W’s costs of providing a competing service. NCCN 500 increased BT’s market power 6.536 C&W submitted, at paragraphs 8.47-8.59 of its submission, that NCCN 500 constituted an abuse of BT’s dominant position in that it increased BT’s market power in NTS call origination and NTS call termination, which for a dominant undertaking such as BT constitutes an abuse in itself. 6.537 C&W submitted that there were a number of case law precedents for such a finding including Tetra Pak International v Commission and Compagnie Maritime Belge Transports v Commission.425 6.538 Ofcom notes that both cases refer to the “special responsibility” of a dominant undertaking and come to the view that: “the actual scope of the special responsibility imposed on a dominant undertaking must be considered in the light of the specific circumstances of each case which show that competition has been weakened.”426 6.539 Ofcom also notes that both cases involved dominant undertakings with a very high market share, and that in Compagnie Maritime Belge Transports v Commission there were only two competitors in the relevant market. 6.540 Ofcom does not hold the view that NCCN 500 increased BT’s market power in the markets for wholesale call origination or NTS call termination/hosting as explained below. 6.541 As set out in Section 5 above, BT’s market power derives mainly from the restricted ability of its competitors to respond to any increase in BT’s charges for NTS call termination. This is essentially a result of features of the relevant market and its relationships with other markets (for example the fact that other TCPs’ charges for NTS call termination are, in effect, regulated via regulation in the wholesale and retail call origination markets, and the limited incentives for BT’s competitors to establish bilateral agreements for NTS transit rather than using BT). Ofcom does not consider that a position of dominance relying on such factors will be either strengthened or weakened by a change in BT’s charges for NTS call termination. 6.542 While BT’s market share is not a determining factor in its position of dominance in the market for NTS call termination/hosting, an increase in BT’s market share attributable to NCCN 500 might offer evidence that BT’s position of dominance had been further strengthened as a result of NCCN 500. However, as discussed in Section 5 above, Ofcom did not observe any increase in BT’s share of the relevant market over the period in question. 6.543 Finally, Ofcom’s analysis of BT’s conduct does not support C&W’s allegation. Had Ofcom concluded, for example, that BT’s conduct had led to a margin squeeze on its competitors (in either call origination or NTS hosting) this might provide evidence 425 Case C-333/94, Tetra Pak International v Commission, judgement of 14 November 1996 and case C-395/96, Compagnie Maritime Belge Transports v Commission, judgement of 16 March 2000. 426 Case C-333/94, paragraph 114. 200 that their competitive position vis-à-vis BT had deteriorated as a result of NCCN 500 and that conversely BT’s market power had increased. However, Ofcom concluded that NCCN 500 did not lead to a margin squeeze on BT’s competitors. 6.544 Ofcom therefore concludes that the charges notified in NCCN 500 do not constitute an abuse of BT’s dominant position on the basis that they increased BT’s market power in a relevant market. NCCN 500 was part of a concerted strategy to dilute competition 6.545 C&W submitted, at paragraphs 8.60-8.66 of its submission, that NCCN 500 constituted an abuse of BT’s dominant position in that it formed part of a strategy to influence the commercial conduct of BT’s competitors in the relevant market. 6.546 In support of this allegation, C&W quoted BT’s response to Ofcom’s NTS call termination market review: “BT took explicit account of the commercial drivers of OCPs, setting pricing that would not cause them either to cease buying BT’s NTS services or to increase retail pricing … the result was pricing that set call origination charges on a broadly reciprocal basis between OCPs, while leaving all parties a reasonable rate of return for an efficient operator … NCCN 500 was aimed at bringing retail OCP retentions closer into line [with the rates Ofcom permitted BT to retain].”427 6.547 C&W submits that on the basis of this statement: “It is clear that NCCN 500 is designed, and was designed from the outset, to control the commercial behaviour of competitors, thereby relieving competitive pressure on BT in related markets such as the NTS services hosting market and NTS retail services markets.”428 6.548 Ofcom considers that BT’s intentions in increasing its prices for NTS call termination via NCCN 500 might be relevant to its decision in this case. 6.549 In particular, had Ofcom concluded that BT’s conduct constituted an abuse of its dominant position, evidence as to BT’s intentions would have been relevant in calculating any appropriate penalty that could be imposed. 6.550 The CAT in Napp stated that an infringement is committed “intentionally” if the undertaking must have been aware that its conduct was of such a nature as to encourage a restriction or distortion of competition.429 6.551 However, on the basis of the analysis set out above, Ofcom has concluded that BT’s conduct did not constitute an abuse of its dominant position in NTS call termination/hosting in the UK. 6.552 Nonetheless, as part of the investigation, Ofcom considered BT’s intentions in relation to the NCCN 500 price increase and found no evidence of exclusionary intent. 427 C&W submission of 15 March 2005, paragraph 8.61, quoting BT’s response to the market review consultation, pages 33 and 35 (original emphasis). 428 C&W submission of 15 March 2005, paragraph 8.63. 429 [2002] CAT 1, paragraph 456. 201 6.553 In particular, as part of its investigation, Ofcom required BT to provide internal documents relevant to its strategy in issuing both NCCN 500 and NCCN 651, as well as documents relating to the largest hosting contracts that it bid for during the period that NCCN 500 was in force (in order to assess how NCCN 500 affected BT’s competitiveness). This information was requested in section 26 Notices of 26 April 2005, 14 November 2005 and 24 April 2006. 6.554 Ofcom’s analysis of BT’s responses does not support C&W’s allegation that NCCN 500 formed part of a concerted strategy to dilute competition. NCCN 500 appears to have had two aims: • to increase BT Group revenues; and • to encourage Ofcom to reconsider the NTS regulatory regime. Evidence of intention to increase BT Group Revenues 6.555 An internal briefing note on the first complaint stated that: “The driver for this pricing decision was that BT felt there was up to £[] of incremental revenue that could be generated without directly triggering a pricing impact for dialling consumers.”430 6.556 The source of this additional revenue was the disparity between BT’s retail charges and those of its competitors: BT offers retail discounts for calls to 0845/0870 numbers, whereas other originating providers do not. The original intention was “to reduce the ‘average discounts’ value in other operators’ NTS formula calculations”, until it was pointed out that: “the same outcome could be achieved simply by raising BT’s termination prices for these services, as BT had no ex ante pricing constraints on these services.”431 Evidence of intention to encourage Ofcom to reconsider the NTS regulatory framework 6.557 In addition, NCCN 500 was seen as having a longer term strategic value. It was commented that the proposal: “fully seeks to exploit current regulation inconsistencies.”432 and that it: “closed what could be called a ‘regulatory loophole’ the altnets were exploiting.”433 6.558 BT appears to have seen NCCN 500 as a “bargaining chip”, which BT was prepared to give up in return for concessions from its competitors: 430 Briefing paper about the first complaint, 12 May 2004, provided by BT in response to section 26 Notice of 22 April 2005 numbered 1.11. 431 Ibid. 432 BT response to Ofcom section 26 Notice of 22 April 2005, item numbered DK241. 433 BT response to Ofcom section 26 Notice of 22 April 2005, item numbered KE674. 202 "I am expecting that, if we manage to agree a new NTS arrangement with the industry, there will be a contractual piece that makes the charges reciprocal in some way. This would ostensibly foreclose NCCN 500 and similar efforts. However, that does not mean we would lose the VALUE of NCCN 500 (etc.), as these would be live pieces of cash that would be in the melting pot for discussion (e.g. we would want to see something like that £[] per annum in our pockets, but perhaps in a different form such as improved retail margins). This would represent a giveaway of what we consider to be our commercial freedom, so it is a valuable bargaining chip…I have never made a secret of the view that I believe NCCN 500 gives us the only valuable bargaining chip in the game, when dealing with other operators.”434 6.559 NCCN 500 was also viewed as a way of putting pressure on Ofcom to review the broader NTS regime: “This proposal will help to add force to BT’s view that the regime needs to be overhauled by demonstrating that BT can use the current structure to its advantage, just as other operators have used it to their advantage in the past. The price change will also emphasise that the regime is highly uncertain and is difficult to plan for.”435 Conclusions on BT’s intent 6.560 Ofcom’s conclusion, on the basis of the documents supplied by BT, was that BT’s intentions in issuing NCCN 500 were to generate additional revenue for BT Group and to draw attention to anomalies in current regulation that meant that BT was free to raise its charges for NTS call termination, since BT believed that the framework gave its competitors freedoms from which BT did not benefit. 6.561 BT’s decision to withdraw the price increases notified in NCCN 500 appears to have been motivated by factors other than attempting to end conduct that BT knew to be anti-competitive (see paragraph 5.138 above, second bullet), including improving relationships with both Ofcom and wholesale customers. 6.562 []. 6.563 Therefore, the withdrawal of NCCN 500 was seen as: [] 6.564 In a paper to the BT Operating Committee (“OC”) of 19 September 2005, the view is set out that: [] 6.565 In the face of a likely forced withdrawal of NCCN 500, BT believed that its best option was to voluntarily withdraw it in order to improve relationships with Ofcom and wholesale customers: 434 435 BT response to Ofcom section 26 Notice of 22 April 2005, item numbered CA255. BT response to Ofcom section 26 Notice of 22 April 2005, items numbered 1.8 and DK929. 203 [] 6.566 [] 6.567 Ofcom believes that BT’s intention in the setting of NCCN 500 was not to reduce competition, but to generate additional income while also highlighting a regulatory ‘loophole’ that its competitors were exploiting. The decision to withdraw the price changes in NCCN 651 was made when senior management weighed up the benefits of increased revenue against the costs of provoking Ofcom and wholesale competitors/customers. Evidence on hosting contracts 6.568 Ofcom also requested information into the top ten hosting contracts bid for by BT during the period NCCN 500 was in place, in order to ascertain whether NCCN 500 had the effect of enabling BT to win more business. 6.569 The information received from BT served to highlight the highly competitive nature of the market, both in price and non-price competition, but did not suggest that BT was better placed to win new business than it had been before NCCN 500. 6.570 When discussing a particular contract in internal emails, one of BT Wholesale’s Sales team members states that: “This particular customer will trawl the marketplace for an 0845 provider offering zero ppm. We need to have a zero on 0845 to ensure that we keep the business and that [customer x] don’t go off to Opal/C&W who would jump at offering zero for this amount of business (approx 60 million minutes per year).” 6.571 Ofcom also noted from the information received from BT that it has stringent regulatory controls in place for approving new bids, and in certain circumstances this can prove a hindrance to winning new business, as a member of the Sales team notes: “Having to receive sign off levels from Legal, Finance, Business Operations and a Bid Manager for every single bid is not an effective use of time considering we only win 30% of the business we bid for and in most cases the net revenue is well under £500k.” 6.572 Ofcom therefore concludes that the charges notified in NCCN 500 do not constitute an abuse of BT’s dominant position on the basis that they were part of a concerted strategy on the part of BT to reduce competition. Responses to draft decision: impact on BT’s competitors and BT’s intentions 6.573 Four respondents commented on the impact of NCCN 500 on BT’s competitors and on BT’s intention when it issued NCCN 500. C&W 6.574 C&W stresses that “we do not necessarily accept that it is necessary to show effect or intention for BT’s conduct to be abusive”. Nevertheless, C&W believes that BT’s 204 competitors were adversely impacted by NCCN 500 and that BT was aware that this would be the outcome.436 6.575 C&W argues that Ofcom has not fully analysed the impact that BT’s conduct had on competitors. It states that: “(w)hile Ofcom appears to acknowledge the lack of competitive constraints facing BT and therefore its ability to raise its prices in the manner that it did, Ofcom appears to actually fall short of acknowledging the damage that this caused to BT’s competitors.”437 6.576 C&W submits that the inability of BT’s competitors to respond to the price change “must have negatively impacted BT’s competitors’ ability to effectively compete against BT”. C&W submits that Ofcom has failed to take this into account “other than for concluding that BT held a position of dominance on the market”.438 6.577 C&W notes that, as set out in its original submission, it lost £[] “as a direct result of NCCN 500” and that “other [communications providers] made losses as a direct result of NCCN 500”.439 6.578 C&W submits that Ofcom has been “less robust” than it should have been in assessing the impact of NCCN 500 because the price rises were in effect reversed with NCCN 651, and that Ofcom should have made an assessment of the likely impact of NCCN 500 “over time”. C&W claims that: “we would have been unable to sustain that level of loss in the long run without having to make commercial decisions about our continued ability to stay in the market.”440 6.579 The fact that BT claims it was “exploiting a regulatory loophole” is, C&W argues, no defence for its behaviour “if [BT] understood that to exploit it would be damaging to competition”.441 6.580 C&W states that, as Ofcom acknowledges, BT’s competitors’ responses to NCCN 500 may have been influenced by the knowledge that Ofcom was investigating, and that BT’s competitors may therefore have decided to absorb the additional cost imposed by NCCN 500 even though such a strategy would not have been viable in the long term: “BT knew that any increase in revenue [from NCCN 500] would be a direct result of losses that its competitors would face. We fail to see how this could not equate with an intention to harm competition. It must have been foreseeable by BT that if its competitors would lose revenue in direct proportion to the increase in BT’s revenue, that this would not be sustainable over time and could lead to its competitors exiting the market for the provision of NTS services. This in itself must be an intention to distort competition.”442 436 C&W’s response to the draft decision, page 20. C&W’s response to the draft decision, page 18. 438 C&W’s response to the draft decision, page 19. 439 C&W’s response to the draft decision, page 29. 440 Ibid. 441 C&W’s response to the draft decision, page 51. 442 C&W’s response to the draft decision, page 50. 437 205 6.581 C&W also states that: “while there is no need to demonstrate intention, BT’s stated reasons for NCCN 500 do not constitute defences for its behaviour. If anything they exacerbate it.”443 6.582 C&W submits that it is “implausible” that the fact BT was aware that its competitors would be limited in their ability to retaliate “cannot be considered as anti-competitive” per se. C&W mentions that the CAT, in Napp, held that an infringement may be considered “intentional” if the undertaking knew its behaviour was likely to distort competition.444 CPW 6.583 CPW states that NCCN 500: “represented a substantial transfer of revenue from CPW to BT, which has had a negative overall impact on CPW’s ability to compete...CPW believes that BT introduced NCCN 500 to harm its competitors for the simple reason it had discovered a flaw in the NTS regulatory framework that allowed it to do so.”445 Thus 6.584 Thus claims that it: “experienced a significant adverse financial impact as a result of BT’s decision to raise prices through NCCN 500.”446 6.585 Thus has said that “BT knew from the outset that its competitors were likely to make a pence per minute loss on each call originated to a BT NTS number.” In addition, BT “picked pricing that was uplifted enough to damage their competitors and provide them with a healthy unregulated revenue stream”.447 6.586 Thus believes that Ofcom should undertake further analysis to establish precisely what BT’s intentions were. Thus states that it finds it “surprising” that the information supplied to Ofcom by BT did not appear to contain any internal assessment of BT’s competitors’ likely reactions to the price change, or any internal debate over precisely where prices should be set.448 6.587 Thus states that “we would urge Ofcom to request again details of all BT Group deliberations over BT’s decision to issue NCCN 500 and the level at which to set the pricing…If no further written documentation is forthcoming we believe gathering oral evidence would be appropriate in this instance. This would help to substantiate our view that this was a deliberate and carefully thought out scheme designed 443 C&W’s response to the draft decision, page 51. C&W’s response to the draft decision, page 20. 445 CPW’s response to the draft decision, page 1. 446 Thus’s response to the draft decision, page 1. 447 Thus’s response to the draft decision, page 3. 448 Thus’s response to the draft decision, page 4. 444 206 to inflict damage on competitors, and hence a more serious breach than would otherwise have been the case.”449 6.588 In addition, Thus believes that “it was probably no coincidence that at the same time as NCCN 500 was announced BT Retail embarked on a heavy consumer marketing campaign…..[which] focused on highlighting the pricing differences between BT and their competitors over the pricing of 0870 and 0845 calls. Thus urges “Ofcom to request that BT supply documentation relating to the planning and timing of these specific marketing initiatives and explore any linkage between the campaigns themselves at the advent of NCCN 500.”450 Virgin Media 6.589 Virgin Media states that it “had very limited options to respond” to the price increases notified in NCCN 500. Virgin Media states that suffered a material incremental increase in costs during the period that NCCN 500 was in place, and questions “how such a scale of impact cannot be regarded as competitively damaging”. Ofcom comments on responses 6.590 Ofcom has not expressed a view that it is necessary to demonstrate intent in order for BT’s conduct to be considered an abuse. Rather, Ofcom considered that any intention on BT’s part to distort competition would have been a relevant supporting factor in assessing evidence of an actual or likely effect on competition. 6.591 Ofcom does not dispute that there is no need to show an actual effect on competition in order for an abuse under Article 82 or Chapter II to occur, and agrees that it is sufficient to show that conduct likely to lead to a (detrimental) effect on competition is sufficient to support a finding of abuse. 6.592 Ofcom’s conclusions under each allegation of abuse are that the conduct did not lead to an actual or likely (detrimental) effect on competition. In this case, the conduct complained about was in place for a period of 20 months – a time in which any anticompetitive effects of the conduct might have been expected to have become evident. 6.593 The respondents suggest that the inability of BT’s competitors to respond to NCCN 500 is evidence that they could not compete effectively with BT, and that BT’s conduct is therefore abusive. 6.594 Ofcom did come to the conclusion that BT’s competitors did not have the ability to constrain the pricing behaviour of BT (see paragraphs 5.132-5.135 above). However, while that conclusion means that BT has the ability to set a price that is above a competitive level, it does not necessarily follow that BT did so in this case. Ofcom’s assessment of BT’s conduct shows that it did not have any actual or likely effect on competition. 6.595 C&W suggest that the fact BT was aware that its competitors would be limited in their ability to respond to the price increase was, in itself, “anti-competitive”. 6.596 Ofcom is of the view that even if BT was aware its competitors had limited ability to respond to the price increase notified by NCCN 500, that is not equivalent to being 449 450 Ibid. Ibid. 207 aware that conduct is of such a nature as to encourage a restriction or distortion of competition. Ofcom considers that such an awareness is more relevant to the intent of the conduct. Ofcom’s assessment of the conduct and the actual evidence of what occurred in the relevant market did not show any effect on competition. 6.597 Thus suggests that there must have been extensive debate within BT about the likely response of its competitors to the price increase and the level at which to set prices, and suggests that Ofcom ask BT for this information again. 6.598 Ofcom has no evidence that BT has failed to comply with Ofcom’s original request for information. The internal correspondence supplied by BT in response to Ofcom’s section 26 Notice of 22 April 2005 included discussions of the appropriate level of prices, in that it referred to BT’s decision to set the level of the price change with reference to discounts offered by other OCPs (see paragraph 6.556 above). It also included some analysis of competitors likely reactions, although this was not referred to in Ofcom’s draft decision.451 Ofcom notes that a number of extracts from BT’s internal correspondence were redacted from the draft decision provided to complainants as they contained information that was commercially confidential to BT. 6.599 The respondents state that they were adversely impacted financially by NCCN 500. C&W claims that, although competitors absorbed the additional cost imposed by NCCN 500 in the short term, this would not have been a viable long-term strategy and, had NCCN 500 remained in place, market exit may have occurred. 6.600 The intent of Article 82 and Chapter II is to prevent the distortion of competition. This has been expressly stated in previous cases including Bronner v Media Print: “.. it is important not to lose sight of the fact that the principal purpose of Article [82] is to prevent distortion of competition – and in particular to safeguard the interests of consumers – rather than to protect the position of particular competitors.”452 6.601 This position was also endorsed in the BHB case: “As Jacobs A-G said in Bronner (cited above), the principal of Article 82 of the Treaty is the protection of consumers, […] not of business competitors. In our judgment this is correct, even if it is the competitors and not the consumers who are alleging abuse of dominant position. We need to look at ATR’s immediate interests to the market served by ATR. There is little, if any, evidence that competition in the market is being distorted by the demands made by BHB upon ATR.”453 6.602 In Ofcom’s view, an adverse financial impact on a competitor does not necessarily equate to distortion of the market and our analysis of the conduct did not show that NCCN 500 would have made BT’s competitors unprofitable, thereby leading to a distortion of competition or, in the limit, market exit (see in particular Ofcom’s analysis of C&W’s allegation of margin squeeze at paragraphs 6.37 et seq above). 451 See, for example, items DK16, DK207 (OCPs increasing retail prices), CA435, CA196 (OCPs raising termination charges, DK165 (OCPs refusing to connect calls to BT), BT’s response of 13 May 2005 to Ofcom section 26 Notice of 22 April 2005. 452 [1998] ECR 1-17791 at 7811, paragraph 58. 453 Case No: A3/2006/0126 [2007] EWCA Civ 38, paragraph 215. 208 Conclusion 6.603 The following paragraphs summarise Ofcom’s analysis of BT’s conduct and conclusions on C&W’s various allegations. 6.604 C&W’s three principal allegations were that the price increases notified in NCCN 500: • imposed a margin squeeze on C&W; • discriminated in favour of BT and against C&W on price; and • were excessive. 6.605 Before finding an infringement of Article 82 and/or the Chapter II prohibition, Ofcom must satisfy itself on the evidence that an infringement is more likely than not to have occurred.454 The undertaking being investigated is entitled to the presumption of innocence and to any reasonable doubt that there may be. Ofcom must therefore proceed on the basis of an analysis that is robust and soundly based.455 Margin squeeze 6.606 Beginning with the question of an alleged margin squeeze, Ofcom considered whether an operator as efficient as BT would be able to compete in the relevant downstream markets identified in the investigation, had it faced the additional cost imposed by NCCN 500. 6.607 Communications providers supply calls to NTS numbers as part of a bundle of services including different types of calls (e.g. calls to geographic numbers and mobile numbers) and in some cases access – they do not compete to supply NTS calls only. Ofcom therefore concluded that it was appropriate to conduct its margin squeeze analysis across these service bundles and not, as the respondents to its draft decision suggested, across only the call types affected by NCCN 500. Ofcom carried out a number of different variations of its retail margin squeeze test to capture the different types of communications providers competing with BT in the retail origination and calls markets. 6.608 Ofcom found that, across the relevant service bundles, BT earned high margins, with higher margins on some call types (for example, national and international calls and calls to mobiles) compensating for losses on other services, including those call types affected by NCCN 500. This was true even when Ofcom discounted all the returns associated with local calls. 6.609 Ofcom concluded that, had NCCN 500 applied to all calls originated by BT, the impact on BT’s margin across the relevant product set would have been insufficient to lead to a margin squeeze. 6.610 In addition to its assessment of the margin squeeze allegation outlined by C&W, Ofcom carried out an analysis of whether the additional revenue that BT gained from NCCN 500 could have enabled it to fund higher payments to its NTS service providers that its competitors were unable to match, leading to a margin squeeze in NTS hosting (see paragraphs 6.158 et seq above). As in the retail margin squeeze 454 Napp Pharmaceutical Holdings Limited v Director General of Fair Trading [2002] CAT 1, paragraph 109. 455 JJB Sports plc and Allsports Limited v Office of Fair Trading [2004] CAT 17, paragraphs 191-199. 209 analysis, Ofcom considered different variants to capture the different bases on which other communications providers compete with BT; in this case whether they are providing NTS hosting in voice or data. 6.611 Ofcom found that BT’s conduct did not lead to a margin squeeze in NTS hosting. Discrimination 6.612 On the question of discrimination by BT, Ofcom considered whether BT’s charges for NTS call termination, as notified in NCCN 500, might amount to anti-competitive price discrimination. 6.613 As set out at paragraph 6.249 et seq above, BT appears to have acknowledged internally that NCCN 500 imposed a charge on its competitors that was different from the charge paid by its downstream business. This was of concern to Ofcom However, Ofcom’s detailed analysis of financial data supplied by BT showed that there was no “hard” charge in BT’s account that might provide a relevant comparator for the charges that BT’s competitors were required to pay while NCCN 500 was in force. There was not robust evidence of a “hard” charge to BT that was different from the charge that applied to BT’s competitors. 6.614 Ofcom therefore concluded that, in the circumstances the relevant test for determining whether BT was discriminating in favour of itself in an anti-competitive manner is whether BT would have been able to make a profit had it paid the charges notified in NCCN 500, taking into account profits earned on all the relevant services, i.e. whether BT’s conduct led to the operation of a margin squeeze on other communications providers. As set out above, Ofcom concluded that NCCN 500 did not lead to a margin squeeze on BT’s competitors in retail access and calls markets. Excessive pricing 6.615 As set out at paragraphs 6.323 et seq above, Ofcom considered the tests for excessive pricing, to the extent that they were relevant in this case. 6.616 As noted at paragraph Error! Reference source not found. above, Ofcom is mindful of the Court of Appeal’s decision in the BHB case. Ofcom did not find any evidence that BT’s conduct in this case led to a distortion of competition or that it led to an adverse effect on consumers generally. 6.617 The excessive pricing tests offered conflicting evidence as to whether the prices notified in NCCN 500 were excessive. Ofcom found that BT’s prices appeared to be significantly above its FAC, but below SAC. 6.618 Overall, in the circumstances of this case and the need for compelling evidence, Ofcom concluded that there was insufficient evidence to demonstrate to the required standard of proof that the prices notified in NCCN 500 were excessive and thereby constituted an abuse of BT’s dominant position. 6.619 Having investigated each aspect of C&W’s complaint, applied appropriate tests and examined the relevant case law, Ofcom has concluded that there are no grounds for action in this case. 210 Annex 1 1 NCCN 500 NETWORK CHARGE CHANGE NOTICE For Number Translation Services NCCN NUMBER: 500 SERVICE: BT LoCall Services BT NationalCall Services BT Schools 0820 Service EFFECTIVE DATE: 1st May 2004 Associated Terms and Conditions are in Schedules 105, 106, 111, 116 and 168 of Annex C of the Network Charge Control Standard Interconnect Agreement dated 1st October 1997 SUBMITTED TO OFTEL ON: 1st April 2004 BY: Michael Barford ROLE: Manager, Regulatory Pricing BT Wholesale SIGNED: Michael Barford 1 City Place Gatwick West Sussex RH6 0PA Tel 01977 593648 211 Network Time of Day Gradient Retail Time of Day Gradient Current Charge P/Min Proposed Charge P/Min @ 01/05/2004 Percentage Change % BT LoCallTM Short Duration Calls 0345 & 0845 Daytime Evening Weekend 1.459 0.668 0.526 1.491 0.605 0.513 5.0674 1.2953 1.3724 6.9482 1.8274 1.9046 37.1% 41.1% 38.8% BT LoCallTM Long Duration Calls 0345 & 0845 [NOTE 2] Daytime Evening Weekend 1.459 0.668 0.526 1.491 0.605 0.513 1.7517 0.2809 0.3580 2.5682 0.4874 0.5646 46.6% 73.5% 57.7% BT NationalCallTM Calls 0990 & 0870 [NOTE 1] Daytime Evening Weekend 1.459 0.668 0.526 1.491 0.605 0.513 5.8137 3.0929 0.9723 6.0462 3.2144 1.0166 4.0% 3.9% 4.6% BT 0820 Internet Caller Long Duration Calls (covers Schools Internet, Primary Schools Internet, Public Institutions Internet) Daytime Evening Weekend 1.459 0.668 0.526 1.491 0.605 0.513 0.0000 0.2809 0.3580 0.0000 0.4874 0.5646 0.00% 73.5% 57.7% BT 0820 Internet Caller Short Duration Calls (covers Schools Internet, Primary Schools Internet, Public Institutions Internet) Daytime Evening Weekend 1.459 0.668 0.526 1.491 0.605 0.513 0.0000 1.2953 1.3724 0.0000 1.8274 1.9046 0.00% 41.1% 38.8% BT Number Translation Services Notes: 1. 2. These charges shall also apply to BT International Incoming Calls to the service and number ranges specified. These charges shall also apply to BT International Incoming Calls to the Short and Long duration services and number ranges specified. 212 Annex 2 2 NCCN 651 NETWORK CHARGE CHANGE NOTICE For Number Translation Services NCCN NUMBER: 651 SERVICE: BT LoCall Services BT NationalCall Services BT Schools 0820 Service EFFECTIVE DATE: 1st January 2006 Associated Terms and Conditions are in Schedules 105, 106, 111, 116 and 168 of Annex C of the Network Charge Control Standard Interconnect Agreement dated 1st October 1997 SUBMITTED TO OFTEL ON: 2nd November 2005 BY: Zack Thompson ROLE: Regulatory Pricing BT Wholesale SIGNED: Zack Thompson PP W2.157E BT Brentwood One London Road Brentwood, Essex CM14 4QP Tel 01277 323461 213 Network Time of Day Gradient Retail Time of Day Gradient Current Charge P/Min Proposed Charge P/Min @ 01/01/2006 Percentage Change % BT LoCallTM Short Duration Calls 0345 & 0845 Daytime Evening Weekend 1.459 0.668 0.526 1.536 0.549 0.498 6.9482 1.8274 1.9046 5.0185 1.9947 2.1018 -27.8% 9.2% 10.4% BT LoCallTM Long Duration Calls 0345 & 0845 Daytime Evening Weekend 1.459 0.668 0.526 1.536 0.549 0.498 2.5682 0.4874 0.5646 2.1076 0.4535 0.4574 -17.9% -7.0% -19.0% BT NationalCallTM Calls 0870 Daytime Evening Weekend 1.459 0.668 0.526 1.536 0.549 0.498 6.0462 3.2144 1.0166 5.8119 3.0862 0.9867 -3.9% -4.0% -2.9% BT 0820 Internet Caller Long Duration Calls Daytime Evening Weekend 1.459 0.668 0.526 1.536 0.549 0.498 0.0000 0.4874 0.5646 0.0000 0.3888 0.4574 0.00% -20.2% -19.0% BT 0820 Internet Caller Short Duration Calls Daytime Evening Weekend 1.459 0.668 0.526 1.536 0.549 0.498 0.0000 1.8274 1.9046 0.0000 1.9947 2.1018 0.00% 9.2% 10.4% BT Number Translation Services 214 Annex 3 3 Calculation of critical elasticity for calls to 0845 numbers A3.1 Ofcom calculated what value of retail price elasticity would be required for NCCN 500 to be unprofitable for BT with respect to 0845 calls. Ofcom has termed this value the “critical elasticity”. This depends partly on whether the price increase is assumed to be passed on at its full ppm value on all 0845 calls, or “diluted” to reflect the fact that it only applied to BT terminated calls which are some []% of the total. A key variable in this calculation is therefore the dilution parameter. A3.2 The calculation is based on 0845 calls since this is where BT applied the biggest proportionate price increases in NCCN 500. A3.3 The following are the key variables in the calculation: • E = the retail own price elasticity of demand for 0845 calls • X = the proportion of the termination charge increase under NCCN 500 assumed to be passed on in the retail price. • W1 = BT’s average gross termination charge for 0845 calls immediately before NCCN 500 based on BT’s price list • W2 = BT’s average gross termination charge for 0845 calls immediately after NCCN 500 based on BT’s price list • P1, = the pre-NCCN 500 average retail price for 0845 of a major OCP, ntl, assuming BT’s average time of day profile and an average call duration of one hour. It is necessary to assume an average call duration in order to calculate an average call set-up charge per minute. This unrealistically long average call duration would tend to understate the pre-NCCN 500 retail price and consequently also the extent to which a BT price increase would be profitable • P2 = P1 + (W2 -W1 )*X, the retail price that would be charged following NCCN 500 • Q1 = the volume of 0845 calls at price P1 • Q2 = Q1 * (1+ E * (P2 - P1)/P1), the volume of 0845 calls at price P2, A3.4 Ofcom then calculated values of E such that profits are unchanged for values of X equal to 1) []% and 2) 100%. A3.5 Note that, because this does not allow for the reduction in volumes to lead to cost savings for BT, other than savings from reduced revenue share payments, this calculation would tend to understate the extent to which a BT price increase would be profitable and produce a lower bound for the critical value of elasticity. 215 Annex 4 4 Related markets A4.1 For the purposes of this investigation, Ofcom has considered: • the relevant market in which BT may hold a dominant position and in which the conduct in question has taken place, i.e. the market in which NTS call termination/hosting falls; • any other upstream markets, in which BT’s position may influence the extent to which it faces competitive constraints in the relevant market; and • downstream markets in which any effect of BT’s conduct in NTS call termination/hosting may be felt. Retail calls to NTS numbers A4.2 This section discusses the downstream market(s) in which the effect of BT’s conduct in this case might be felt. A4.3 C&W alleged that NCCN 500 would have the effect of creating a margin squeeze between BT’s wholesale charges and its retail call prices, since it increased the costs incurred by OCPs other than BT in providing retail calls to 0845 and 0870 NTS numbers.456 These services, referred to in the following discussion as “retail NTS calls” are downstream of NTS call termination/hosting because NTS call termination is a necessary input to a retail end-to-end NTS call. Product market A4.4 In 2005 Ofcom carried out an analysis that briefly considered the relevant market for retail NTS calls (“the August 2005 review”).457 A4.5 In the August 2005 review, Ofcom outlined its view that calls to NTS numbers are part of a product market that is distinct from the markets for other call types. A4.6 Ofcom argued in the August 2005 review that fixed and mobile calls are in separate markets, because: A4.7 456 • demand-side substitution between the two call types is restricted by differences in price and functionality; and • barriers to entry prevent constraints between the two call types being imposed by means of supply-side substitution. Ofcom argued that calls to non-geographic numbers are not in the same market as calls to geographic numbers, because: With only BT-terminated calls affected by NCCN 500. Review of BT’s network charge controls, statement of 18 August 2005 (see footnote 14), paragraph A5.20 et seq. This analysis was based on a summary of the narrowband market reviews carried out by Oftel in 2003 (see http://www.ofcom.org.uk/consult/condocs/narrowband_mkt_rvw/). 457 216 A4.8 A4.9 • the two call types are not substitutable from a demand-side perspective because they are used for different purposes, e.g. it would not be possible, using a nongeographic number, to reach the same called party as could be obtained on a geographic number (and vice versa); and • there is no scope for supply-side substitution between the two call types to provide further constraints because all suppliers of retail non-geographic calls are already supplying geographic calls. Ofcom argued that calls to different types of non-geographic numbers (e.g. DQ, PRS, other NTS ranges) are in different markets since: • these different call types are not substitutable from a demand-side perspective, since they are used for different purposes (in a manner analogous to geographic and non-geographic calls); and • there is no scope for supply-side substitution between different types to provide further constraints, because all suppliers of non-geographic calls are already supplying all non-geographic call types. However, the issue of whether there might be distinct markets within the broader category of calls to NTS numbers was not considered. A4.10 Ofcom also concluded that residential and business calls are in separate markets, since: • there is limited scope for demand-side substitution between the two call types, given that suppliers are able to identify residential and business customers, and charge different tariffs; and • there is limited scope for supply-side substitution between the two because of the tendency of residential and business customers to be located in distinct geographic areas.458 A4.11 Taken together, the arguments set out in the August 2005 review suggest that there are two distinct product markets for: • NTS calls made by business users; and • NTS calls made by residential users. A4.12 It is, however, important to consider two further issues, namely: i) whether narrower definitions might be appropriate, i.e. whether there is a distinct product market for retail calls to 0845 numbers, and so on; and ii) what implications, if any, the buying patterns of these services have for market definition. A4.13 When carrying out a SSNIP test analysis for calls from fixed lines, the appropriate starting point, as outlined in Ofcom’s most recent full review of retail markets, is for 458 Paragraph A5.28 et seq. of the August 2005 review. 217 calls between a pair of individuals, i.e. a call from person X to person Y.459 Such a market definition will typically not be broadened to encompass calls to other individuals by demand-side substitution, but might be broadened by supply-side substitution to the extent that cost-based termination is available, and it does appear that suppliers of retail calls face a common pricing constraint across all recipients in setting call prices within a call type (e.g. local calls, calls to mobiles, and so on). A4.14 Taken in isolation, this factor might arguably suggest that there is a distinct economic market for every call type across which a common pricing constraint applies, i.e. one market for local calls, another for calls to mobiles, another for calls to 0845 numbers, and so on. A4.15 However, in relation to point ii), it is important to note that calls to NTS numbers are only ever bought as part of a package of services. This suggests that competition tends to take place within a market that includes a larger variety of products and service than calls to NTS numbers alone (a fact arguably supported by the widely differing prices for these calls on different networks), with providers competing against each other to supply a bundle of services. A4.16 There are therefore compelling reasons to think that defining narrow calls markets for these call types would be inappropriate. For example, when considering one of the key determinants of market power, namely the scope for market entry, it would not be the case that entry would only take place in relation to 0845 calls following an increase in the price of this type of calls. A4.17 As discussed in Section 6, communications providers use a number of different business models in order to sell retail calls services to end users, namely: • all retail calls plus access; • all retail calls but not access (CPS all calls); and • some retail calls but not access (CPS national and/or international calls). A4.18 The smallest bundle of services within which retail NTS calls are provided therefore includes a number of other call types, in particular local calls and calls to mobiles. A4.19 Downstream retail market definitions are not material to assessing whether the charges notified in NCCN 500 charges were excessive or discriminatory. This is because the key reason for defining the relevant retail market(s) definition is to assist with Ofcom’s assessment of C&W’s allegation that NCCN 500 imposed a margin squeeze on OCPs other than BT. A4.20 As explained in Section 6, Ofcom’s view is that the appropriate breadth of a test to ascertain whether BT’s conduct led to a margin squeeze in wholesale call origination is not solely determined by the scope of the economic market(s) in which any margin squeeze would be felt. Having analysed BT’s costs and revenues for 2004/05, Ofcom concluded that NCCN 500 did not create a margin squeeze. 459 Review of the fixed narrowband wholesale exchange line, call origination, conveyance and transit markets, statement of 26 August 2003 (see footnote 66)Fixed Narrowband Retail Services Market Identification and analysis of markets, Determination of market power and Setting of SMP conditions Explanatory statement and Notification, 26 August 2003, http://www.ofcom.org.uk/static/archive/oftel/publications/eu_directives/2003/fix_narrow_retail0803.pdf 218 A4.21 On this basis, Ofcom’s view is that it is not necessary to formally conclude on the relevant retail market definition for the purpose of this decision, beyond the fact that calls to NTS numbers are always supplied as part of a bundle of services. Given that NCCN 500 did not create a margin squeeze, Ofcom’s view is that there are separate downstream markets for services sold to residential and business customers, and that those markets should be further delineated on the basis of a bundle that is no narrower than the minimum bundle of services in which retail NTS calls may be offered as a component, principally the following retail products within the UK: • local calls; • calls to mobiles; and • calls to NTS numbers. A4.22 It is also worth noting that some (voice) NTS calls are originated on one of the five mobile networks. Ofcom has considered whether it is necessary to carry out a market definition exercise for such calls and concluded that it is not, for the following reasons. First, Ofcom’s upstream market definition encompasses the termination of calls originated on both fixed and mobile networks callers, meaning that considering the market definition for mobile calls would not have any further impact on the upstream market definition. Second, it has not been suggested that NCCN 500 could have an impact on competition in calls markets that include calls made by mobile customers. Geographic market A4.23 As noted in Section 4, the SSNIP test is relevant for defining the geographic market as well as the product market. In this context, the SSNIP test assesses whether a hypothetical monopolist in one geographic area would be constrained by consumers switching to other suppliers in other areas, or by other suppliers switching to the area in which the hypothetical monopolist supplies. A4.24 Demand-side substitution between different geographic areas tends not to be feasible for any call type (e.g. customers wanting to make a call from Town X to Town Y do not view a call to the same destination but made from Town Z as a substitute). Supply-side substitution may, however, be feasible, depending on the availability of wholesale inputs on reasonable terms. The identification of a common pricing constraint is therefore often an important factor to consider in defining the relevant geographic market. A4.25 For all retail call types, including NTS calls, BT currently charges a uniform price to callers throughout the country. This means that any response by BT to competition in a given area in the form of lower prices would apply throughout the UK (excluding the Hull area), and therefore that the geographical extent of the relevant markets should be regarded as the whole of the UK (excluding the Hull area), and the Hull area. This means that the extent of the geographical market is the whole of the UK, excluding the Hull area where a uniform constraint holds. A4.26 For the purposes of this investigation, Ofcom has not considered this definition in any great depth, since the geographic scope of downstream markets does not have any key implications for the market definition for termination/hosting, with TCPs terminating traffic on behalf of communications providers based throughout the UK, or indeed on the scope of the relevant margin squeeze test, since Ofcom’s analysis 219 has considered the impact of NCCN 500 on competition between BT and all the types of competing providers which operate on a mostly national basis. Other wholesale calls markets Wholesale narrowband call origination The relevant product market A4.27 Wholesale call origination is a key input into all of the retail call types described in the previous section. A4.28 Ofcom’s view is that there is a single market for narrowband call origination (which is used as an input into all of these call types), and that BT has SMP in that market. This view was set out in Ofcom’s 28 November 2003 statement Review of the fixed narrowband wholesale exchange line, call origination, conveyance and transit markets and confirmed in Ofcom’s 18 August 2005 statement Review of BT’s network charge controls.460 A4.29 The delineations discussed in the previous section, based on different call types (e.g. national, international) and customer types (business/residential) do not apply to the corresponding wholesale call origination product, as call origination is a uniform product and does not differ according to the ultimate destination (e.g. a fixed geographic number or a mobile number) of the call. While the charge for NTS call origination differs from that for other types of calls in that it includes an explicit contribution to retail cost recovery, it is part of a cluster of origination services bought by suppliers of retail calls from suppliers of origination. A4.30 OCPs compete to provide a range of services over a single access line, rather than limited services across many access lines. Such competition means that customers choose the provider that can provide the range of services at the lowest price. The fact that retail call origination is always purchased as part of a bundle provides further support for the view that all call origination services should be treated as forming part of a single product market. A4.31 Finally, a common pricing constraint applies to the call origination services that are supplied as an input into all retail call types, including services sold to both residential and business customers. A4.32 Ofcom’s conclusion is that there is a single market for wholesale call origination, used as an input into all call types sold to both residential and business customers. The relevant geographic market A4.33 The application of the SSNIP test to wholesale call origination (and also to transit as discussed below) implies a potentially very narrow geographic market definition. This is because the price of call origination provided by a hypothetical monopolist at, for example, at a particular exchange: 460 Review of the fixed narrowband wholesale exchange line, call origination, conveyance and transit markets, statement of 28 November 2003 (see footnote 12) and Review of BT’s network charge controls, statement of 18 August 2005 (see footnote 14). 220 • would not be constrained by demand-side substitution to call origination in other areas, since customers are only interested in obtaining call origination from a particular location; and • would not be constrained by supply-side substitution by suppliers from other areas, because of the significant time and expense that would be required for supply-side substitution to take place from other areas. A4.34 Ofcom’s view is that the market for call origination (outside the Hull area, where BT does not operate) is national in scope. The key aspect of BT’s position in the market for call origination in the context of NCCN 500 is its regulatory requirement to supply origination for NTS calls at cost-based charges. This remedy has been applied by Ofcom on a uniform basis in all geographic areas. It is not the case that a TCP in any one area in the country is more or less able to negotiate with BT when setting its gross termination charges. The analysis that follows therefore assumes a single national market for call origination (outside the Hull area). Conclusion A4.35 The relevant market is call origination in the UK excluding the Hull area. BT’s position on the relevant market A4.36 Ofcom has designated BT has having SMP in the market for wholesale call origination in the UK excluding Hull. A4.37 The primary source of this position of dominance is BT’s position as the incumbent communications provider, and the only undertaking with a ubiquitous access network. No other provider has so far found it commercially attractive to attempt to replicate this network reach. A4.38 BT’s market share of call origination has fluctuated between about 70% and 80% for a number of years.461 A4.39 No other provider is in a position of countervailing buyer power, since they all compete with BT at the downstream level, and BT’s own retail business is by some distance the biggest buyer of call origination A4.40 Ofcom imposed the NTS call origination condition on BT in order to promote competition in NTS, given BT’s SMP in the market for wholesale call origination. Single transit The relevant product market A4.41 Single transit is a service provided by a transit provider in respect of calls originating and terminating on networks other than its own, where the OCP and TCP are directly connected to the same transit provider’s tandem exchange. The call is therefore transited through a single tandem exchange. A4.42 Single transit is in an economic market which is distinct from that for other wholesale calls services such as inter tandem transit (“ITT”) (see below). Since Ofcom has 461 See for example Review of BT’s network charge controls, statement of 18 August 2005 (see footnote 14), paragraph A5.45. 221 determined that no provider has SMP in the market for ITT, Ofcom’s analysis of single transit focuses on constraints imposed on single transit by ITT. A4.43 A hypothetical monopolist of single transit: • would not be constrained by demand-side substitution into ITT. Use of ITT would involve a transmission element, involving higher costs, which may not be required if the operator purchased single transit; and • would not be constrained by supply-side substitution from suppliers of ITT, since such firms would require a much higher level of connectivity with other providers to be able to offer single transit services. This would require significant investment and build to a large number of other providers’ tandem exchanges. It is therefore very unlikely that a provider of ITT would be able to supply in a way that constrained the prices of a hypothetical monopolist of single transit. A4.44 Based on these considerations, Ofcom’s view is that there is a distinct product market for single transit. The relevant geographic market A4.45 BT is the only supplier of single transit services. On this basis, Ofcom considers that BT would have a dominant position in supply of single transit services throughout the UK, regardless of whether the UK was viewed as being a single geographic market or encompassing more than one geographic market. For the purposes of this investigation, Ofcom regards the market for single transit as national (excluding the Hull area), based on the uniformity of competitive conditions throughout the UK (excluding the Hull area). Conclusion A4.46 The relevant market is single transit in the UK excluding the Hull area. BT’s position on the relevant market A4.47 Ofcom has designated BT has having SMP in the market for single transit in the UK excluding Hull. A4.48 The primary source of this position of dominance is BT’s position as the incumbent communications provider, and the only company with a ubiquitous access network. A4.49 BT is the only provider that provides single transit to any notable extent, meaning that its share of the market is close to 100%. While some communications providers are now interconnected to BT’s tandem switches, at current volume levels the business case for interconnection between competing providers is commercially unattractive. Inter tandem transit The relevant market and BT’s position within the relevant market A4.50 ITT is the service a communications provider supplies to others to enable them to convey calls between its tandem exchanges when a call originates and terminates on networks other than its own. 222 A4.51 In the August 2005 review, Ofcom set out the view that there is a single market for inter-tandem conveyance and ITT on fixed public narrowband networks in the UK excluding the Hull area and that this market is effectively competitive. This represented a change from Oftel’s previous consideration of the market (the November 2003 market review referred to above), which concluded that BT had SMP in the markets for inter-tandem conveyance and transit on fixed public narrowband networks; and single transit on fixed public narrowband networks in the UK excluding the Hull area.462 A4.52 Ofcom’s assessment that this market was competitive was based on: • increasing interconnection to BT exchanges by competing infrastructure providers; and • BT’s declining share of ITT (BT’s share of ITT has been consistently below 40% since Q3 2002/03). A4.53 Since BT’s position in ITT does not form a key part of this investigation, Ofcom does not consider it necessary to reconsider this market definition or assessment for the purpose of assessing the impact of NCCN 500 on competition. 462 Review of the fixed narrowband wholesale exchange line, call origination, conveyance and transit markets, statement of 28 November 2003 (see footnote 12). 223 Annex 5 5 ROCE and ROS A5.1 In order for a margin to be considered adequate, it must be sufficient to allow the firm to earn its required minimum rate of return on capital. A5.2 When using historical accounting information to assess profitability, a commonly used measure of the rate of return actually earned is return on capital employed (“ROCE”). Using this measure, an activity is considered not to be profitable if the ROCE that it earns falls below the level that would be required by investors in order to compensate them for any risk incurred by investing in the activity, i.e. below the activity’s weighted average cost of capital (“WACC”).463 A5.3 The use of ROCE as a measure of profitability is not, however, appropriate in all circumstances, particularly in the case of activities that inherently require relatively little physical or working capital. This applies strongly to BT’s retail calls and access activities, since outpayments to TCPs, including its own wholesale business, account for a high proportion of the total underlying cost base (see Annex 7). Most of BT’s fixed capital is included in its wholesale business, with the required return included in the outpayment from BT Retail. At the retail level therefore, capital employed is typically small relative to turnover or even negative. A5.4 A frequently-used alternative in such circumstances is Return on Sales (“ROS”).464 ROCE is generally considered to have greater economic significance than ROS, which has the disadvantage that there is no theoretical benchmark with which to compare it. This means that the “required” return on sales will vary directly with the capital intensity of the firm and its cost of capital. It is therefore necessary to benchmark the return on sales with that earned by other firms of similar capital intensity and risk. A5.5 Competition authorities have provided indications of an appropriate return on sales where capital intensity is low. In the 1999 report by the then Monopolies and Mergers Commission (“MMC”) on BT, instead of using return on capital employed, the MMC based its assessment of the profitability of BT’s calls business on return on turnover (“ROT”). 465 The MMC considered that the reason this approach could be applied to BT’s call business was the “very high proportion of turnover accounted for by boughtin services”’.466 The MMC took the view that a return on turnover of 1.5% would be 463 See, for example, Ofcom’s approach to risk in the assessment of the cost of capital, 18 August 2005 (see footnote 371), and OFT414a. 464 Also referred to (as in BSkyB) as Return on Turnover, RoT. 465 British Telecommunications Plc: A report on a reference under section 13 of the Telecommunications Act 1984 on the charges made by British Telecommunications Plc for calls from its subscribers to phones connected to the networks Cellnet and Vodafone. MMC, 21 January 1999. The MMC concluded (paragraph 2.113) that calculating a return on net assets employed was an unreliable basis for setting a reasonable return as the mean net assets employed in call activities are not only relatively small but they consisted for the most part of working capital items which could fluctuate considerably from year to year. 466 Paragraph 2.116. In the case of BT, ‘over 80% of the retail price to consumers represents the cost of bought in services’ (Paragraph 2.112). 224 appropriate for BT’s calls-to-mobiles activity. In contrast, in its report on Scottish Hydro-Electric plc a return of 0.5% was adopted.467 The MMC considered that “the potential for competition from new operators and the speed with which it could impact on BT are factors which we believe differentiate BT’s calls to mobile activity from the circumstances of Scottish Hydro-Electric.”468 A5.6 For these reasons, in the historical accounting analysis set out in Section 6, Ofcom uses ROS as a reasonable measure of the minimum required return on investment, and has used as a benchmark a rate consistent with the MMC’s analysis.469 A5.7 The figures in the following Annex show the importance of payments and the low capital employed in supplying retail services to residential, business and all customers respectively. These tables show that most, if not all, of the BT retail services in question will have ROCEs that do not lend themselves to meaningful comparison with company cost of capital estimates: negative capital employed means that high returns lead to negative ROCE; whereas low (positive) capital employed leads to services having triple-digit ROCE. The usefulness of ROCE as a measure of profitability in this case is therefore significantly reduced. 467 Scottish-Hydro-Electric plc, A Report on a reference under section 12 of the Electricity Act 1989, MMC, June 1995, paragraph 2.85. 468 Scottish-Hydro-Electric plc, A Report on a reference under section 12 of the Electricity Act 1989 (see footnote 467), paragraph 2.117 469 OFT 657, paragraphs 1.24-1.27. 225 Annex 6 6 The importance of capital employed in BT’s retail services A6.1 Shaded cells indicate inputs supplied by BT, with unshaded cells describing calculations carried out by Ofcom. Table 31: Payments and capital employed – BT retail products (products sold to residential customers), 2004/05 Charges from BT Wholesale Markets £m Payments Total CCA Operating Costs Charges from Mean BT Wholesale capital & payments employed % of TC £m £m % £m Products: P501,P502,P503,P504 2,170 (IDD Residential) - 2,360 [] [] P241 Inland Calls BT to Mobile Residential 21 373 466 [] [] P317 PSTN Local calls: Residential 137 17 253 [] [] P319 PSTN National calls: Residential 86 10 142 [] [] P455 PSTN Res Connections (Std) 115 - 139 [] [] P461 Residential Telephone Line Rentals 2,170 - 2,360 [] [] TOTAL 4,699 400 5,718 [] [] Source: BT 226 Table 32: Payments and capital employed – BT retail products (products sold to business customers), 2004/05 Payments Total CCA Operating Costs Charges from BT Wholesale & payments - % of TC Mean capital employed £m £m £m % £m P014 IDD Calls to Overseas Fixed Business [] [] [] [] [] P240 Inland Calls BT to Mobile Business [] [] [] [] [] P316 PSTN Local calls: Business [] [] [] [] [] P318 PSTN National calls: Business [] [] [] [] [] P454 PSTN Bus Connections (Std) [] [] [] [] [] P460 Business Telephone Line Rentals [] [] [] [] [] TOTAL [] [] [] [] [] Charges from BT Wholesale Markets Source: BT 227 Table 33: Payments and capital employed – BT retail products (products sold to both customer types), 2004/05 Payments Total CCA Operating Costs Charges from BT Wholesale & payments % of TC Mean capital employed £m £m £m % £m P056 BT LinkLine [] [] [] [] [] P059 BT ValueCall [] [] [] [] [] P171 Freephone BT to OLO [] [] [] [] [] P313 BT to OLO Lo-call [] [] [] [] [] P314 BT to OLO Nationalcall [] [] [] [] [] P315 BT to OLO Valuecall [] [] [] [] [] P346 BTnet Services - Dial Access [] [] [] [] [] TOTAL [] [] [] [] [] Source: BT 228 Charges from BT Wholesale Markets Annex 7 7 Alleged margin squeeze in call origination: data sources and calculations A7.1 Shaded cells indicate inputs supplied by BT, with unshaded cells describing calculations carried out by Ofcom. Note that BT accounts for BT-to-BT and BT-toTCP NTS calls separately. Table 34: alleged margin squeeze in call origination: data sources and calculations Data series Source/notes (a) FAC business and residential financial statements, i.e. profit and loss accounts and statements of capital employed, for various geographic call types. BT response to section 26 Notice of 11 October 2005. (b) FAC business and residential financial statements for access products, i.e. connections and rentals. BT response to section 26 Notice of 11 October 2005. (c) FAC financial statements for BT-OCP NTS calls – combined business and residential: P171 (Freephone); P313 (Lo-call); P314 (Nationalcall); P315 (Valuecall i.e. PRS). BT response to section 26 Notice of 11 October 2005. (d) Total volume (both data and voice traffic) data for BT originated traffic – split by BT-OCP and BT-to-BT by NTS number range. BT response to section 26 Notice of 11 October 2005. (e) Ofcom-derived FAC financial statements for retailing of all NTS calls by number range (i.e. both BT-to-BT and BT-to-TCP calls, rather than just BT-to-TCP, as in the data supplied by BT. These statements were constructed by multiplying all of the key financial measures provided in (c), i.e. costs and revenues, by a multiplier equal to the ratio of the sum of BTto-BT and BT-to-TCP call volumes divided by BT-to-TCP call volumes. (c); (d) (f) Information on the split between traffic (minutes) originated by business and residential customers in the case of geographic calls (local and national). BT response to section 26 Notice of 11 October 2005. Also voice-data split for all BT-to-BT 084x calls (15%-85%). 229 (g) (k) Data series Source/notes Ofcom-derived business and residential financial statements for each NTS number range, calculated by splitting all figures (e.g. turnover, opex, payments) in (e) on the basis of (f). For 084X, Ofcom assumed that all data traffic (and associated revenue/costs) was residential. The financial impact of NCCN 500 split by business and residential if higher rates were faced on all BT-to-BT traffic. C&W submission of 15 March 2005. For all other NTS number ranges, including 084X voice, Ofcom assumed the same businessresidential proportions as per the average of BT-originated local and national geographic calls. BT response to section 26 Notice of 11 October 2005. Based on: 230 (i) total volumes of BT-to-BT calls; BT response to section 26 Notice of 25 August 2005. (ii) a time of day split for the above; and NTS Calculator. (iii) the differential between BT and altnet termination prices. Volume splits implied in (f), (g). Annex 8 8 Alleged margin squeeze in call origination: sensitivity analysis A8.1 As discussed in Section 6, NCCN 500 did not lead to a margin squeeze across a relevant set of services. This is because the set of calls that NCCN 500 applied to represents a small proportion of the retail calls and access services sold by BT to its residential and, especially, its business customers. A8.2 This Annex sets out the sensitivity analysis carried out by Ofcom to test this result, which consists of three further tests: • NTS calls assumed to account for a significantly larger proportion of calls originated by BT; • business customers assumed to generate as many data calls to NTS numbers as they do voice calls; • traffic is originated by BT according to the same time of day profile as another OCP. NTS calls account for a larger proportion of all calls A8.3 Ofcom calculated what the impact of NCCN 500 would have been if NTS calls had accounted for a significantly larger proportion of the calls originated by (and hence of the value of) the services sold by BT. A8.4 This analysis assessed the extent to which NCCN 500 might have had an impact on any competitor generating a particularly large proportion of NTS calls. Ofcom has no reason to think that any of BT’s competitors actually operates such a business model. A8.5 The results below are calculated on a fully allocated cost basis, and calculated across all call types plus access. 231 Table 35: BT’s ROS, business multiplier Pre-NCCN 500Post-NCCN 500 1 2 3 4 [] 5 6 7 8 9 10 Table 36: BT’s ROS, residential multiplier Pre-NCCN 500Post-NCCN 500 1.00 1.25 1.50 [] 1.75 2.00 2.25 2.50 2.75 3.00 3.25 A8.6 The results set out in the tables above show that, for services sold to business customers, BT would need to generate over five times as many NTS calls as it actually does in order to create margins of an order that might be suggestive of NCCN 500 imposing a margin squeeze across all calls and access. For services sold to residential customers, BT would need to generate roughly twice as many NTS calls in order to create margins of an order that might be suggestive of NCCN 500 imposing a margin squeeze across all calls and access. A8.7 As, in practice, none of BT’s competitors have an output mix resembling those described in the tables above and, moreover, this test is based on FAC rather than incremental cost, these results suggest that Ofcom’s findings are robust. Sensitivity: business customers generate same proportion of data calls as voice calls A8.8 Ofcom assumed in its analysis that business customers do not generate any data traffic to NTS numbers affected by NCCN 500 – i.e. that businesses do not use dialup internet access. A8.9 In fact, they do generate some: Ofcom’s SME tracking research (which looks at businesses in the UK with between 1 and 250 employees with £50K turnover or 232 more) found from October-December 2005 that a number of businesses of this size do still use dial-up internet: • 85% claimed to be connected or in the process of connecting to the internet (unweighted base size = 896); and • among those connected to the internet, 19% claim to be using 'dial-up/ordinary phone line' (unweighted base size = 839). A8.10 The survey did not record what proportion of this was metered traffic, i.e. calls affected by NCCN 500. A8.11 Ofcom therefore assumed for this sensitivity analysis that business customers account for the same share of NTS data calls as they do of voice traffic, which is likely to be an overestimate and therefore a conservative assumption for these purposes. Table 37: Results (base) Turnover [a] [b] [c] [d] [e] [f] [g] [h] [i] Base case (all calls + access) Calls only CPS all calls CPS national calls Margin of [c] over [d] CPS international calls Margin of [c] over {[d]+[f]} CPS all calls except local Margin [g] excl. local calls [] Turnover [a] Base case (all calls + access) [b] Calls only [c] CPS all calls [d] CPS national calls [e] Margin of [c] over [d] [f] CPS international calls [g] Margin of [c] over {[d]+[f]} [h] CPS all calls except local [i] Margin [g] excl. local calls Residential customers Return ROS pre-NCCN500 post-NCCN500 pre-NCCN500 post-NCCN500 Business customers Return ROS pre-NCCN500 post-NCCN500 pre-NCCN500 post-NCCN500 [] 233 Table 38: Results – assume that some business NTS data traffic Turnover [a] [b] [c] [d] [e] [f] [g] [h] [i] Base case (all calls + access) Calls only CPS all calls CPS national calls Margin of [c] over [d] CPS international calls Margin of [c] over {[d]+[f]} CPS all calls except local Margin [g] excl. local calls [] Turnover [a] [b] [c] [d] [e] [f] [g] [h] [i] Residential customers ROS Return pre-NCCN500 post-NCCN500 pre-NCCN500 post-NCCN500 Business customers Return ROS pre-NCCN500 post-NCCN500 pre-NCCN500 post-NCCN500 Base case (all calls + access) Calls only CPS all calls CPS national calls Margin of [c] over [d] CPS international calls Margin of [c] over {[d]+[f]} CPS all calls except local Margin [g] excl. local calls [] A8.12 This sensitivity tends to increase the apparent profitability of BT’s sales to the less profitable residential sector, and to reduce the profitability of sales to business customers. However in both cases, margins are significantly above the relevant benchmark, even on an FAC basis. Sensitivity: C&W time of day weights A8.13 In its analysis, Ofcom calculated the impact of NCCN 500 on BT’s profitability using actual time of day weights for BT-to-BT traffic. As a sensitivity, Ofcom replaced these with time of day weights supplied by C&W relating to traffic that it originates. Table 39: Results – assume same T.O.D profile as C&W Turnover [a] [b] [c] [d] [e] [f] [g] [h] [i] Base case (all calls + access) Calls only CPS all calls CPS national calls Margin of [c] over [d] CPS international calls Margin of [c] over {[d]+[f]} CPS all calls except local Margin [g] excl. local calls [] Turnover [a] [b] [c] [d] [e] [f] [g] [h] [i] 234 Base case (all calls + access) Calls only CPS all calls CPS national calls Margin of [c] over [d] CPS international calls Margin of [c] over {[d]+[f]} CPS all calls except local Margin [g] excl. local calls Residential customers Return ROS pre-NCCN500 post-NCCN500 pre-NCCN500 post-NCCN500 Business customers Return ROS pre-NCCN500 post-NCCN500 pre-NCCN500 post-NCCN500 [] A8.14 Applying C&W’s time of day weights reduces the ROS for residential customers somewhat, although it would remain above benchmark levels even on an FAC basis. Moreover, C&W customers are mostly businesses, suggesting that the business analysis is more relevant, which is unaffected by the change of weights. 235 Annex 9 9 FAC profitability analysis: basis of preparation A9.1 BT’s regulatory product groups P056 and P346 reflect all the non-geographic number ranges used by BT to provide the voice and data services described in the previous section, not just those NTS number ranges affected by NCCN 500. In order to ascertain the costs associated with the prices notified in NCCN 500, it was therefore necessary to attribute shared costs between number ranges. Ofcom also attributed the revenues associated with other services provided in conjunction with NTS call termination e.g. advanced voice call management or data port rentals (on the hosting side of the two-sided market) between the different number ranges reflected within P056 and P346. A9.2 BT’s regulatory product group financial statements are prepared on a current cost basis. When assessing profitability, asset values should, where possible, reflect their current cost of acquisition rather than past acquisition costs. This approach means that when assessing whether prices are high in relation to the costs of supply, up-to-date measures of the cost of supply are taken into account. FAC profitability analysis: cost basis for wholesale services A9.3 BT’s regulatory product group financial statements reflect the costs of wholesale services, e.g. geographic call termination or use of BT’s intelligent network (“IN”) incurred in servicing purchasers of NTS call termination and NTS service providers, not at underlying cost, but at either published tariff where it exists (e.g. geographic call termination) or at CCA cost including its regulated return on capital employed (e.g. intelligent network).470 This is referred to in the following discussion as the “transfer charge basis”.471 A9.4 Ofcom however wished to establish the costs actually incurred by BT on an “endto-end basis”, reflecting the full profitability that it as a vertically integrated undertaking with its own network derived from its NTS call termination/hosting operations on a fully attributed basis. Ofcom therefore sought to use BT’s actually incurred costs consistently throughout its analysis of its profitability. A9.5 As an intermediate step Ofcom also produced cost analyses based on the “transfer charge basis” which was used to conduct the hosting margin squeeze test and as a comparator in the SAC analysis set out below, where the “transfer charge basis” 470 A technology used in conjunction with traditional telephone networks to enable them to support value-added services such as NTS and personal numbering. The core functions of switching, call progress tones/announcements and dialled digit recognition continue to be supplied by the network, but call control for value added services is provided by a service layer that interacts with the network layer. Services are defined entirely in software, enabling services to be defined without the need to modify the network switches. Intelligent Network architecture and standards are defined in ITU-T Q.1200 series recommendations. 471 In BT’s regulatory financial statements, “transfer charges” are notional charges made to an activity for the use of a service from another activity within the same vertically integrated entity. The activities making or receiving the charges are generally defined in line with normal business value chains, with notional charges recorded between each activity stage. 236 better simulated the costs faced by the hypothetical competitor modelled (which in each case did not operate its own network). FAC profitability analysis: stages of analysis A9.6 The following discussion applies equally to termination of voice and data NTS calls. There were, however, some differences between BT’s voice and data NTS call termination businesses, which led Ofcom to treat certain cost and revenue items within these businesses in a specific way. Stage 1: substitution of OCP-to-BT NTS call termination/hosting revenues with Ofcom’s own calculation A9.7 The wholesale call revenue figures supplied by BT in relation to OCP-to-BT calls in response to Ofcom’s information requests (see Table 2) did not generate the relationships Ofcom would have expected, based on a weighted-average comparison with pre- and post-NCCN 500 published tariffs. Using volume data supplied by BT and multiplying it by the relevant unit prices, Ofcom calculated weighted average unit prices pre- and post-NCCN 500 to arrive at an overall percentage price increase attributable to NCCN 500. This differed significantly from the equivalent weighted average unit price rise derivable using the revenue and volume figures BT supplied to Ofcom for the purposes of its financial analysis. A9.8 It appeared to Ofcom that this resulted primarily from BT’s inability to account precisely for the impact of non-geographic number portability, both in terms of numbers ported from BT to other TCPs (i.e. where BT was the donor operator) and to BT from other TCPs (i.e. where BT was the recipient operator) and to attribute call revenues accurately between voice and data services, since BT does not capture at source whether a call is voice or data.472 A9.9 In order to reliably isolate the change in profitability attributable to NCCN 500, Ofcom therefore substituted the call revenue figures supplied by BT for OCP-to-BT revenues with its own calculations using the volume of data calls provided by BT originated on other networks. A9.10 This approach ensured that the percentage revenue increases reflected published charges. Ofcom took into account the relevant time of day profiles for voice and data calls separately, and factored in Ofcom’s estimate of the split of total volumes by voice and data for short 0845 calls terminated by BT in each period.473 A9.11 The methodology used to calculate these wholesale revenues was the same used to calculate wholesale call revenues for BT-to-BT calls (stage 2), except that it used the charges notified in NCCN 500, rather than the NTS Calculator rates for 2004/05. Stage 2: elimination of call retailing and origination activities A9.12 Ofcom eliminated retail call revenues and related costs (i.e. transfer charges for call origination and an estimate of BT’s retail costs, based on the NTS retail 472 For a further discussion of non-geographic number portability and the process for porting numbers between “donor” and “recipient” operators, see: http://www.ofcom.org.uk/telecoms/ioi/numbers/num_port_info/non_geogr_num_portab/. 473 Volumes were split 50:50 between voice and data as their time of day profile lay in between the two. 237 uplift474) from the data provided by BT, and replaced retail revenues from BT-to-BT calls with revenues calculated using the relevant average NTS Calculator rates. Stage 3: number ranges analysed A9.13 Ofcom separated out BT’s NTS call termination/hosting activities between the principal number ranges it operated under, and in the case of data, unmetered internet access (which for the purposes of the analysis were treated as a distinct NTS number range). Ofcom’s analysis of the number ranges impacted by NCCN 500 was further split between BT-to-BT calls and OCP-to-BT calls. A9.14 Ofcom first allocated costs and revenues to those number ranges where, using the cost causation principle, the costs or revenues could unambiguously be assigned to a single identified number range.475 Ofcom then apportioned all other remaining costs and revenues based on relevant call volumes (see tables 36 and 37 for the various call proportions used. A9.15 At this stage of the analysis Ofcom had constructed cost, revenue and volume information for BT’s NTS call termination/hosting activities, for each number range reflected in the relevant regulatory product groups, prepared on a CCA FAC basis with wholesale services reflected not at cost but at their transfer charge values. Stage 4: substitution of transfer charges with underlying costs A9.16 Ofcom then constructed cost, revenue and volume information for each relevant number range on the “end-to-end basis” for the wholesale services. It did this by substituting transfer charge values relating to network activities with the underlying cost and MCE information supplied by BT (see tables 36 and 37 for an indication of the amounts involved). This information was also on a CCA FAC basis. Stage 5: calculation of profitability measures A9.17 Using this information, Ofcom then calculated ROCE and ROS for the number ranges affected by NCCN 500. FAC profitability analysis: differences between voice and data A9.18 The preceding section sets out an overview of the stages of the analysis in a way that applies equally to voice and data. Specifics relating to Ofcom’s analysis of BT’s voice and data services are set out in tables 34 and 35. 474 See Charges between Communications Providers: Number Translation Services Retail Uplift charge control and Premium Rate Services bad debt surcharge, September 2005, http://www.ofcom.org.uk/consult/condocs/NTSfin/statement_nts_uplift/ 475 The cost causation principle, as set out in BT’s Regulatory Accounting Principles document, is that revenue (including appropriate transfer charges), costs (including appropriate transfer charges), assets and liabilities shall be attributed to network components, wholesale services and retail products in accordance with the activities which cause the revenues to be earned or costs to be incurred or the assets to be acquired or liabilities to be incurred. 238 Table 40: Methodological issues for FAC analysis (voice) Stage of analysis Specifics Notes Source information Sub-analysis of product group P056: BT Linkline 1. Substitution of OCP-toBT NTS call termination revenues with Ofcom’s own calculation Equivalent inputs and same approach taken for voice and data 2. Elimination of call retailing and origination activities Equivalent inputs and same approach taken for voice and data 3. Number ranges analysed Number ranges identified The principal ranges were: • 0800; • 0845 (split between BT-to-BT and OCP-to-BT); and • 0870 (split between BT-to-BT and OCP-to-BT). For the purposes of this analysis 0844 was aggregated with 0845, and 0871 with 0870, as BT did not account separately for revenues and costs associated with these number ranges. In both financial years analysed, 0844 and 0871 volumes were an immaterial proportion of total 0845 and 0870 volumes (e.g. 0.1% and 1.5% respectively in 2004/05). Cost and revenue allocations Cost and revenue apportionments The principal items allocated between number ranges were: • BT-to-BT and OCP-to-BT NTS call termination revenues for the relevant number ranges (0845 and 0870); and • public payphone supplementary call conveyance charges (PPSCCC) – allocated to Freephone calls. The principal items apportioned between number ranges were: • other revenues from NTS service providers not directly related to call volumes; • call conveyance transfer charges; • intelligent and other network 239 Stage of analysis Specifics Notes transfer charges; • remaining retail costs; • mean capital employed. The principal items apportioned between originating operator (i.e. between BT-to-BT and OCP-to-BT calls) were: • revenues from NTS service providers relating to 0845 calls; • other revenues (as above); and • revenue share payments. The % split of volumes between number ranges, and therefore the basis of the above cost apportionments, is given at table 36. 4. Substitution of transfer charges with underlying costs Wholesale services reflected on a transfer charge basis were substituted with their corresponding underlying (i.e. network component) costs provided by BT as relating to P056 as follows: • an adjustment was made to call conveyance network component costs to eliminate the origination element. The origination element was assumed to be the same proportion that origination transfer charges (see stage 2 above) was to total call conveyance transfer charges. • for other network costs elements, no further adjustments were necessary Once these cost substitutions had been made, the cost and revenue apportionments (immediate step above) were recalculated to reflect these changes. 5. Calculation of profitability measures 240 Both ROS and ROCE were calculated in the same way for voice and data. Table 41: Methodological issues for FAC analysis (data) Stage of analysis Specifics Notes Source information Sub-analysis of product group P346: Dial IP 1. Substitution of OCP-toBT NTS termination revenues with Ofcom’s bottom up calculation Equivalent inputs and same approach taken for voice and data 2. Elimination of call retailing and origination activities Equivalent inputs and same approach taken for voice and data 3. Number ranges analysed Number ranges identified Cost and revenue allocations Cost and revenue apportionments The principal ranges were: • 0800; • 0845 (split between BT-to-BT and OCP-to-BT); and • unmetered. The principal items allocated between number ranges were: • BT-to-BT and OCP-to-BT NTS call termination revenues for the relevant number ranges (i.e. for 0845 only); • revenues for unmetered ports; and • charges for unmetered wholesale services (e.g. flat rate internet access call origination (FRIACO) and product management, policy and planning (PPP) for FRIACO. The principal items apportioned between number ranges were: • other revenues from NTS service providers not directly related to call volumes (primarily metered port revenues); • IP conveyance and other network transfer charges; • remaining retail costs; and • mean capital employed. The principal items apportioned by OCP (i.e. between BT-to-BT and OCPto-BT) for 0845 were: • metered port revenues (as above); 241 Stage of analysis Specifics Notes and • revenue share payments. The % split of volumes between number ranges, and therefore the basis of the above cost attributions, is given at table 37. 4. Substitution of transfer charges with underlying costs Wholesale services reflected on a transfer charge basis were substituted with their corresponding underlying (i.e. network component) costs provided by BT as relating to P346 as follows: • call conveyance network component costs were apportioned between metered and unmetered based on their respective proportion of total metered and unmetered conveyance call transfer charges. An adjustment was then made to eliminate the origination element from metered costs. • PPP network component costs were apportioned between metered and unmetered based on their respective proportion of total metered PPP and unmetered PPP transfer charges • for other network costs elements e.g. for the IP network, no adjustments were necessary Once these cost substitutions had been made, the cost and revenue apportionments (immediate step above) were recalculated to reflect these changes. 5. Calculation of profitability measures Both ROS and ROCE were calculated in the same way for voice and data. A9.19 A major element of Ofcom’s analysis was the apportionment of costs and revenues not directly attributable to a specific number range between number ranges (i.e. ”cost and revenue apportionments” in the above tables). Ofcom apportioned these costs and revenues in direct proportion to relevant call volumes. A9.20 The relevance of these volume splits is best illustrated by example: • 242 Voice calls to NTS number ranges incur geographic call termination costs. The total cost of geographic call termination was therefore apportioned between number ranges in relation to their percentage of total call volumes. For example, in 2004/05 calls to Freephone numbers represented 31% of total voice call volumes, so picked up 31% of total geographic call termination costs. • Data calls to NTS number ranges incur some of the cost of BT’s IP network, which is used to provide both unmetered and metered internet access services. In 2004/05, 85% of total volumes, both metered and unmetered, related to unmetered internet access, so the NTS number ranges used to provide unmetered services picked up 85% of these costs. A9.21 The effective bases (i.e. volume proportions) used to apportion costs and revenues (“cost and revenue apportionments” are set out below in table 36 and table 37. Table 42: Apportionment of BT’s costs as reported in P056 (voice) All volumes relate to 2004/05 Shared costs to be apportioned £ millions** Distribution of volumes across number ranges (percentages) Freephone 0845 BT-to- 0845 OCP- 0870 BT-to- 0870 OCPBT to-BT BT to-BT Apportionment categories All number ranges 0845 number range only 0870 number range only Total [] ** 2004/05 profit & loss end-to-end costs given here for illustrative purposes; MCE and certain revenues also apportioned Table 43: Apportionment of BT’s costs as reported in P346 (data) All volumes relate to 2004/05 Distribution of volumes across number ranges (percentages) Metered Unmetered Total Shared costs to be apportioned £millions** Apportionment categories Metered & Unmetered Metered only 0845 0845 BT-to-BT OCP-to-BT Freephone & other Sub-Total [] BT-to-BT v OCP-to-BT ** 2004/05 profit & loss end-to-end costs given here for illustrative purposes; MCE and certain revenues also apportioned 243