IN THE HIGH COURT OF NEW ZEALAND WELLINGTON REGISTRY CIV [2014] NZHC 690. Appellant. COMMERCE COMMISSION First Respondent

Transcription

1 IN THE HIGH COURT OF NEW ZEALAND WELLINGTON REGISTRY CIV [2014] NZHC 690 UNDER the Telecommunications Act 2001 IN THE MATTER OF BETWEEN AND an appeal from a decision of the Commerce Commission CHORUS LIMITED Appellant COMMERCE COMMISSION First Respondent VODAFONE NEW ZEALAND LIMITED Second Respondent TELECOM NEW ZEALAND LIMITED Third Respondent ORCON LIMITED Fourth Respondent 2TALK LIMITED AND OTHERS Fifth Respondent INTERNET NZ, CONSUMER NEW ZEALAND AND TELECOMMUNICATIONS USERS ASSOCIATION NEW ZEALAND INC Interveners Hearing: March 2014 Counsel: D J Goddard QC with A M Peterson and T D Smith for Appellant J A Farmer QC with J D Every-Palmer and J L W Wass for First Respondent P J Radich with T Thursby for Second Respondent S E Fitzgerald with C R Shrive for Third Respondent M B Wigley for Fourth Respondent and Interveners No appearance for Fifth Respondents Judgment: 8 April 2014 CHORUS LIMITED v COMMERCE COMMISSION & ORS [2014] NZHC 690 [8 April 2014]

2 JUDGMENT OF THE HON JUSTICE KÓS Index Introduction... [1] Issues for determination... [8] Appeals under s [10] Statutory framework... [22] Fletcher Inquiry [23] The Act [30] Commission s pre-decision process... [44] Before the draft determination [45] Draft determination [48] UBA conference [62] Professor Vogelsang s report [63] Commission s Update Paper [71] Decision appealed... [83] Issue 1: Did the Commission in fact adopt the Swedish price as an upper bound, and thereby err in law?... [100] Submissions [100] Discussion [108] Conclusion [119] Issue 2: Did the Commission in fact fail to consider s 18 other than in the context of price point selection, and thereby err in law?... [121] Submissions [121] Discussion [129] Conclusion [141] Issue 3: Did the Commission in fact fail to determine what inferences could reliably be drawn from the two benchmarks about the likely cost of providing the UBA service in New Zealand, and thereby err in law?... [143] Discussion [152] Conclusion [157] Issue 4: Did the Commission in fact fail to consider whether a price above $10.92 would be more consistent with s 18, and thereby err in law?... [158] Issue 5: Did the Commission fail to consider whether a price reduction of this magnitude was consistent with the s 18 purpose, having regard to the limited information available to it about the cost of providing the UBA service in New Zealand, and thereby err in law?... [158] Submissions [159] Discussion [165] Conclusion [172] Issue 6: If yes to any of the above, what remedy should be granted?... [173] Conclusions... [174] Result... [180]

3 Introduction [1] Unbundled bitstream access (UBA) is a service offered by Chorus. It involves carrying broadband traffic back and forth between end users (consumers) and access seekers (service providers) across Chorus s copper local loop, through its exchange (or a remotely located cabinet), and through Chorus s data switch. There it is handed over to (or from) the access seeker s network. [2] Chorus s pricing for this service is regulated. The price paid is the sum of two parts: the unbundled copper local loop price (the UCLL price, $23.52 per month) and the UBA additional component price (the UBA price). 1 Until 1 December 2014 the UBA price is grandfathered, at $21.46 per month. The Telecommunications Act 2001 (Act) requires the Commerce Commission to determine the UBA price. There are two stages. First it has to undertake a review of the UBA price using an Initial Pricing Principle (IPP). The IPP is based on international benchmarks. Secondly, but only if application is made, the price is reviewed using the Final Pricing Principle (FPP). That involves creating a model of the total service long run incremental cost (TSLRIC) of providing the UBA service in New Zealand. [3] The Commission was to use reasonable endeavours to undertake an IPP price determination by 1 December Its process was consultative, involving a draft determination and a conference. It had its complexities. Not least because only a small number of benchmarks could be found on which to base the IPP exercise. The Commission was not ready to release its determination in December But that month it released a draft determination. It suggested a UBA price of $8.93 per month. The Commission did not issue its final IPP determination until 5 November That set a new UBA price of $10.92 per month. [4] Chorus (and four other parties) have applied for the price to be reviewed using the FPP. In addition, Chorus has appealed the November 2013 IPP determination to this Court. Appeals are restricted by the Act to questions of law. 1 These concepts are more easily understood if reference is made to the diagram at [31], below.

4 [5] In essence, Chorus says the Commission erred in law in adopting the $10.92 figure. That was the comparable Swedish price, and the highest figure in the benchmark range. But only two benchmark countries (Denmark and Sweden) had been found to be sufficiently comparable. Chorus says the Commission wrongly directed itself not to go beyond the benchmark range, thereby treating the Swedish price as a bookend for the IPP exercise. In addition, Chorus says the Commission failed to consider the overall purpose principle, s 18 of the Act, throughout the process. Rather, it considered it only later on, at the price point selection stage. Chorus submits the Commission failed to determine the inferences that could reasonably be drawn from the two benchmarks about the likely cost of providing UBA service in New Zealand, failed to consider whether a price above $10.92 would be more consistent with s 18, and failed to consider whether a price reduction from $21.46 to $10.92 per month was consistent with s 18 of the Act. [6] The Commission (supported by other respondents and interveners in this proceeding) says that the benchmark selection process was robust and consistent with the criteria stated in the Act. That process provided the key evidence the Commission could work from. The Commission did not bookend itself. Rather it stated it was open to adopting a plausible range extending beyond the benchmark range. It considered methods of doing so. But, the Commission says, there was no reliable evidence available to cause it to move, ultimately, outside the benchmark range. The Commission says there was nothing unlawful about it giving primary consideration to s 18 at the price point selection stage of the process. It says that, after balancing all the available (but necessarily limited) evidence, and applying s 18, it selected a price of $ The Commission says it was not bound to move beyond the evidence, or seek separate connecting evidence to TSLRIC values in New Zealand. The IPP process was supposed to be a rapid one. If any party was disgruntled with the outcome, the Act provides the remedy through an FPP-based review. Chorus has done so, seeking a price increase. Others have done so, seeking a price reduction. [7] The impatient reader may, if he or she wishes, go straight to my conclusions. They begin at [174], below. Exigency compels early delivery of this judgment.

5 More time may have altered the expression of my reasons. But not my overall conclusions, which are clear. Issues for determination [8] The agreed issues for determination are: 2 (a) Issue 1: Did the Commission in fact adopt the Swedish price as an upper bound, and thereby err in law? (b) Issue 2: Did the Commission in fact fail to consider s 18 other than in the context of price point selection, and thereby err in law? (c) Issue 3: Did the Commission in fact fail to determine what inferences could reliably be drawn from the two benchmarks about the likely cost of providing the UBA service in New Zealand, and thereby err in law? (d) Issue 4: Did the Commission in fact fail to consider whether a price above $10.92 would be more consistent with s 18, and thereby err in law? (e) Issue 5: Did the Commission fail to consider whether a price reduction of this magnitude was consistent with the s 18 purpose, having regard to the limited information available to it about the cost of providing the UBA service in New Zealand, and thereby err in law? (f) Issue 6: If yes to any of the above what remedy should be granted? [9] Issues 1 and 2 dominated the appeal. Issues 4 and 5 in effect merged in argument. 2 The content of this list was developed with counsel at the hearing. It ultimately reflects terms counsel for Chorus were content with. Counsel for the Commission dissented slightly from the expression of Issue 5, which had been added by counsel for Chorus, but did not suggest any alternative wording. I have included the former discrete Issue 6 ( if yes to any of the above, was that an error of law? ) within each of Issues 1 5, thus obviating its further inclusion.

6 Appeals under s 60 [10] Section 60 of the Act provides that a party to a determination may appeal against that determination on a question of law. [11] There was little disagreement before me on the relevant principles applying to appeals of law under Part 2 of the Act. The governing principles may be drawn from either appeals on questions of law, or judicial review proceedings involving alleged error of law. As the Court of Appeal of England and Wales observed in E v Secretary of State for the Home Department, it has become a generally safe working rule that these grounds for intervention are identical across the two streams. 3 [12] An appeal on a question of law is not a general appeal by way of rehearing. It is not a merits review. A functional merits review for the present determination can in fact be found in s 42 of the Act. That provides that a party to a designated access service price determination (such as here) may apply for review of the price set in that determination. In that case, as we shall see, the Commission (which conducts the review) is obliged to approach pricing on a rather different basis. It has to build a model which establishes what the Commission concludes to be the TSLRIC for the service. Chorus and four other parties have taken that opportunity in this case. So there is an appeal (on law, to the High Court) and a merits review (by the Commission), both proceeding concurrently. [13] It was common ground before me that this Court may only interfere with a tribunal s decision, on a question of law appeal, if it considers that the tribunal: (a) applied the wrong legal test; or (b) took into account matters which it should not have taken into account; or 3 E v Secretary of State for the Home Department [2004] EWCA Civ 49 at [40]. Applied in Queenstown Lakes District Licensing Agency Inspector v Turnbull Group Ltd [2011] NZAR 554 (HC) at [32].

7 (c) failed to take into account matters which it should have taken into account; or (d) came to a view without evidence, or reached a view which, on the evidence, it could not reasonably have come to. [14] Logically those grounds break into two broad categories or manifestations. The first is where the tribunal has misinterpreted what is required of it under the statutory framework i.e., ground (a) above. The second is where what the tribunal has done in its fact-finding function is so misconceived that it constitutes an unlawful decision i.e. grounds (b) to (d) above. [15] In relation to the first manifestation of error of law applying the wrong legal test I bear in mind an important submission made to me by Mr Radich (on behalf of Vodafone). The point he made was that statutory language is not necessarily wholly prescriptive. There will be a correct interpretation, of course. But that interpretation will often leave open options to a decision-maker. In which case, judgment is required. 4 In the province of criminal law, for instance, statutory language is likely to be relatively more prescriptive. Economic regulation, on the other hand, is notoriously difficult to prescribe, given the extraordinary variety of business practices, markets and circumstances that fall to be addressed. 5 The reality of economic regulation is that statutes present a chart of medium scale at best. The exact route to be taken is left to the judgment of the navigator, the decision-maker. Usually, as here, an expert tribunal for that very reason. In such cases, the decisionmaker may have an area of judgment. 6 [16] In Vodafone New Zealand Ltd v Telecom New Zealand Ltd 7 Blanchard J adopted the observations of Lord Mustill, delivering the advice of the House of Bennion rightly distinguishes between the statutory requirement of a decision-maker to exercise judgment, and the conferral of a discretion: Bennion on Statutory Interpretation (5 th ed, LexisNexis, London, 2008) at Tanner, Confronting the Process of Statute-Making in Bigwood (ed) The Statute: Making & Meaning (LexisNexis, Wellington, 2003) 49, at 96. As Lord Bingham of Cornill CJ put it in R v Criminal Cases Review Commission ex parte Pearson [1999] 3 All ER 498, at 523. Vodafone New Zealand Ltd v Telecom New Zealand Ltd [2011] NZSC 138, [2012] 3 NZLR 153 at [54] [55]; R v Monopolies and Mergers Commission ex parte South Yorkshire Transport Ltd [1993] 1 WLR 23 (HL).

8 Lords in R v Monopolies and Mergers Commission ex parte South Yorkshire Transport Ltd. That case concerned the construction of the words a substantial part of the United Kingdom in legislation governing the investigation and mergers of transport services. As Blanchard J put it: 8 Lord Mustill drew attention to the protean nature of the word substantial, ranging from not trifling to nearly complete. He cautioned against taking an inherently imprecise word and by redefining it thrusting on it a spurious degree of precision. Blanchard J went on to approve Lord Mustill s observation: 9 Once the criterion for a judgment has been properly understood, the fact that it was formerly part of a range of possible criteria from which it was difficult to choose and on which opinions might legitimately differ becomes a matter of history. The judgment now proceeds unequivocally on the basis of the criterion as ascertained. So far, no room for controversy. But this clear-cut approach cannot be applied to every case, for the criterion so established may itself be so imprecise that different decision makers, each acting rationally, might reach differing conclusions when applying it to the facts of a given case. In such a case the Court is entitled to substitute its own opinion for that of the person to whom the decision has been entrusted only if the decision is so aberrant that it cannot be classed as rational: Edwards v Bairstow. [17] A similar conclusion was reached in an earlier Supreme Court decision, Unison Networks Ltd v Commerce Commission. 10 There legislation required the Commission to set thresholds for the control of electricity lines companies. Delivering the judgment of the Court, McGrath J said: 11 The requirement that it set thresholds could have been lawfully tackled by the Commission in one of two ways. First, the Commission could have set thresholds which ascertained those companies operating inefficiently and restricted the prices they could each charge by imposing controls. This would have included individual profit adjustments. The difficulties in implementing this approach, due to the time it would take and the resources that would be involved in collecting information, made it unattractive. Alternatively, the Commission could seek to set price paths based on the information it had available for the industry as a whole, fine-tuning them through resetting as it became better and more specifically informed. Both approaches were within the terms of the provisions of the subpart. The Commission s chosen approach for the initial and revised thresholds was the latter one and it was lawful At [54]. R v Monopolies and Mergers Commission Ex Parte South Yorkshire Transport Ltd [1993] 1 WLR 23 (HL) at 32. Unison Networks Ltd v Commerce Commission [2007] NZSC 74, [2008] 1 NZLR 42. At [77].

9 [18] It follows, then, that statutory language may be imprecise, and decisionmakers may have to exercise judgment among options available on the statutory words. The only question, then, is whether the judgment made was a permissible one, or one open to the decision-maker. [19] Where the second manifestation of error of law is involved error involving serious misconception in the tribunal s fact-finding function a very high hurdle is involved. 12 As the Supreme Court put it in Bryson v Three Foot Six Ltd: 13 Provided that the [tribunal] has not overlooked any relevant matter or taken account of some matter which is irrelevant to the proper application of the law, the conclusion is a matter for the fact-finding [tribunal], unless it is clearly unsupportable. What matters is whether the finding being appealed was a permissible option or open to the tribunal. 14 [20] I accept the submission made to me that greater latitude in reviewing the exercise of a tribunal s fact-finding function will be given where that function involves determination within the tribunal s particular area of expertise. That much was acknowledged by the Supreme Court in Unison Networks Ltd v Commerce Commission: 15 Often, as in this case, a public body, with expertise in the subject matter, is given a broadly expressed power that is designed to achieve economic objectives which are themselves expansively expressed. In such instances Parliament generally contemplates that wide policy considerations will be taken into account in the exercise of the expert body s powers. The courts in those circumstances are unlikely to intervene unless the body exercising the power has acted in bad faith, has materially misapplied the law, or has exercised the power in a way which cannot rationally be regarded as coming within the statutory purpose Bryson v Three Foot Six Ltd [2005] NZSC 34, [2005] 3 NZLR 721 at [27]. At [25]. Piggott Brothers & Co Ltd v Jackson [1992] ICR 85 (CA) at [92]; Lewis v Wilson & Horton Ltd [2009] 3 NZLR 546 (CA) at [62]. Unison Networks Ltd v Commerce Commission [2007] NZSC 74; [2008] 1 NZLR 42 at [55]; see also Major Electricity Users Group Inc v Electricity Commission [2008] NZCA 536 at [56] [57].

10 [21] Although that was an observation made in a judicial review challenge, as I have already said the principle applies equally in appeals confined to questions of law. 16 Statutory framework [22] I turn now to describe the statutory framework, and in particular the relevant provisions of the Act (as amended by the Telecommunications Amendment Act 2011). Before doing so, I need to look at the genesis of that legislation. That was the report of the Ministerial Inquiry into Telecommunications (the Fletcher Inquiry) of September Fletcher Inquiry [23] The Fletcher Inquiry was established by the Government in March 2000 to consider whether changes needed to be made to the very limited regulatory framework that had applied to telecommunications since the former state-owned incumbent, Telecom New Zealand, was privatised in Apart from the Commerce Act 1986, telecommunications services were substantially unregulated. As the Inquiry noted, New Zealand had been one of the first countries in the world to privatise and deregulate telecommunication services at the same time. Service growth, and some price falls, had occurred. But this had happened just as dramatically in other countries. And no other country had taken New Zealand s approach of almost exclusive reliance on pure competition law to govern access. As the Inquiry put it, without exception, [other countries] have opted for industryspecific regulation. [24] The Inquiry issued a draft report in June It supported a greater measure of regulation, albeit still light-handed, of certain designated services. Where interconnection was designated, a new regulatory pricing model would be needed. It noted support for an international benchmarking ( read across ) approach. 17 That was because a cost-based approach was more complex, expensive and time consuming. And it was unlikely to result in agreement, given different views on the At [11]. Ministerial Inquiry into Telecommunications, Draft Report (June 2000) at [7.1.1].

11 large number of assumptions necessary. The Inquiry saw considerable merit in international benchmarking, with New Zealand able to piggy-back off cost-based modelling from other countries with similar characteristics to New Zealand. 18 As the Inquiry noted in its draft: 19 The key advantage of a benchmarking approach is that, because interconnection prices are readily available in other markets, they can be applied quickly and cheaply (relative to cost-based models). [25] The Inquiry received submissions. It then issued a final report in September The final report recommended a dual approach: an initial price determination and (if required) a second price review determination. It recognised that different pricing principles might have to apply, given the short time frame for an initial determination to be given. The report said: 20 The initial determination would ideally get sufficiently close to the efficient price so that both parties accept the determination and decide not to progress to the (longer and more costly) pricing review determination. Initial pricing principles therefore would need to: reproduce as far as practicable the likely results of applying the pricing principles that would apply to any pricing review determination; and have regard to what is practicable to apply within the timeframe for initial determinations (30 working days, or 50 with public hearings). [26] The final report went on to recommend the use of benchmarking for the initial determination. It said: 21 The Inquiry recommends using benchmarking for the initial determination, using OECD countries that regulate such services using a TSLRIC type methodology and that do not have a state owned incumbent. Benchmarks should be adjusted for factors outside the control of the incumbent that materially impact the cost of providing the service, such as differences in customer density and traffic patterns. The Inquiry notes that a benchmarking approach would give a range of prices that could be ranked from lowest to highest by country or, as in the United States, by operator/state. A judgement would then be required as to where New Zealand should, at any time, fit within the ranking. The Inquiry It noted that the benchmarks would need to be amended where necessary for factors in New Zealand different to the benchmark countries, such as differences in regards to customer density and traffic patterns. At [7.1.3]. Ministerial Inquiry into Telecommunications, Final Report (September 2000) at [6.4]. At [7.2.1].

12 recommends that this judgement be made by the Commissioner on the basis of his/her best estimate of where New Zealand would fall if a full TSLRIC assessment were undertaken. This is consistent with the earlier expressed view that the initial pricing principle should be a proxy for the pricing principle to be used in the pricing review determination... This is to minimise the difference in resultant prices, both to achieve efficient pricing and to reduce the likelihood of a party seeking a pricing review determination. [27] A number of inferences can be drawn: (a) The initial determination might well be different from the pricing review determination if the latter was sought. (b) Ideally, however, it would get close to the price expected on review, so that parties would not need to implement that step. (c) Benchmarking (using OECD countries regulating the relevant designated service, using a TSLRIC-type methodology, and without a state-owned incumbent) would be used. (d) The benchmarks would need adjustment for factors outside the incumbent s control materially impacting cost of supply. (e) Judgment would then be made by the regulator, on the basis of its best estimate of where New Zealand would fall if a full TSLRIC assessment were undertaken, evidently from within the benchmark range (as adjusted). [28] The result would then be a proxy for the result in any pricing review determination. All parties before me on this appeal agree that that principle carried through into the legislation. [29] I move now directly to the Act, and the 2011 Amendment Act. 22 Counsel advised me that no further extrinsic illumination could be found in the Parliamentary debates. 22 Telecommunications (TSO, Broadband, and Other Matters) Amendment Act Herein, the Amendment Act.

13 The Act [30] The principal purpose of the Act is to regulate the supply of telecommunication services. The Act provides for the regulation of designated services listed in sch 1. [31] One designated service is Chorus s unbundled bitstream access service i.e. the UBA. Another is Chorus s unbundled copper local loop network service or UCLL. How these two fit together is demonstrated by a useful diagram provided by the Commission: [32] To explain: access seekers wishing to offer broadband services using the copper network typically purchase the UCLL service or the UBA service from Chorus. Both are now regulated under the Act. When Chorus provides an access seeker with the UBA service, it carries the broadband traffic between the end user and the handover point on behalf of the access seeker. It provides access to the copper local loop, the exchange or cabinet (including a DSLAM multiplexer) and along the aggregation path carrying aggregated DSL traffic from the DSLAM to the handover point (the first data switch that traffic from the end user comes to in Chorus s network). [33] Part 2 of the Act, providing for regulation of designated services, commences with a statement of purpose. It is set out in s 18:

14 Purpose (1) The purpose of this Part and Schedules 1 to 3 is to promote competition in telecommunications markets for the long-term benefit of end-users of telecommunications services within New Zealand by regulating, and providing for the regulation of, the supply of certain telecommunications services between service providers. (2) In determining whether or not, or the extent to which, any act or omission will result, or will be likely to result, in competition in telecommunications markets for the long-term benefit of end-users of telecommunications services within New Zealand, the efficiencies that will result, or will be likely to result, from that act or omission must be considered. (2A) To avoid doubt, in determining whether or not, or the extent to which, competition in telecommunications markets for the long-term benefit of end-users of telecommunications services within New Zealand is promoted, consideration must be given to the incentives to innovate that exist for, and the risks faced by, investors in new telecommunications services that involve significant capital investment and that offer capabilities not available from established services. (3) Except as otherwise expressly provided, nothing in this Act limits the application of this section. (4) Subsection (3) is for the avoidance of doubt. [34] Subsection 2A was added by the Amendment Act in The dominant provision remains s 18(1). The purpose of Part 2 is to promote competition in telecommunications markets for the long-term benefit of end users of telecommunication services in New Zealand. Subsections 2 (focusing on efficiencies) and 2A (focusing on incentives to innovate, and risks faced by, investors in new telecommunications services involving significant capital investment and offering capabilities not available from established services) are specified for the purpose of assisting analysis under s 18(1). It is common ground that efficiencies refer to both static and dynamic efficiencies. The latter are concerned with innovation, either of existing services or by the introduction of new services. In balancing those, and in light of s 18(2A), the Commission has said it should give greater weight to dynamic efficiencies, given the Act s emphasis on promoting competition aimed at end-user welfare over the long term. [35] Section 19 of the Act then requires the Commission, in making any determination, to consider the purpose set out in s 18. The Commission must make

15 the determination it considers best gives, or is likely to give, effect to the purpose set out in s 18. [36] So that is how regulation of the designated services under the Act begins. [37] Subpart 2 of the Act enables an individual access seeker to apply to the Commission for determination of the terms (including price) on which a designated service (including UBA and UCLL) must be supplied. Subpart 2A, inserted in 2006, provides for the Commission to instead make a determination of the terms, including price terms, on which a designated service must be supplied to all access seekers. The present decision was made under subpart 2A. Standard terms for the supply of the UBA service were set by the Commission in Decision 611 (12 December 2007). I was told that the price aspect of the standard terms determination has been reviewed on a number of occasions since then, the last occasion prior to the present decision being in November [38] Matters were then changed, fundamentally, by the Amendment Act in That changed IPP from a retail-minus price to a forward-looking cost-based price. The amendment coincided with the structural separation of Telecom New Zealand. Chorus was one of the by-products. The previous price for UBA access $21.46 per month was frozen by the Amendment Act for three years, from 1 December 2011 (separation date) to 1 December [39] After the three year period, the Act required pricing to be set on the basis of the initial pricing principle (IPP) or (if reviewed) final pricing principle (FPP). For UBA service, sch 1 of the Act defines those thus: Initial pricing principle applicable after the expiry of 3 years from separation day. The price for the designated access service entitled Chorus s unbundled copper local loop network plus benchmarking additional costs incurred in providing the unbundled bitstream access service against prices in comparable countries that use a forwardlooking cost-based pricing method (emphasis added) 23 Amendment Act, s 75; and Act, sch 1, pt 2, sub-pt 1.

16 Final pricing principle applicable after the expiry of 3 years from separation day: The price for Chorus s unbundled copper local loop network plus TSLRIC of additional costs incurred in providing the unbundled bitstream access service. [40] Moving, after the three year freeze period, from a price based on retail-minus to a TSLRIC-based price in effect, the costs of an efficient access provider was structurally, and potentially economically, a significant change. The regulatory impact statement provided by the Ministry of Economic Development at the time of the Amendment Act anticipated a drop in UBA pricing. 24 Mr Goddard QC, for Chorus, conceded that his client accepted that a price reduction was more likely than a price increase. [41] Section 77 of the Amendment Act required the Commission to undertake a review of the UBA price under s 30R to implement the new pricing principles, with effect from 1 December Section 77(1) required the Commission to make reasonable efforts to review UBA service standard terms (including price) by 1 December 2012 a year after separation day. The exercise, however, proved protracted. The decision under appeal was not issued until 5 November [42] Section 78 of the Amendment Act then provides that a party to that determination may apply for a review under s 42. In that review FPP, rather than the IPP, is applied. Section 78(3) requires the Commission to make reasonable efforts to complete that review by 1 December As I have said already, Chorus and four other parties have applied for a s 42 review. Albeit with different expectations of the outcome. It was common ground before me that it would be difficult for the Commission to complete the s 42 review by 1 December Counsel for the Commission evinced some optimism. That did not seem generally to be shared. If the Commission issues its s 42 review determination by 1 December 2014, the IPP price (the subject of this appeal) will not take effect. It was suggested to me that even if the FPP price is determined after 1 December 2014, it is open to the 24 Ministry of Economic Development Regulatory Impact Statement: Regulatory Issues if Telecom Becomes a Partner in the Ultra-fast Broadband Initiative (March 2011) at [40].

17 Commission to backdate the effect of the determination to 1 December express no view on that proposition. I [43] Section 53 of the Act provides some guidance as to process to be adopted by the Commission: Procedure for determinations For a determination made under this Part, the Commission (a) (b) (c) is not bound by technicalities, legal forms, or rules of evidence: may inform itself of any matter relevant to the determination in any way it thinks appropriate: must consider all submissions made in relation to the determination and all information and opinions presented or expressed at any conference or public hearing in relation to the determination. Commission s pre-decision process [44] I turn now to the process the Commission adopted, which Chorus condemns. Before the draft determination [45] As already noted, the present determination was required by s 77 of the Amendment Act. The Commission was required to determine, using the IPP, a new forward-looking cost-based price to come into effect on 1 December The legislation required the Commission to use reasonable efforts to do that by 1 December [46] In June 2012 the Commission published a discussion paper. In it the Commission noted that its proposed approach to benchmarking would involve identifying countries setting forward-looking cost-based prices, identifying services (or service components) similar to the UBA service, and identifying whether the countries were comparable. If the benchmark service included the equivalent of UCLL and its cost, that would need to be deducted. The forward-looking cost-based 25 Telecom New Zealand Ltd v Commerce Commission HC Auckland CIV , 8 April 2005 and Telecom New Zealand Ltd v Commerce Commission CA 75/05, 25 May 2006.

18 price would need to be geographically averaged. Relativity between the UBA and UCLL services would need to be considered. [47] Submissions and cross-submissions were received in August and September Draft determination [48] In December 2012 the Commission published its draft determination. Extensive reference to this document was made by counsel for both Chorus and the Commission. I will, therefore, spend a little time on it. [49] As the draft determination noted, the IPP set out at [39] above had two parts. The first was the UCLL standard price. The second part was the benchmarking element. [50] The UCLL price had been updated on 3 December 2012 to $ The price for UBA would, therefore, be $23.52 plus the additional costs incurred in providing the UBA service determined by the benchmarking exercise. The Commission noted that the IPP was intended to be a proxy for the FPP. There is general agreement that that direction is correct. After looking at the definition of TSLRIC, the Commission noted that s 19 of the Act required it to consider the purpose in s 18, and it set that provision out in full. [51] The draft determination recorded the Commission had used two main criteria to identify countries to be included in the benchmark set: (a) Forward-looking cost-based pricing method: only countries regulating bitstream prices on a forward-looking cost-based pricing methodology would be included. (b) Comparable countries: the Commission said that the country characteristics that are relevant to the cost of providing a wholesale bitstream service must be similar to those of New Zealand.

19 In fact a third criterion service similarity was also employed. 26 [52] The Commission then identified 31 countries as possible benchmark candidates. That exercise was based on a number of sources, including questionnaire responses received from overseas regulators approach by the Commission. [53] The first criterion forward-looking cost-based pricing method was broken down further. The Commission used a four-step filtering process. The first filter was whether cost-based regulated pricing was used. That eliminated 20 of the 31 countries. The second filter was whether countries used a forward-looking TSLRIC methodology (or equivalent) in calculating regulatory price. That eliminated a further five countries: France, Spain, Bahrain and the United Kingdom used a fully distributed cost (FDC) methodology. The third filter was whether current costs were used in regulated pricing. Poland used historic cost accounting in determining its TSLRIC based price. Out it went, then. The fourth filter was whether the cost model was designed, or expressly reviewed and approved, by the regulator. Three countries (Switzerland, Greece and Slovakia) did not have regulator-verified models. [54] That left three countries that could pass through the four filters: Belgium, Denmark and Sweden. [55] Next, having looked at which countries had forward-looking cost-based pricing methods, the Commission considered the second criterion country comparability. It sought to check whether the countries identified had broadband market characteristics broadly similar to New Zealand. As the Commission put it, the size of the DSL markets in these countries is particularly important, as UBA services are provided using DSL. It focused on the three countries that had met the first criterion. Considering both the number of broadband subscriptions and the degree of broadband penetration in the three countries, the Commission was satisfied that all three were suitable as benchmarks for setting the New Zealand UBA service price. 26 See [56] [57] below.

20 [56] The Commission then applied a third overall criterion whether the benchmark services were sufficiently similar to the UBA service. That is not expressly referred to in the IPP. But it was common ground before me that it is implicit in the IPP that benchmarking comparable countries must involve comparable services. Here, the Commission focused on the location of the handover point, class of service, speed of service and the technology used to provide it. [57] The result of this third criterion analysis was that the Commission found that only Denmark and Sweden had broadly similar UBA services to New Zealand. But Belgium was different. In particular, because its handover point was co-located with the DSLAM. Reference back to the diagram at [31] will demonstrate the significance of that difference. The Commission considered whether it could undertake adjustments to allow for the difference in handover point. But it felt there was no intuitive and transparent way of determining such an adjustment. [58] The result was the Commission s draft analysis found that only Denmark and Sweden had regulated UBA pricing consistent with the IPP. [59] The draft determination then considered how to determine UBA price on the basis of that two member benchmark set. The Commission first undertook a currency conversion exercise. That produced a Danish UBA price (excluding UCLL) of $10.51, and a Swedish price of $ The median (which is the same as the mean in a two sample set) was $8.93. (Had Belgium been included, its bitstream price would have been $6.76, and the mean would have been lower, $8.21.) The Commission said its favoured approach was to start with that two sample set median of $8.93, and then consider whether there were grounds to deviate from that point because of expected differences in cost, or to address s 18 concerns such as relativity. [60] The Commission then discussed s 18. It noted that UBA pricing would affect UCLL unbundling in the near term, and might also affect transition to fibre in the medium term. Should a higher price point than the mean benchmark UBA price be chosen to encourage investment? Its preliminary view was not. The Commission 27 These figures were corrected, for a number of reasons, subsequently. Ultimately the comparable service prices were found to be $8.88 for Denmark, and $10.92 for Sweden.

21 felt that an adjustment to that price was not required to maintain appropriate incentives for access seekers to invest in copper loop-based services. The Commission noted the specific matters required by ss 18 and 19, and the complexity of setting such a price to start in two years time. It sought submissions on relativity between the pricing of UBA and UCLL, implications for investments in those services, whether there were asymmetric economic costs in setting the UBA price too high or too low, the likely impact on incentives to invest in broadband services, whether over copper or fibre, and effects on end users. [61] The Commission specifically noted s 18(2A) in its draft determination. It considered, tentatively, that the issue was the same for both UCLL and ultra-fast broadband (UFB): whether the UBA price encouraged investment and a take-up of competing services that would benefit end users. The Commission was unclear whether a UBA price higher than benchmark sample mean was likely to lead to investment in new innovative services, over copper or fibre, given access seekers would have an incentive to upgrade to fibre in order to differentiate their services from copper-based ones. The draft determination concluded, our preliminary view is that the mean price point best gives effect to the interests of end users. So that would be a UBA price of $8.93 per month. UBA conference [62] Submissions and cross-submissions on the draft determination were received in February and March A UBA conference was then held in June It lasted two days. Considerations of particular importance at the conference included the robustness of setting the IPP using a two country benchmark set; adjustments and amendments to provide additional robustness; the potential inclusion of other countries in the benchmark set through the relaxation of the benchmarking criteria used by the Commission; comparable service speed; and (particularly important for present purposes) the impact of fibre on the migration from copper, the application of s 18 of the Act, and the possibility that the price determined could fall outside the benchmark range.

22 Professor Vogelsang s report [63] The Commission then commissioned a report from Professor Ingo Vogelsang, an economics professor at Boston University specialising in the regulation of network industries. The report (which was produced on 5 July 2013), has a lengthy title: What effect would different price point choices have on achieving the objectives mentioned in s 18, the promotion of competition for the long-term benefit of end users, the efficiencies in the sector, and incentives to innovate that exist for, and the risks faced by investors in new telecommunication services that involve significant capital investment and that offer capabilities not available from established services? The professor recorded that the Commission had asked him to conduct an independent analysis of the key s 18 economic considerations, when undertaking a price review for UBA, by answering the question forming the title of his paper. [64] The approach Professor Vogelsang took (whether of his own initiative, or on instruction, it is not clear) was that the Commission first had to determine the feasible benchmark set, and the resulting range of outcomes, before addressing the s 18 issues. That approach is a point in issue in this appeal. 28 [65] The key point being addressed by Professor Vogelsang was this: the IPP was unlikely to produce the true UBA cost (i.e. the TSLRIC, measured under the FPP). Given that fact, what were the effects of an increase or a decrease in the UBA price compared to the true UBA cost in terms of the s 18 objectives? [66] As a preliminary point, Professor Vogelsang was clear that the true UBA cost established in a TSLRIC model would be substantially below the current UBA retail-minus price of $21.46 per month. This was, he said, clear from the evidence presented in the draft determination. And it was clear from his own knowledge of other areas of the world. As the professor put it: I would have expected a ballpark figure of about NZ$10 per month, based on international data, certainly nothing in the range of the retail-minus price of $ Issue 2. See [121] et seq.

23 [67] Professor Vogelsang proceeded, also, on an assumption (based on the balance of the submissions) that expected true TSLRIC of UBA for New Zealand would lie above the median identified in the draft determination. 29 He then noted what he considered to be the highly uncertain nature of adjustments necessary to account for different line densities. DSL density in Sweden was slightly lower than in New Zealand, but substantially higher in Denmark. As the professor put it: A clean (econometric) analysis for adjusting the benchmark results to reflect the New Zealand circumstances would be quite complex (including taking care of the adjustments and equipment of network architecture for different densities) and would, in my view, totally defeat the purpose of the IPP. It would, in my view, be out of proportion and would catapult the IPP both in terms of resource use and time requirement to the same level as the FPP. We therefore do not want to go further into the nitty-gritty of explaining or adjusting the benchmark figures for Denmark and Sweden. Nevertheless, an adjustment by the weight given to each of the two countries may be warranted. For example, Sweden may actually be closer in geographic properties to New Zealand than Denmark is. In conclusion, if the benchmarking shows Denmark with a lower UBA cost than Sweden, but Sweden has lower UBA density than Denmark, then the cost differences between the two countries may well be explained by the differences in density. Because New Zealand s density is very close to that of Sweden, the Swedish observation is probably much closer to the true expected value for New Zealand than the Danish observation. Consequently, a value at the 75 per cent or even 100 per cent mark between the benchmark costs of Denmark and Sweden appears to be justified. [68] Professor Vogelsang then undertook an indicative error analysis. He noted that the probability distribution of the true cost would not necessarily fit in the interval spanned by the corrected Danish costs at the lower bound and Swedish costs at the upper bound. 30 Cognitively, the two benchmark observations might define a distribution, but almost certainly the true distribution would extend well beyond those points. That true distribution was probably also asymmetric. He therefore suggested a better way of looking at the relevant range was between the Danish price (as a lower bound) and 150 per cent of the inter-benchmark difference. This was depicted diagrammatically: Professor Vogelsang noted a number of submissions (supported by evidence) in favour of a price higher than median, but none for a price lower. That is the correct ordering. Since the draft determination it had become apparent that the calculated Danish cost was overstated and calculated Swedish cost understated. As noted before, the final determination had Denmark at $8.88 and Sweden at $10.92 (median $9.90). The diagram, which is exactly as set out in the report, appears representative rather than drawn to scale.

24 [69] The lower arc thus was the beginnings of what later came to be called the plausible range. The actual UBA cost could not be stated. But this conceptual exercise helped in providing direction for the optimal price, which was clearly above the expected value (in this example, the benchmark range median that the Commission had settled on in the draft determination). [70] Professor Vogelsang concluded: The objectives of ss 18 and 19 are fulfilled by a price certainly not below but possibly above true cost. In addition, the error analysis indicates that the price should be above expected costs. Furthermore, the clear value of Chorus s UBA costs appears to be above the median between the UBA cost of Denmark and Sweden. Combining these factors justifies a UBA price with the measured UBA cost of Sweden. Such a price at the 100 per cent mark would still be compatible with the requirement to stay within the benchmarking range. In my view it could even exceed this value. Since objective measures are not available, these statements are based on subjective probability assessments. Commission s Update Paper [71] Professor Vogelsang s report was attached to an Update Paper issued by the Commission on 13 August 2013.

25 [72] It dealt, first, with updates for the Danish and Swedish pricing. Costs and exchange rates were updated. An error in the Danish benchmark was corrected. The Commission indicated that it agreed with submitters arguing that a weighted average speed approach should be adopted. All of this resulted in updated prices of $10.92 for Sweden and $8.91 for Denmark. [73] The paper indicated that it would present the Commission s views on three matters: (a) The option of weighting individual benchmarks most comparable to New Zealand to derive the most likely forward-looking costs of the UBA service. (b) A proposed approach to deriving a plausible range of the forwardlooking costs of the UBA service. 32 (c) The application of s 18 to the choice of a price point for the UBA service. [74] The purpose of the paper was to seek further submissions on these points. [75] The Update Paper again set out the requirements of ss 18 and 19 (and again set out s 18 in full). The Commission indicated its continued preference to benchmark only against countries setting their prices using TSLRIC. That approach had been taken by the Commission in previous determinations for UCLL and mobile termination access services. Further submissions on that point were not sought. [76] The Commission was clearly agonising (as Ms Fitzgerald, who appeared for Telecom put it and I agree) over the fact that it was dealing with a small benchmark set. It recognised that a small benchmark set would increase uncertainty under IPP in determining the price that is an unbiased estimator of the forwardlooking costs of UBA service. And it would increase uncertainty under the IPP as to the plausible range of such costs. As the Commission saw it: 32 This was the first reference to the concept of plausible range.