IN THE HIGH COURT OF REGISTRY CIV 2000-485-673

BETWEEN THE COMMERCE COMMISSION Plaintiff

AND TELECOM CORPORATION OF NEW ZEALAND LIMITED First Defendant

AND TELECOM NEW ZEALAND LIMITED Second Defendant

Hearing: 20-23, 27-30 August, 3-7, 10-11, 18-19 and 24-26 September 2007

Court: Rodney Hansen J Mr MC Copeland

Counsel: JA Farmer QC, BD Gray QC, S Glass and JS McHerron for Plaintiff D Shavin QC, JE Hodder, PR Jagose, S Dalzell and H Northover for Defendants

Judgment: 18 April 2008

JUDGMENT OF RODNEY HANSEN J AND MR MC COPELAND

This judgment was delivered by me on 18 April 2008 at [10.00 a.m ], pursuant to Rule 540(4) of the High Court Rules.

Registrar/Deputy Registrar

Date: ………………………….

Solicitors: Commerce Commission, P O Box 2351, Wellington for Plaintiff Chapman Tripp, P O Box 993, Wellington for Defendants

THE COMMERCE COMMISSION V TELECOM CORPORATION OF NEW ZEALAND LIMITED And Anor HC AK CIV 2000-485-673 [18 April 2008] TABLE OF CONTENTS

Introduction [1]

Confidentiality [4]

Issues [5]

Markets

Background [8]

Market definition [17]

Fixed-line retail phone service to residential customers [19]

Wholesale network access [22]

Retail ISP services for residential customers [26]

Wholesale ISP services [27]

Dominance [28]

Fixed line retail phone services to residential customers [30]

Wholesale network terminating access [46]

Use of dominance

Legal test [55]

Development of 0867 service [56]

Counterfactual analysis [77]

Purpose [91]

Commission’s case [93]

Stated purpose [97]

Inferred purpose [106]

Result [114] Introduction

[1] The advent and dramatic expansion of the internet during the 1990s caused a revolution in telecommunications worldwide. In New Zealand it provided a further and unexpected challenge to a telecommunications industry that was still adapting to the competitive environment created by the economic and regulatory reforms of the late 1980s. Burgeoning internet traffic greatly added to the pressures on the telecommunications infrastructure and radically changed the commercial environment. This led to the introduction by the defendants (Telecom) in 1999 of its 0867 service which is the genesis of this litigation.

[2] The rapid and unforeseen growth in internet traffic had two consequences which the 0867 service was intended to address. The first was a massively increased loading on Telecom’s network. The second was an adverse and growing imbalance in fees payable by Telecom to Clear Communications Limited (Clear) under an agreement made in 1996 which had not anticipated the impact the internet would have on the industry.

[3] While there was and is no issue that Telecom was entitled to take steps to address these concerns, Clear and others claimed that in doing so by means of the 0867 service, Telecom acted in breach of s 36 of the Commerce Act 1985 (the Act). They complained to the Commerce Commission that the introduction of the 0867 service involved the use by Telecom of a dominant position for the purpose of preventing Clear and others from engaging in competitive conduct in a market. Clear and other complainants ultimately reached commercial settlements with Telecom. The Commission continued this proceeding, seeking a declaration that Telecom was in breach of the Act and a pecuniary penalty.

Confidentiality

[4] Some of the information received by the Court is subject to confidentiality orders. Any confidential information referred to in this judgment will be enclosed in square brackets. The confidential information will be included in copies of the judgment distributed to the parties but omitted from any copies accessible to the public.

Issues

[5] Section 36(1) of the Act (prior to its amendment in 2001) relevantly provided:

Use of dominant position in a market

(1) No person who has a dominant position in a market shall use that position for the purpose of -

(a) Restricting the entry of any person into that or any other market; or

(b) Preventing or deterring any person from engaging in competitive conduct in that or any other market; or

(c) Eliminating any person from that or any other market.

[6] In order to establish a breach of s 36, three elements must be present:

a) A person who has a dominant position in the market;

b) Who has used that dominant position;

c) For the purpose of the matters referred to in paras (a), (b) and (c) of subs (1)

Telecom Corp of NZ Limited v Clear Communications Limited [1995] 1 NZLR 385 at 402 (PC).

[7] Proof of these elements raises four primary issues for consideration. They are:

a) To identify the relevant markets.

b) To determine whether Telecom was dominant in any market. c) If Telecom was dominant, to decide whether in introducing the 0867 service it used that dominance.

d) If there was a use of dominance, to decide whether it was for an anti- competitive purpose.

Markets

Background

[8] Telecom was established in 1987 as a state owned enterprise to take over the telecommunications network services previously operated by the Post Office. From 1 April 1989 all telecommunication markets had been opened to competition. The process of privatising Telecom was completed by its sale in September 1990 to a consortium for $4.25b. Clear soon emerged as Telecom’s first competitor, commencing business in May 1991.

[9] A feature of the telecommunications industry in New Zealand is Telecom’s ownership of the public-switched telecommunications network (PSTN) and associated infrastructure. The PSTN comprises a network of copper wires that connects customers’ premises to local exchanges (the local loop or last mile) and the core network or backbone that connects the exchanges to each other.

[10] As a precondition of the sale of Telecom, the government took steps to ensure that private ownership of the PSTN would not adversely affect the price of telephone services to residential telephone users. It did so by means of the Kiwi Share Obligation (KSO), which attached to a share in Telecom, known as the Kiwi Share, held by the Minister of Finance on behalf of the Crown. In 1999 the rights of the Kiwi shareholder were set out in the First Schedule to Telecom’s constitution.1 Those rights could not be varied without the consent of the Minister. They were legally enforceable against Telecom at the suit of the Minister. Among other things, the KSO provided:

1 The KSO ceased to exist in 2001 when, on the passing of the Telecommunications Act 2001, its key provisions were replaced by the telecommunications service obligation (TSO). • A local free-calling option would be maintained for all residential customers.2

• Telecom would charge no more than the standard residential rental for ordinary residential telephone services and, from 1 November 1989, the standard residential rental would not be increased in real terms provided that overall profitability of subsidiary regional operating companies was not unreasonably impaired.

• The line rental for residential users in rural areas would be no higher than the standard residential rental (as defined) and Telecom would continue to make ordinary residential telephone services as widely available as at 11 September 1990.

[11] Competition in the nascent telecommunications markets began when Clear started offering a toll service. One of its shareholders, NZ Rail Limited, had existing fibreoptic cables running along its train tracks which enabled it to commence business as a supplier of toll services. It was able to offer customers toll services by bypassing Telecom’s core network (known as toll bypass services). Telecom agreed to carry calls from its customers to the Clear network (known as originating access) and from Clear’s network to the customer receiving the call (terminating access). In 1991 Telecom and Clear entered into an interconnection agreement (ICA) under which Telecom agreed to provide Clear with originating and terminating access.

[12] In 1996 Telecom and Clear entered into a further ICA which enabled Clear to provide local interconnections as well as toll bypass interconnections. The agreement followed the 1994 decision of the Privy Council in Telecom v Clear which finally determined the basis on which Telecom could charge Clear for local interconnection, that is for connecting calls that originated on one network and terminated on the other network. The ICA provided for payment for terminating (and originating) access. The charges were, however, asymmetric. Clear was required to pay Telecom a higher rate per minute than Telecom was required to pay Clear for terminating access on Clear’s network.

2 By free-calling option is meant that the cost of local calls would be covered by a fixed monthly charge. [13] The charges for terminating access were on a “calling party pays” basis, whereby the network originating the call is paid by the customer making the call, albeit for Telecom by way of the flat monthly charge permitted under the KSO for local calls. The network terminating the call receives no income from the customer making the call. The originating network then passes some of the income received from the calling customer to the terminating network in payment for the termination service provided. The “calling party pays” method of payment is to be contrasted with a “bill and keep” arrangement whereby the originating network keeps all of the income received from the calling customer.

[14] Computer owners accessed (and continue to access) the internet through telephone calls to internet service providers (ISPs) who facilitate access to the global network of computers. As one witness described it, ISPs act like a local post office. In the main centres Clear established its own access network, connecting business customers in the central business districts. It attracted a number of ISPs to its network. A feature of ISPs is that they receive a large volume of telephone calls but make virtually none. They are what is termed a “call sink”. Furthermore, unlike voice calls, calls to an ISP are of much longer duration than voice calls. Because most calls to ISPs came from the Telecom network, termination fees became a lucrative source of revenue for Clear. On the other hand, because Clear had very few residential customers, Telecom’s revenue for providing terminating access to calls originating on Clear’s network was relatively low. Telecom found itself paying out much more in termination fees than it received. One of the reasons Telecom introduced the 0867 service was to counter this development.

[15] The 0867 service introduced a new 0867 number for internet calls. Telecom maintained calls to the 0867 number were not covered by existing ICAs. Carriers were required to enter into a new agreement with Telecom to obtain the 0867 number range and would not be paid termination fees for calls to those numbers. Residential customers could continue to obtain unlimited access to internet services by using the 0867 service but if they chose not to, they would be charged two cents per minute after ten hours of use per month. The two cent per minute charge did not, however, apply to calls to the 0873 number used by Telecom’s ISP, Xtra. [16] Residential customers who accessed internet services for more than ten hours per month were accordingly incentivised to use the 0867 scheme or the services of Xtra. If carriers such as Clear did not enter into a new agreement with Telecom, they risked losing the ISPs on their network.

Market definition

[17] Markets must be defined by reference to s 3(1A) of the Act which provides:

Every reference in this Act, except the reference in section 36A[(2)](b) and (c) of this Act, to the term “ market“ is a reference to a market in New Zealand for goods or services as well as other goods or services that, as a matter of fact and commercial common sense, are substitutable for them.

[18] The purpose of market definition is to identify the area or areas of close competition and to identify the constraints which operate on participants. It is the first step in determining whether Telecom is dominant in any relevant market and it is necessary to deciding whether any use of dominance was for an anti-competitive purpose. There is general agreement on the dimensions and characteristics of the relevant markets in which Telecom may have been dominant and which would be affected by any use of that dominance.

Fixed-line retail phone service to residential customers

[19] There is a market for fixed-line retail access to residential customers which covers the provision of services to residential customers to make local voice or internet calls. At least in 1999, it excluded the provision of toll call services and mobile telephone calls. Services to business customers are also excluded. They were not subject to the KSO. The 0867 package applied only to residential customers.

[20] The only difference between expert witnesses related to the geographical dimension of the market. Professor Jerry Hausman, one of Telecom’s experts, suggested there should be separate regional markets. He argued that residential services were not substitutable between separate geographical areas. He also claimed that, because the costs of providing telephone access to residential customers varies widely between geographic areas (the cost of access in rural areas, for example, is higher than in urban areas), whereas the KSO imposes a uniform price across the country, price discrimination markets exist in which different margins occur for different customers. On orthodox principles, such markets would be analysed separately.

[21] While acknowledging the force of Professor Hausman’s arguments, we propose to adopt the national market. We do not think there is any advantage to be gained by analysing a whole series of separate geographic markets. The objective of exposing the play of market forces upon industry participants will not be compromised by adopting a national market. In particular, the analysis of dominance will take full account of the special characteristics flowing from the KSO.

Wholesale network access

[22] As mentioned earlier, both Telecom and Clear provided (and continue to provide) access to each other’s networks for toll bypass and local interconnection services. That is necessitated by what is known as the requirement of ubiquity. Any supplier of telecommunication services must provide connections with all, or almost all, other telephone users. We have explained that Clear was able to offer toll bypass services by purchasing both originating and terminating access to Telecom’s network and using its own core network to transmit the call from an exchange in one locality to an exchange in another locality. For the purpose of offering a local service, a firm such as Clear required terminating access only. Having established its own network linking to business customers in Auckland, Wellington and Christchurch, it needed only terminating access to residences and businesses connected to Telecom’s network. That was secured by the 1996 ICA..

[23] Arguably, there is a market for originating and terminating wholesale network access (for firms offering toll bypass services) and a separate market for terminating wholesale network access (for firms selling local access services). Telecom priced originating and terminating wholesale services at different rates from terminating wholesale services; it charged Clear more for terminating services than for originating and terminating services. The 0867 package potentially affected termination fees for termination wholesale network access but not fees for originating and terminating access for providers of toll bypass services.

[24] The Commission’s expert, Dr Gustavo Bamberger, preferred a combined market, as did Professor Lewis Evans who also gave evidence for Telecom. Professor Hausman thought the relevant market should be confined to terminating access. We agree with him that the narrower definition is better suited to a consideration of dominance. At the wholesale level, the 0867 service related only to the termination of local calls. That is the relevant field of rivalry. We note also Dr Bamberger’s acknowledgement that nothing in his evaluation of dominance turned on this aspect of market definition.

[25] In closing submissions, Mr Shavin QC hypothesised the existence of a series of small markets each corresponding to the networks to which terminating access is sought, rather than a single wholesale network access market. If the wholesale access market were defined in this way, each carrier would be dominant in the market for terminating access on its own network. We do not find this a helpful means of assessing the potential dominance of Telecom in relation to terminating access and the effect of the 0867 service. It forecloses any exploration of the constraints which operated on Telecom; the market definition itself would become conclusive of dominance rather than the tool for analysis it is intended to be. We consider the appropriate definition for the purpose of assessing Telecom’s dominance is the market for wholesale network terminating access.

Retail ISP services for residential customers

[26] All experts agreed that there was a national market for the provision of ISP services to residential customers. This is the market in which ISPs such as Xtra, Ihug, CallPlus and others supplied internet services to members of the public. Wholesale ISP services

[27] Again, all experts agreed that there was a market for the supply of services by network companies, such as Telecom, Clear, and Saturn, to ISPs for the purpose of enabling ISPs to receive internet calls from residential customers to connect them with the worldwide web, send and receive emails and receive other services provided by ISPs.

Dominance

[28] Before its repeal in 2001, s 3(8) of the Act defined “dominance”. It provided:

(8) For the purposes of sections 36 and 36A of this Act, a dominant position in a market is one in which a person as a supplier or an acquirer of goods or services either alone or together with any interconnected body corporate is in a position to exercise a dominant influence over the production, acquisition, supply, or price of goods or services in that market and for the purposes of determining whether a person is in a position to exercise a dominant influence over the production, acquisition, supply, or price of goods or services in a market regard shall be had to—

(a) The share of the market, the technical knowledge, the access to materials or capital of that person or that person together with any interconnected body corporate:

(b) The extent to which that person is constrained by the conduct of competitors or potential competitors in that market:

(c) The extent to which that person is constrained by the conduct of suppliers or acquirers of goods or services in that market.

[29] There was no dominance analysis of the markets for retail ISP services for residential customers and wholesale ISP services. We assume they were contestable. They have relevance to the purpose and effect of the introduction of the 0867 service but not to the use of dominance itself. For that purpose, the relevant markets are those for fixed line retail phone services to residential customers and wholesale network access. Fixed line retail phone services to residential customers

[30] The statutory test of dominance requires us to decide whether Telecom was in a position to exercise a dominant influence over the production, acquisition, supply, or price of goods or services in the market. By reference to conventional criteria (including those specifically referred to in s 3(8)), a finding of dominance would be irresistible.

[31] In 1999 Telecom accounted for over 99% of residential lines in New Zealand. Its PSTN was (and still is) connected to almost every residential location. Telecom had substantial “sunk costs” because of its status as the legacy carrier. Entry into the market for fixed line retail phone service to residential customers was difficult because economies of scale and density are important determinants of the cost of supplying local loops. New competitors such as Clear built local networks primarily in the CBDs of the largest cities in New Zealand and generally did not attempt to enter the retail market for residential customers. An exception was Saturn which began offering a local residential phone service in selected residential areas in the Wellington region in 1998. However, it did so by offering telephone services as an adjunct to its pay television service.

[32] The argument against Telecom being dominant focuses on the effect of the KSO. Mr Shavin submitted that the conventional indicia of dominance – market share and entry barriers – are answered and explained by the KSO. He argued that both barriers to entry and Telecom’s market share are the result of the requirement under the KSO that Telecom provide a free residential local calling option, essentially across the entire country. The KSO was said to have determined the essential structure of the market and to have acted as a constraint on any dominance its features might otherwise have conferred on Telecom.

[33] The argument continues that the KSO requires Telecom to maintain the provision of residential local access but prevents it from raising the price of the service. It is accordingly a constraint on Telecom either raising price or reducing output. By reference to the classical formulation of market power in Re Queensland Co-op Milling Association (1976) 8 ALR 481 at 515, the KSO prevents Telecom from “charging more” or “giving less”, and “keeps that power in check”. In terms of s 3(8)(c), it is a constraint applied by the Crown, as representative of acquirers of residential local access services.

[34] The experts differed on the extent to which the KSO in fact constrained Telecom’s ability to price significantly above the competitive price (to adopt Professor Hausman’s test of the existence of market power). Dr Bamberger pointed out that, in approving the 0867 package, the government accepted Telecom’s position that the internet dialup charge (IDC) was not constrained at all by the KSO. He also argued that, as the KSO set an average price across all New Zealand, it may be expected that in some local telephone markets, price would exceed cost. As a result, Dr Bamberger said the existence of the KSO cannot inform an analysis of whether competition constrained Telecom’s ability to impose the IDC, or the level at which the IDC would be set. He said the fact that Telecom loses money providing services in some areas because of the KSO is a sustainable outcome for incumbent telecommunications carriers, common in many countries, because such carriers are allowed to charge higher-than-competitive rates in other areas. He said that typically the same national price for the service implies a cross-subsidy from urban areas to rural areas because unit costs tend to be substantially lower in urban areas. Dr Bamberger contended that the ability to subsidise rural prices by charging higher- than-competitive rates in urban areas could not be sustained in the presence of competition. The existence of the KSO therefore demonstrates that Telecom’s prices for fixed line residential services were not constrained to competitive levels by rivals.

[35] Telecom’s experts argued the KSO prevented Telecom from earning a supra- normal return on invested capital or a return above the risk-adjusted rate of return. Professor Hausman said the evidence indicates that Telecom’s own price elasticity3 for residential access lines is well below the (1.0) level to be expected of a firm that is not subject to price constraints. He relied on a report by officers of the Commission into complaints that Telecom breached ss 27 and 36 of the Act in lowering its prices in response to the entry of Saturn into the market in the Wellington region – Termination Report: Telecom’s Pricing of Fixed Telephone Services in Lower Hutt: 30 July 1998. In concluding that no breach had occurred, the Commission said (at para 58) that as a direct consequence of the KSO, it is likely that Telecom did not recover its long run average incremental cost of providing a bare access service across New Zealand. For Telecom it was submitted that the Commission cannot resile from its own analysis that the constraint provided by the KSO is such that the capped rental does not cover the costs of residential access. Telecom also relied on a passage from the High Court judgment in Clear v Telecom4 which recorded Clear’s acceptance that Telecom’s revenue for non-business rentals and local calls did not cover its costs.

[36] We are unable to finally determine whether, as a result of the KSO, Telecom’s price for residential access was below its actual costs. While the Saturn report is a substantial and carefully researched document, it relied for its key findings on information provided by Telecom which was not in evidence or tested before us. The reports of later inquiries are equivocal on the issue.

[37] A ministerial inquiry into regulatory arrangements for the New Zealand telecommunications sector, chaired by Mr Hugh Fletcher (hence known as the Fletcher Inquiry), found in its report - Ministerial Inquiry into Telecommunications: Final Report, 27 September 2000 – (at 9.1.3) that Telecom’s estimate of $100m for the losses it incurred in performing its Kiwi Share obligations was an upper estimate. Against this, the Inquiry found:

... there are other legacy businesses of Telecom, such as its directory business, which are delivering above cost of capital returns.

The Kiwi Share standard residential line rental price cap is high by international standards and it is not a diminishing price cap in real terms. It was also set before much of Telecom’s significant productivity improvement and before the recent global technological advances and associated dramatic decline in the costs of an efficient operator.

[38] The Fletcher Inquiry recommended that the delivery of cost-efficient, timely and innovative telecommunications services would be enhanced by regulation. It

3 The percentage change in demand in response to a percentage change in price. 4 Clear Communications Limited v Telecom Corporation of New Zealand Limited HC WN CP590/91 22 December 1992 recommended legislation, later enacted as the Telecommunications Act 2001 which, as earlier mentioned, replaced the KSO with the TSO. Under the TSO Telecom is required to continue to provide an 0867 service.

[39] Under Part 3 of the Telecommunications Act 2001, the Commission is annually required to determine the cost incurred for the supply of services by Telecom under the TSO. A key component of its calculation is to determine the cost of providing TSO services to what are called commercially non-viable customers (CNVCs). The Commission’s report for the period 20 December 2001 – 30 June 2002 dated 17 December 2003, found (at para 470) that the numbers of CNVCs constituted 5% of the 1.3m residential customer lines modelled. It calculated that Telecom was losing $65m in providing services to those customers.

[40] Mr Shavin acknowledged that it would not be realistic to finally determine the extent to which Telecom was constrained by the KSO, but argued that, in the face of the Commission’s own findings in the Saturn report and Clear’s concession in its litigation with Telecom, there is an evidential onus on the Commission to show that the KSO did not operate to constrain prices.

[41] There can be no doubt that the KSO did constrain Telecom’s prices and, consequently, its levels of profitability in the residential access market. It also operated in other ways as a constraint on Telecom’s freedom to act; it required, for example, government approval of the response to marketplace pressures that became the 0867 service.

[42] However, the relevant question is much broader. It is whether the constraints which flowed from the KSO meant that in 1999 Telecom was not in a position to exercise a dominant influence over the production, acquisition, supply or price of goods or services in the market. That, as Cooke P said in Telecom v Commerce Commission [1992] 3 NZLR 429 at 434 (CA), requires a consideration of “the extent to which a person is free from the practical constraints of competition”.

[43] In our view, the constraints offered by the KSO were not equivalent to the practical constraints of competition. For example, cross-subsidies existed between urban and rural customers and between heavy users and light users which would not have been sustainable in a competitive market. Without the KSO, it may be that the barriers to entry and the high market share, which are suggestive of dominance, would not have existed. But it does not follow that the KSO prevented Telecom from exercising a dominant influence over the production, acquisition, supply or price of goods or services in the market. For, as we will come to consider in more detail when addressing the possible use of dominance, it could not be suggested – and was not suggested – that the KSO or some other means of control on Telecom would create a competitive market. All experts agreed that, for the purpose of determining what a non-dominant firm might do, there would need to be postulated a number of competing firms and the absence of barriers to entry.

[44] It cannot be overlooked, either, that the KSO operated as constraint only on voice calls. The government, while not conceding that the 0867 package did not involve a breach of a KSO, did not object to its introduction provided the interests of residential customers were protected. Telecom was at liberty to charge customers for internet dialup calls, where previously the service was free. Even if Telecom did not have market power in relation to voice calls, it was not constrained in its ability to increase prices for internet dialup calls.

[45] We conclude that in 1999 Telecom was dominant in the market for fixed line retail services to residential customers.

Wholesale network terminating access

[46] The Commission’s position, as articulated by Dr Bamberger, is that Telecom was dominant in the market for wholesale network access. His arguments apply equally to a market confined to terminating access. He relied on direct economic evidence that Telecom had market power as shown by the terms of the 1996 ICA with Clear which required Clear to pay Telecom substantially more for terminating services than Telecom paid Clear. On the assumption that Clear’s price exceeded its costs and that Clear’s costs were similar to or higher than Telecom’s, it could be inferred that Telecom’s prices were substantially higher than its costs.5

[47] Dr Bamberger also pointed to Telecom’s ability to price discriminate by charging Clear a higher price for terminating access than for originating and terminating access, although the costs of the termination service alone could not be higher. Even assuming the price of originating and terminating access to be at competitive levels, Telecom was pricing termination services at above competitive levels, unconstrained by competition or the threat of entry.

[48] Dr Bamberger also relied on the indirect economic evidence that, because of its ownership of the PSTN, any firm offering telecommunication services in New Zealand would have no reasonable alternative to acquiring wholesale network access services from Telecom.

[49] Professor Evans, while disagreeing with Dr Bamberger’s approach, did not challenge his conclusion of dominance in the wholesale network access market. Professor Hausman acknowledged that Telecom may have substantial market power but pointed out that so had Clear as shown by the supra-normal profits it had earned for termination on its network, permitting it to pay significant subsidies to ISPs. Mr Shavin submitted that, as only one person can have a dominant influence over any relevant aspect of the market at one time (see Telecom v Commerce Commission at 434 and Limited v Todd Energy Limited [2007] NZCA 304 at [171]-[175]), Telecom’s market power did not equate to dominance for the purposes of s 3(8).

[50] In 1996 when the ICA with Clear was concluded, Telecom was clearly dominant in the market for terminating (and originating) access for the reasons given by Dr Bamberger. Telecom acknowledged that termination charges were fixed above cost. The question is whether it had ceased to be dominant as a result of the intervening change in market conditions and, in particular, the supra-normal profits

5 Both the Commission and Telecom accepted that prices for terminating services in the 1996 ICA were above the costs of both Telecom and Clear. The Fletcher Inquiry found that the terminating charges were not efficient (cost-based) - the Fletcher Report at para 7.2. Clear was generating by virtue of the terminating access provided to ISPs on its network.

[51] In our view, the changes in the market since 1996 had been such as to negate Telecom’s dominance by the time the 0867 service was introduced. Clear had successfully established its own networks in the CBDs of the three largest cities. It was competing on equal terms with Telecom for business customers in those areas. Those customers relevantly included the ISPs which, by virtue of the asymmetric above-cost charges for terminating access in the 1996 ICA, became a lucrative source of terminating access revenue for Clear.

[52] These changes in the market were not transient or short-run phenomena. Clear’s position was not simply the product of the asymmetric terms in the 1996 ICA which could be expected to come to an end when the ICA fell to be renegotiated in December 2000. Clear’s established network and a continuing growth in internet traffic meant that when the ICA came to be renegotiated (whatever the fate of 0867) Telecom would be required to offer cost-based terminating access. That is what in fact occurred. As we will later explain, Telecom and Clear reached a commercial settlement and a new ICA was agreed under a cost-based bill and keep arrangement.

[53] The Fletcher Inquiry concluded in 2000 that, because of the ubiquitous nature of Telecom’s network, and the need of other providers to interconnect to that network, Telecom had market power which had allowed it to charge inefficiently high interconnection prices over the previous decade. Clear’s entry into the business sector of the market (associated with the rapid growth of internet traffic) changed the conditions which had made that possible. Notwithstanding Clear’s modest market share and its comparatively limited network, it had become an effective competitor in the market for terminating access whose conduct demonstrably operated as a significant constraint on Telecom. We are satisfied that in 1999 Telecom was no longer dominant in the market for wholesale network terminating access.

[54] We conclude that Telecom was dominant only in the market for fixed line retail phone services to residential customers. Use of dominance

Legal test

[55] A dominant firm does not use its dominance for one of the proscribed purposes in s 36 if it acts as a non-dominant firm otherwise in the same position would have acted in a competitive market – Carter Holt Harvey v Commerce Commission [2006] 1 NZLR 145 at [52] (PC). This is the basis of the counterfactual test. In Carter Holt at [60](a), the Privy Council reaffirmed that it is both legitimate and necessary to apply the test in order to determine whether a firm has used its position of dominance. In the present context, the counterfactual test involves an enquiry into the way in which a firm in Telecom’s position would have acted in a competitive market. If it would have been able to introduce the 0867 service in a competitive market, it cannot be said to have used its dominant position.

Development of 0867 service

[56] Before considering the way in which Telecom could have acted in a competitive market, it is necessary to say something more about the circumstances in which the 0867 package was developed and implemented.

[57] The ICAs, including that made between Telecom and Clear in 1996, were negotiated on the basis that local calls passing between different carriers’ networks would largely be in balance. It was assumed that local calls would be voice calls which have a number of predictable characteristics:

• The majority are made by residential customers.

• They tend to be of short duration – the average call is around three minutes.

• Calling is two-way, that is, people generally make the same number of voice calls as they receive. • The average number of voice calls is stable, particularly in New Zealand where residential customers enjoy free local calls.

• The peak demand for voice calls occur around 7.00 p.m. on weekdays although the exact time may vary.

[58] These features determined that termination payments in the ICA would be calculated on a price per minute basis and that the local voice traffic would be broadly balanced; Telecom’s customers would call other carriers’ customers about as often and for as long as other carriers’ customers called Telecom customers. That has in fact proved to be the case.

[59] Internet calls differ from voice calls in a number of respects:

• They are longer in duration – in 1999 the average internet call was 22 minutes compared to 3 minutes for voice calls.

• Internet calling can be automated. A computer can be programmed to redial an ISPs number repeatedly until it establishes a connection.

• Internet calling is one-way, that is, customers make calls to ISPs but not the other way round.

• Peak demand occurs around 9.00 p.m. on a week day, later than the peak for voice calling.

[60] The growth in internet traffic was dramatic and underestimated. In the space of a few years, internet usage went from nothing to become the dominant use of the PSTN. An internal Telecom paper dated 13 March 1998 notes the number of host addresses going from near zero in 1992 to around 15,000 in 1995, rising to 53,000 in 1996, 84,000 in 1997 and 169,000 in 1998. There was an increase in the number of residential customers using the internet, the number of calls each customer was making and the average length of those calls. By March 1999, 31% of residential local traffic was internet traffic, almost double the volume of a year before. In considering the 0867 business case, Telecom forecast continuing exponential growth in internet traffic from a thousand million minutes in the 1999/2000 financial year to six thousand million minutes in the 2001/2002 financial year.

[61] Some of the growth had been stimulated by flat rate pricing schemes, known as “all you can eat”, initially introduced by Ihug and later adopted by Xtra and Clearnet. The growth in internet traffic was matched by the number of ISPs and reflected in fierce competition between them. Between 1995 and 2000 the number of ISPs grew from 15 to over 100.

[62] The combination of above-cost terminating payments under the ICAs and free residential calls to ISPs created arbitrage opportunities which were exploited by competing carriers and ISPs. These conditions led to what were described in a key Telecom memorandum dated 18 May 1999, which laid out the business case for the proposed 0867 service, as “perverse incentives” for generating additional dialup internet traffic. In setting out the background to the proposal, the memorandum said:

The present local interconnection agreement was established on the basis of long-standing voice traffic characteristics, in particular, the approximate balance of traffic originating and terminating between networks. The agreement provides for a per minute payment to be made from the carrier originating a call to the carrier terminating a call. The advent of internet service providers as large “sinks” of traffic (i.e. the traffic is not balanced; the much higher proportion terminates at the ISP) has provided local interconnect carriers with a readily identifiable customer base that provides the carriers with large interconnect revenues. The interconnect payments are not offset by revenue for Telecom as there is no balancing traffic from the carrier, and there is no revenue from the end user because their local calls to dial-up the ISPs are invariably free residential calls. The interconnect revenue thus earned by the terminating carrier is substantial and can be shared with the ISPs as an incentive to join the carrier’s network.

The concern for Telecom is that this revenue can then be further shared with the ISPs’ customers, to encourage them to remain on-line. As we show in the analysis that follows, the presence of these perverse incentives creates a likely scenario of much higher dial-up internet traffic than forecast, either leading to unacceptable PSTN congestion for all services (including basic voice services) or requiring substantial investment to increase the capacity of the PSTN and its associated systems.

This scenario is not at all unlikely with ISPs in the UK now offering free internet access off the back of termination payments from carriers.

(Emphasis added) [63] As the memorandum notes, the rapid growth in internet traffic and the features of internet calling patterns caused congestion on the PSTN. Internet calls were of much longer duration than voice calls. As ISPs generally had links to a particular local exchange, every call going to an ISP would tie up one line in that exchange leading to a phased shutdown if capacity overload continued. The problems of capacity overload were exacerbated by a phenomenon known as “attack dialling” by which, if a call is unsuccessful, repeated calls are made until a connection is made.

[64] Telecom sought to address the increased burden on the PSTN network by several measures. It developed a system called the Intelligent Network (IN). It was not a separate network but a system of call management which provided enhanced management of the existing network. Bruce Parkes, who had primary responsibility for developing the 0867 service, described it as a brain for Telecom’s PSTN. It could help manage internet traffic but only if the customer dialled a special number such as 0800. It was seen as having the capacity to reduce problems associated with limited switching capacity in the PSTN, but only limited potential to solve the long- hold problem and other issues arising from increased internet usage. In order to address these issues, Telecom needed an alternative network that could be used to send internet data traffic to the ISPs. The IPNet was developed to meet that need.

[65] The IPNet sought to remove internet traffic from the network by converting internet calls to data format at the originating exchange (instead of at the ISP’s premises) and sending the calls over a data network directly to the ISP. By this means, internet calls were taken off the PSTN before they reached the terminating exchange and ISP’s access links.

[66] IPNet did not succeed as anticipated. It did not appeal to major ISPs (other than Xtra) who preferred to control their network infrastructure and found cheaper alternatives. It did not achieve its goal of becoming the predominant means of customers accessing the internet.

[67] Telecom was hesitant about large scale investment in an expanded PSTN because of the risk of “stranded assets”. It was foreseen that the emerging broadband technology would in time see internet traffic migrating to non-PSTN platforms. The expectation that the new technology would take some years to implement has proved to be the case. It is only in recent years that broadband customers have grown to exceed internet dialup internet customers.

[68] Mr Parkes explained in detail how the 0867 proposal evolved. He said Telecom had to resolve the ICA arbitrage problem. The incentive to generate traffic for traffic’s sake was untenable because it was causing a level of unbilled growth that would have rapidly caused parts of the network to become unusable. He explained how the genesis of the 0867 service was a proposal to combine a charge for using the PSTN for internet calls while offering a free alternative utilising the IPNet technology. However, as the IPNet was a unique Telecom service, a free calling option using the IPNet would only be available to ISPs that were based on the Telecom network.

[69] As the volume of internet traffic increased, Telecom perceived that a permanent solution would require an element of charging for residential calls to the internet on the PSTN. The free call IPNet project was put on hold while there was focus on a separate project to charge for internet calls which became known as PSTN Internet Charging (PIC). As the PIC project developed, it was recognised that if the solution involved making IPNet calls free while charging for PSTN calls, there would be a perception that ISPs were being forced to use IPNet or another carrier’s comparable service. The solution was to utilise the IN instead of the IPNet by allowing free residential calls to ISPs using IN numbers. This would permit free calls to:

• Telecom’s IPNet service.

• A competing carrier’s IPNet equivalent.

• A non-IPNet IN number terminating on Telecom lines provided to the ISP or on another carrier’s network. There would be a charge only for calls to a PSTN number terminating on the lines of Telecom or another carrier’s network. A further advantage of the proposal, as seen by Telecom, was that the use of IN numbers would mean that the free residential internet calls would fall outside the existing ICAs.

[70] The 0867 proposal, essentially in its final form, but before the final digit 7 was chosen, was set out in an internal Telecom memorandum dated 2 March 1999. The problem was identified as follows:

Telecom faces spiralling growth from residential local calls to the internet. Internet calls are now over 20% of all residential call traffic and growing strongly. It is estimated in 2-3 years that this figure will be in excess of 40%.

The growth is driven by decreasing ISP costs which allow for reasonably priced “all you can eat” packages and free local calls.

It is clear that the lack of economic incentives on customers is driving inefficient overuse of the network and if not addressed, will lead to either huge investment in the network and/or congestion which will impact on voice traffic which the PSTN is designed for.

This 086X proposal is designed to address this.

[71] The business case for the proposed 0867 service, by then a combination of the internet dialup access (IDA) and PIC workstreams known as Project Morecambe, was set out in a memorandum to Theresa Gattung, the Chief Executive Officer of Telecom, dated 18 May 1999. It sought approval for expenditure of $2.556m to implement the project. The financial base for the business case was said to be entirely cost-avoidance: it would deliver no new revenue but would prevent large spending on Telecom’s infrastructure over the succeeding four years. The rationale for the project was set out as follows:

The rationale for the expansion of the scope of PIC to include IDA and combine under Project Morecambe is that IDA and PIC complement each other:

1. Both developments have been initiated in order to manage costs resulting from the growth of Internet dial up traffic on the PSTN without affecting the natural growth trend. PIC presents the residential market with the local call cost for dial-up Internet access. IDA provides Telecom with a cost effective framework with which to manage dial-up Internet traffic. 2. PIC applies to local calls to an ISP in order to access the Internet, but does not apply to calls on IDA or IPNet. Thus, IDA and IPNet will be preferred by residential customers to local call dial-up. Furthermore, IDA will not affect residential customers’ normal Internet access behaviour.

3. Financially, IDA is expected to reduce the revenue resulting from PIC.

The memorandum went on to say:

Abnormal Dial-up Growth

As well as managing the current growth of Internet traffic on the PSTN, a set of incentives were identified between other access providers and ISP’s that could accelerate this growth of dial-up Internet traffic at an uncontrolled rate. The set of incentives results in ISPs subsidising residential customers’ use of Internet access, creating abnormal growth in dial-up Internet traffic. Currently, Telecom has no mechanism to manage the consequential growth in access lines and call volume other than Telecom’s physical capacity to support this growth. The detail of the incentives is presented in Appendix D.

Telecom Benefits

• Removal of incentives which artificially grow PSTN traffic.

• Much improved management capability for ISP dialup traffic, particularly during periods of serious overload because this traffic can be prioritised behind voice calls.

The memorandum recorded that under the status quo, it was expected that additional capital expenditure of $205m would be required over the next three years to support projected network usage growth and that if the proposal was not implemented, there would be unsustainable growth in PSTN traffic, creating congestion, or a requirement for investment which would not yield a return and increase the risk of further stranded assets.

[72] The proposal to introduce the 0867 service was publicly announced on 10 June 1999. The reception from industry participants was described by Ms Gattung as “mixed”. Telecom sought the approval of the Kiwi Shareholder. The Crown received legal advice that the 0867 initiative was a breach of the KSO but ultimately agreed not to object to its introduction provided that conditions designed to protect the interests of residential customers were agreed to. Those conditions, recorded in a letter of 15 September 1999 from the Minister of Finance, were that:

• Calls to 0867 numbers must be free to residential customers.

• Telecom must ensure residential calls to 0867 numbers are of no worse quality than ordinary residential calls to the internet.

• Telecom must monitor standards to ensure quality is maintained.

• Telecom must not implement the IDC before 1 November 1999.

• 0867 calls to commercial organisations using 0867 numbers or to a carrier network interface to be free.

[73] After the change of government in November 1999, there were discussions with the incoming Labour led government which had by then announced its intention to initiate the Fletcher Inquiry. Telecom agreed to temporarily suspend the two-cent charge under PIC which had been implemented in November 1999 pending the completion of the Inquiry.

[74] Competing carriers and ISPs who shared termination payments paid to carriers were described as “very exercised” by the announcement of the 0867 package. Telecom embarked on negotiations with Ihug, the biggest ISP after Xtra. On 8 July 1999, a settlement was reached. [

] That was achieved by 29 October 1999. With a further thirty ISPs signed up to use 0867, in total 80% of residential internet traffic (including Xtra) had moved off the PSTN and onto the IN.

[75] Carriers generally resisted the 0867 package. Telecom was able to reach an agreement with Saturn on 31 October 1999. [

] Clear maintained its opposition. Negotiations continued while the Fletcher Inquiry was underway and an agreement was finally reached in October 2000. Clear took the 0867 service. In return Telecom paid Clear a total of $21.75m and forgave it $40m in payments due under the 1996 ICA.

[76] With this necessarily lengthy, but in fact highly condensed account of the background to the introduction of the 0867 service, we pass to a consideration of whether a firm in Telecom’s position in a competitive market could have successfully introduced the 0867 service.

Counterfactual analysis

[77] The first step in the counterfactual analysis is to identify the essential characteristics of a competitive market. There was general agreement that they would include:

• At least three firms operating in the relevant market – Telecom (Company X), Clear (Company Y) and at least one other. None would be dominant. No firm will be earning supra-normal profits.

• Like Telecom and Clear, Companies X and Y would be parties to an ICA with the essential features of the 1996 ICA.

• Like Telecom, Company X would be bound by the KSO.

• Both Company X and Company Y, like Telecom and Clear, would have their own ISP and have other ISPs on their network. It must also be assumed that, as Telecom did with Ihug, Company X was able to come to an agreement with an ISP which gave it no less than 75% in aggregate of the ISPs terminating traffic

• Finally, the counterfactual must assume that Company Y has the small number of residential customers which Clear had relative to its ISP terminating traffic, so as to replicate the net imbalance of termination fees which Telecom experienced. [78] There seems no reason to suppose that in the counterfactual Company X would not achieve the commercial benefits that the 0867 service was designed to secure. In summary they were:

• To bring to an end the high and increasing outflow of charges for terminating access to ISPs on Clear’s network.

• To deny Clear the ability to attract ISPs to its network by sharing terminating revenue, in some cases by permitting ISPs to offer free access to residential internet users.

• To provide an incentive for ISPs to move from Clear’s network to Telecom’s network if Clear refused (as it did) to enter into a new agreement with Telecom to obtain the 0867 number range. Under the 0867 package residential customers not using the 0867 service faced a per minute charge after ten hours of use per month. If an ISP did not switch to Telecom’s network, its residential customers would be encouraged to switch to ISPs who were on Telecom’s network.

• To save Telecom the capital costs of expanding its network to meet the growth in internet traffic usage for residential customers calling ISPs on Clear’s network. These were projected to be $205m. This figure was criticised by the Commission as including the cost of second lines but excluding the additional line rental revenue generated by them. We do not need to resolve the issue; it is clear that substantial savings would be achieved.

[79] The counterfactual test requires us to decide whether competitors of Company X (Telecom) could have taken steps that would have denied Company X the ability to realise those benefits or at least to negate them. If it would not have been profitable for a firm in Telecom’s position to introduce 0867 in a competitive market, it is to be inferred that it was only able to do so in 1999 by virtue of the dominance it enjoyed. Contrariwise, if it could have profitably introduced the 0867 service in an hypothetical competitive market, it could not have used its dominance when it actually introduced the service.

[80] Dr Bamberger made the preliminary point that in a competitive market Telecom would have been compelled to negotiate with Clear by the threat of legal action. We disagree. There is no evidence that Clear could have succeeded in legal action against Telecom, whether for breach of the ICA or otherwise. Grant Forsyth, the former Manager, Industry and Regulatory Affairs at Clear, acknowledged that Clear settled its claims against Telecom when it became clear the government would not intervene to prevent the 0867 package from being implemented. There is no reason to suppose that Company Y would be better placed than Clear to use the threat of legal action to deter the introduction of 0867 in the counterfactual.

[81] Dr Bamberger argued that in the counterfactual Company Y (Clear) could counter the incentives offered by the 0867 package by itself introducing a per minute charge on residential customers calling ISPs on its network at a level just under that of the 0867 package. However, such a charge would also deter residential customers from remaining with ISPs on Company Y’s network; they could avoid the charge by shifting to an ISP on the network of Company X or any other network prepared to subscribe to the 0867 service.

[82] Against the benefits available to Company X (Telecom) in the counterfactual (and to Telecom in fact), the potential costs of introducing the 0867 service must be considered. Company X would face the prospect of losing residential customers on its network who, rather than switch ISPs to avoid the per minute charge under the 0867 package, chose to remain with their ISP but to switch networks. The potential costs to Company X would depend on the number of customers who may switch networks and the net cost per lost customer.

[83] The number of customers who would choose to switch networks appears to be limited. The majority of Company X’s residential customers would not be at risk. They are unaffected by the 0867 package as are customers of Company X’s own ISP (Xtra), Ihug or other ISPs on Company X’s network. In addition, customers of ISPs who are encouraged by the 0867 package to switch from Company Y’s network to Company X’s network will similarly be unaffected. And with Company Y (Clear) refusing to sign up to the 0867 service, Company X is as likely to acquire new residential customers from Company Y as it is to lose customers to Company Y or another carrier’s network. The net loss in residential customers is accordingly likely to be small; a net gain could even result.

[84] Assuming, however, that Company X lost customers, it does not appear that there would be a cost to it. It saves termination charges that it would have been obliged to pay Company Y while foregoing only the per minute charge payable after ten hours use per month. As the per minute charge is equivalent to the termination charge, there is a net saving equivalent to the termination charges for the first ten hours per month. If other services, such as local voice calls and toll calls, were also taken away, there would be no additional cost to Company X. As there are no supra- normal profits being earned in the counterfactual, lost revenue would exactly match the cost of capital and lost profits could be recouped by investing elsewhere.

[85] We do not think the net effect would differ if, instead of migrating from Company X to Company Y (Clear), it is postulated that Company X’s residential customers moved to another carrier, regardless of whether or not the carrier has an ICA with Company Y and the terms of such ICA. If the termination arrangements are bill and keep, the implications of customer migration are exactly the same as if the move had been to Company Y. On the other hand, if the ICA involves asymmetrical termination charges, Company Y would continue to receive termination revenue. But it would be illogical for another carrier to seek business on the very terms that prompted Company X (Telecom) to introduce the 0867 service in the first place.

[86] Dr Bamberger theorised that economies of scale or scope could generate savings that would exceed the cost of capital liberated by lost customers. No evidence was presented to support his contention. We accept that within the telecommunications industry it is reasonable to assume that even in a hypothetical competitive market there are fixed costs and some economies of scale or scope. However, the loss of customers will result in cost savings which must be set off against the (unquantified) benefit of these economies. The evidence showed that in the environment in which Telecom was operating, the retention of customers required additional capital expenditure. Such expenditure would be saved if customers were lost as would, of course, the termination fees they generated. There is no evidential basis which would permit us to find that economies of scale and scope would outweigh the benefits of introducing 0867 in a competitive market. For this purpose, we disregard the improvement to network services that would likely follow the introduction of 0867 and would provide a further benefit to Company X (Telecom) in the counterfactual.

[87] For the purpose of the counterfactual analysis, it is irrelevant whether the KSO is assumed to apply only to Company X or to all firms in the market. It is arguable that in order to achieve the objectives of the KSO, it would have to be applied to all participants in the market. If that were not so, the business of Company X would be rendered unsustainable by its competitors “cherry picking” its most profitable customers and preventing the cross-subsidies inherent in the KSO. It may also be, as Mr Shavin argued, that even if only Company X were bound by the KSO, other firms in a competitive market would be compelled to match its prices.

[88] However, even if the KSO is assumed to apply only to Company X, the outcome will be the same. Company Y (Clear) and other carriers would not be required to maintain a free calling option. They could choose to charge residential customers in order to generate revenue from termination payments and Company Y may even continue to subsidise ISPs on its network. However, in this scenario there is a reduced incentive for residential customers to move from Company X, simply for the purpose of staying with a particular ISP on Company Y’s network. Because there is also a heightened risk that ISPs on Company Y’s network will lose customers, they may choose to relocate to a network that has agreed to provide the 0867 service.

[89] This analysis leads inexorably to the conclusion that a non-dominant firm in Telecom’s position in a competitive market would have been able to introduce the 0867 service in order to deal with the termination fee and network capacity issues which concerned Telecom. It follows that in introducing the 0867 service Telecom did not use its dominance in the market for fixed line retail telephone services to residential customers.

[90] Although our counterfactual analysis has been confined to the market for residential customers, we would have reached the same conclusion, albeit by a somewhat different route, had we been required (by a finding of dominance) to undertake an analysis of the wholesale network terminating access market.

Purpose

[91] As we have found that Telecom did not use its dominant position when it introduced the 0867 service, it is strictly unnecessary for us to decide whether its actions had an anti-competitive purpose. But the issue is not foreclosed by our finding on use of dominance. A firm may have an anti-competitive purpose but not achieve it by using its dominant position – see the comments of the Privy Council in Telecom v Clear at 402 and see also Carter Holt Harvey v Commerce Commission at [26]. As Telecom’s true (as distinct from stated) purpose was at the forefront of argument (and of industry criticism at the time), it is appropriate that we should state our view on the issue.

[92] Purpose may be established by direct evidence that an anti-competitive outcome was a real or substantial purpose of the introduction of 0867. Alternatively, an illegitimate purpose could be inferred from proof that the 0867 service had an anti-competitive effect.

Commission’s case

[93] The Commission’s case is that the relevant proscribed purpose is that referred to in s 36(1)(b) of preventing or deterring any person from engaging in competitive conduct in either the markets in which Telecom is dominant or in any other market. The Commission says that the affected markets are: • The market for wholesale ISP services, that is, the market in which carriers such as Telecom, Clear and Saturn supply telecommunications services to ISPs.

• The wholesale market for originating and terminating access.

• The retail internet access market in which ISPs and others supply internet services to members of the public.

[94] There is no dispute that there are wholesale internet access and retail internet access markets (see [26] and [27] above). We have decided that for the purpose of evaluating dominance, the wholesale market for originating and terminating access should be confined to terminating access. It is appropriate to define it in the same way when considering purpose.

[95] The Commission contends that the introduction of the 0867 service was anti- competitive in purpose and effect because it sought to eliminate or reduce the terminating charges payable to other carriers under the ICAs. As pleaded, the consequence was to constrain competing carriers from competing in the markets by:

56.4 increasing the costs or eliminating or reducing revenues available to competing carriers to offset against the termination charges payable to Telecom under the various interconnection agreements and thereby reducing the funds available to them;

56.5 reducing or eliminating an opportunity available to competing carriers to negotiate lower interconnection charges and/or cost based interconnection charges and/or terms and conditions more advantageous to competing carriers as the various interconnection agreements were renegotiated;

56.6 raising the costs of and/or refusing to compensate competing carriers for the actual cost of terminating calls on their networks;

56.7 forcing changes in Interconnection Agreements on competing carriers where, in the absence of the use by Telecom of its dominant position, it would have had to have made significant concessions or provide significant benefits to those carriers in order to obtain the contractual change;

56.8 demonstrating Telecom’s ability to unilaterally change and/or dictate the terms upon which it would deal with competing carriers and ISPs; 56.9 demonstrating Telecom’s ability to regulate the markets to its own advantage;

56.10 increasing the risk and uncertainty faced by competing carriers and ISPs and thus deterring investment in assets and infrastructure that would enable them to engage in competitive conduct;

56.11 preventing or deterring innovation in the development of the market for ISP Internet access services;

56.12 demonstrating to ISPs that Telecom could interfere with the terms and conditions and the level and quality of service upon which competing carriers could provide services to ISPs;

56.13 interfering with the terms and conditions and the level of service that competing carriers could offer to ISPs;

56.14 causing confusion in the residential and business consumer market for ISP Internet access (for example in relation to cost and service quality) to the detriment of ISPs other than Telecom Xtra;

56.15 imposing costs on the customers of ISPs that were not shared by customers of Telecom Xtra.

[96] An anti-competitive purpose for the purpose of s 36 does not have to be the sole purpose; it is sufficient if it is a substantial purpose. Prior to its repeal in 2001, s 2(5)(b) of the Act provided:

A person shall be deemed to have engaged, or to engage, in conduct for a particular purpose or a particular reason if –

(i) That person engaged or engages in that conduct for that purpose or reason or for purposes or reasons that included or include that purpose or reason; and

(ii) That purpose or reason was or is a substantial purpose or reason.

By s 2(1A) “substantial” meant real or of substance.

Stated purpose

[97] A great deal of evidence and argument focused on Telecom’s stated justification for introducing the 0867 service. The Commission was adamant that its introduction was a suboptimal solution to the network management problem brought about by growth in internet traffic. It was submitted that other numbers could have been used and were offered by Clear but, as they did not fall outside the ICA, would not achieve the objective of getting rid of the termination payment problem. The Commission submitted that this showed that Telecom had the substantial purpose of eliminating termination payments to rival carriers and this was indeed its dominant purpose. The Commission maintained that the elimination of payments to rival carriers and their ISPs was Telecom’s over-riding purpose. This was said to have been demonstrated by Telecom’s failure in April 2000 to adopt a suggested solution to network congestion problems that would not have also resulted in the removal of Clear’s right to receive termination payments.

[98] We do not doubt that other means of addressing the network management problems were available to Telecom and, taken alone, may even have been preferable. It does not follow that the sole or dominant purpose for the introduction of the 0867 service was to eliminate or reduce termination payments.

[99] The way in which the 0867 service was conceived and developed makes it clear that it was crafted to provide a solution to both problems – network management and termination charges. That is hardly surprising. They were related issues; the “perverse incentives” provided by the termination charges exacerbated the network management problems. It is to be expected that Telecom would seek to address them together. We do not think it is helpful or accurate to characterise one as the dominant or primary purpose. We agree with Mr Parkes’ observation that traffic management and the elimination of termination payments were two equal ranking and inseparable objectives, the two feeding on each other.

[100] The goal of eliminating terminating payments for internet calls was, nevertheless, a substantial purpose. It must be conceded that it was capable of being anti-competitive if, as Mr Farmer QC submitted, Telecom’s real or primary objective was to deprive its rivals of revenue, not to reduce its own costs. He maintained this was the case; Telecom was seeking to prevent or deter Clear from exploiting “a legitimate competitive advantage” it had in competing with Telecom. That competitive advantage derived from the terms of the 1996 ICA and its removal was said to be a foreseeable (and foreseen) consequence of Telecom’s conduct. The Commission argued that Telecom could not have been criticised if it had achieved the reduction in costs (and, consequently, in carriers’ revenues) by “lawful” means such as exercising a contractual right to reduce termination charges or by negotiation or other competitive conduct. It is the unilateral imposition of the 0867 scheme and the manner in which the reduction in costs and revenue was achieved which are said to show that Telecom acted for an unlawful purpose.

[101] This submission, as is to be expected, presupposes a finding that Telecom used a dominant position in introducing the 0867 service. Otherwise, its actions were no less lawful than exercising a contractual right or achieving a reduction in termination charges by negotiation.

[102] In seeking to reduce its costs Telecom was engaging in normal profit maximising behaviour, to be expected of any firm, dominant or otherwise. It would be strange indeed if it had not sought to stem the ever-increasing flow of termination charges. There has been no claim and no finding that the introduction of the 0867 service involved a breach of the ICA. A breach of the Commerce Act apart, we were not asked to find that Telecom acted unlawfully. If, as Mr Farmer acknowledged, it would have been legitimate for Telecom to exercise a contractual right to reduce termination charges or to do so by negotiation or by competitive conduct, there is no reason why it was impermissible for Telecom to achieve the same outcome by the strategy it adopted. To paraphrase the majority judgment of the High Court of Australia in Melway Publishing Pty Limited v Robert Hicks Pty Limited (2001) 205 CLR 1 at [67], Telecom’s relative freedom from competitive constraints did not make it wrong to secure business advantages which might not have been so readily achievable in a competitive environment. The comments of the Privy Council in Telecom v Clear at 402 – 403 are apposite:

A monopolist is entitled, like everyone else, to compete with its competitors: if it is not permitted to do so it “would be holding an umbrella over inefficient competitors”: see Olympia Equipment Leasing Co v Western Union Telegraph Co 797 F 2d 370 (1986) per Posner J; Union Shipping NZ Ltd v Port Nelson Ltd at p 706; New Zealand Magic Millions Ltd v Wrightson Bloodstock Ltd [1990] 1 NZLR 731 at p 761; Queensland Wire Industries Pty Ltd v Broken Hill Pty Co Ltd (1989) 167 CLR 177 at 191 per Mason CJ and Wilson J.

[103] In support of his submission that Telecom’s dominant purpose was to disadvantage rival carriers and Xtra’s competitors in the retail internet access market, Mr Farmer pointed to evidence that showed that Telecom was aware that the removal of termination payments would be to the disadvantage of its competitors. He also relied on:

• What was referred to as the CallPlus incident in which CallPlus, an ISP on Clear’s network, sought to avoid its customers paying the two cents per minute charge by a mechanism known as call readdress. When Telecom attempted to block this manoeuvre, CallPlus successfully obtained an interim injunction.

• Evidence that Telecom rushed the 0867 package into service in the knowledge that it suffered from a higher fault rate than other forms of internet access. Mr Farmer referred to evidence that at the time the 0867 service was launched in September 1999, Telecom was aware of a significantly higher rate of faults with 0867 than with other internet access methods.

• In its dealings with Government and the media, Telecom sought to suppress the fact that one of its purposes was to reduce the level of termination payments.

[104] We accept without question that Telecom knew that its strategy would disadvantage its competitors and those of Xtra. We accept that there were indications of technical problems associated with the introduction of the 0867 service that might have been lessened if its launch had been deferred. We agree that in its public statements and in dealings with Government Telecom focused on the advantages of the 0867 scheme for network management purposes. Mr Parkes conceded candidly that Telecom chose to stress the network management issues as it was felt that this was “a more palatable or sellable public message”.

[105] In our view, this evidence takes the Commission’s case no further. It simply confirms that one of Telecom’s objectives was to address the termination charge problem and that it was well aware of what the consequences would be for rival carriers and ISPs. Its response to the attempt by CallPlus to circumvent the 0867 initiative was consistent with Telecom’s overall strategy and, although in issuing an interim injunction the Court found there was a serious question to be tried, it was never found that Telecom had acted in breach of its contractual obligations. As with all of the litigation spawned by the terms of the ICAs, the CallPlus claim was eventually settled.

Inferred purpose

[106] Finally, we consider the arguments put forward on behalf of the Commission by Dr Bamberger that an anti-competitive purpose is to be inferred from the effects of the introduction of the 0867 service. He referred first to short-term effects.

[107] Dr Bamberger said that by opting to introduce the 0867 service, instead of responding to the termination charges problem by competing more aggressively for wholesale ISP and retail ISP customers, Telecom’s choice likely had the effect of preventing or deterring potential short-term competition in the markets for wholesale ISP services and retail ISP services. He said that Telecom could have offered payments to ISPs on Clear’s network to induce them to move to Telecom’s network. Had Telecom done so, other firms such as Clear, may have responded and further increased competition. The introduction of the 0867 package eliminated potential competition in the market for wholesale ISP services that would have been likely to reduce prices paid by ISPs. Alternatively, Dr Bamberger said Telecom could have renegotiated its ICA with Clear and other networks. This would have benefited competitors since Telecom would have had to agree to substantial compensation to provide other firms to give up their termination rights.

[108] Even on the assumption implicit in Dr Bamberger’s argument that Telecom used its dominant position in introducing the 0867 service, we do not accept that it was injurious to competition even in the short-term. We agree with Professors Evans and Hausman that the introduction of the 0867 package, with the consequence of removing the arbitrage opportunity, was pro-competitive. Termination payments and the resultant subsidies to ISPs, leading to low (even zero) prices to consumers, were not the result of competition. They were the result of the KSO and the operation of the 1996 ICA in an era of rapidly escalating internet use. They let to inefficient outcomes, in particular, in the way in which the network was used. A more competitive environment in both the wholesale ISP and retail ISP markets resulted from the introduction of the 0867 service.

[109] Dr Bamberger argued that the introduction of the 0867 package also had long-term anti-competitive effects. He contended that it led to an unexpected change in economic conditions which benefited one party to the 1996 ICA (Telecom) while harming the other (Clear). He said this sent a signal to the marketplace that Telecom would use its dominance to, in effect, unilaterally rewrite contracts when it was in its interests to do so. That prospect would reduce the incentive of competitors to make the investments necessary to compete more effectively with Telecom, so preventing or deterring competition in telecommunications markets.

[110] Professors Evans and Hausman did not accept that the introduction of the 0867 service would discourage investment by competitors in telecommunications markets. They argued that any such effect was limited by the short-term remaining on the ICA; the circumstances giving rise to the introduction of the 0867 service were exceptional; [

] and the removal of the “perverse incentives” created by the KSO was pro-competitive, leading to an improved quality of service and greater dynamic efficiencies.

[111] There was no evidence to suggest that the introduction of the 0867 service had the effect of deterring investment in the telecommunications markets and we do not think the factors cited by Dr Bamberger provide grounds to draw that inference. The circumstances giving rise to the introduction of the 0867 package were unique and would never be repeated. Telecom’s response could not reasonably be regarded as a reliable predictor of the way it would act in the future. In any event, while denying breaches of contractual obligations, Telecom settled claims by carriers and paid substantial compensation to ISPs. And, of course, new ICAs incorporating the sought-after “bill and keep” arrangements were soon agreed. On these facts there is no reliable basis for finding that the introduction of the 0867 service would have had the long-term consequence of deterring Telecom’s competitors from investing in the industry.

[112] We accept that by removing the “perverse incentives” arising from the KSO, the 0867 service could likely have led to increased dynamic efficiency. That would suggest that the introduction of the service had a public benefit. It does not, however, necessarily lead to a finding that there was no anti-competitive effect and we disregard it for that purpose.

[113] We are satisfied that Telecom did not have an anti-competitive purpose when it introduced the 0867 service and that there is nothing in the consequences which followed that could lead to a contrary view.

Result

[114] In introducing the 0867 package Telecom did not contravene s 36 of the Act. The Commission’s case accordingly fails.

[115] The issue of costs can be addressed in the first instance at a telephone conference to be arranged by the Registry.

Rodney Hansen J MC Copeland