Non confidential version Contents - Ofcom · 7 Non-confidential version 2. Background to regulation of physical access ‐ the backhaul bottleneck The backhaul bottleneck 2.1 It is

W W W . T O W E R H O U S E C O N S U L T I N G . C O M

1 Non-confidential version

Unlocking 4G ‐ Fixing the backhaul bottleneck

Paper for Vodafone

Non‐confidential version

Contents

1. INTRODUCTION AND SUMMARY .............................................................................................. 2

2. BACKGROUND TO REGULATION OF PHYSICAL ACCESS ‐ THE BACKHAUL BOTTLENECK ........... 7

3. HOW TO MAKE IT WORK ........................................................................................................ 18

4. THE BENEFITS OF PASSIVE ACCESS ......................................................................................... 29

5. CONCLUSIONS ......................................................................................................................... 40

ANNEX 1– REGULATION OF PIA IN TODAY’S POLICY ENVIRONMENT ............................................. 41

ANNEX 2 ‐ COMPARISON OF NGA TOPOLOGIES .............................................................................. 50

ANNEX 3 ‐ APPELLATE RULINGS ON RAMSEY‐EFFICIENT PRICING .................................................. 52

ANNEX 4 – PIA REQUIREMENTS EXPRESSED IN CM3 PER ADDRESSABLE HOUSEHOLD ................... 57


1. Introduction and summary

1.1 This paper is about ensuring that backhaul develops to meet the UK’s mobile broadband targets and that UK consumers receive high standards of data services at reasonable prices. We are proposing that physical access to BT’s network should be mandated for mobile backhaul. This represents an excellent, timely opportunity for the regulator to act to ensure that the potential benefits of 4G are harnessed effectively.

1.2 The UK has, historically, been a world leader in mobile voice and data.

1.3 Spectrum suitable for 4G services is now set to be licenced and cleared in spring 2013 following the UK’s 4G auction. Significant further activity is on‐going on the part of the UK Government, Ofcom and the European Community to make available considerably more spectrum in the next few years.

1.4 4G is rightly seen as crucial for the UK economy. Ed Richards, Ofcom’s chief executive recently said:

“The 4G auction will release crucial capacity to support future growth, helping to boost UK productivity, innovation and drive significant improvements to mobile broadband availability across the UK.”

1.5 Maria Miller, the new Secretary of State, shares those sentiments:

“Delivering 4G quickly is a key part of our economic growth strategy.”

1.6 However, releasing spectrum will not, of itself, drive innovation or deliver growth. Unless there is sufficient backhaul at an appropriate cost, actual speed will be throttled and much of the effort spent on the 4G auction will be wasted.

1.7 To put it another way: releasing more spectrum will simply exacerbate the bottleneck. Addressing the scarcity of spectrum must be accompanied by addressing scarcity in backhaul.

1.8 In this paper we argue that Ofcom should act to free‐up access to the links that join the radio segment of mobile networks to the core network and ultimately the internet and the rest of the world ‐ or there will be significant bottleneck problems over the next 3 – 6 years. This will mean poorer quality and more expensive mobile data services for end users in the UK.

1.9 The UK is particularly well‐placed to take advantage of a successful 4G implementation. The UK has highly‐competitive retail markets; a network sector characterised by two strong joint ventures and strengthened by the merger of Vodafone and C&WW; and a highly‐sophisticated consumer base, enjoying high


levels of smartphone penetration. With these advantages, in the correct environment, there is every reason to suppose the UK, in implementing 4G, can match other regulatory success such as fixed consumer broadband (through LLU)1.

1.10 Demand for mobile broadband in the UK is set to grow exponentially over the coming years. But provision of backhaul in the UK is already much less good than in other countries and this is leading to poor customer experiences. The situation will only get worse as backhaul bandwidth requirements increase unless action is taken.

1.11 This paper focusses on the most sensible and readily‐implementable solution to this problem: regulated access to passive elements of BT’s networks, to be used for mobile backhaul. The direct consumer benefits of this would include:

• increased mobile broadband speeds;

• greater coverage;

• improved customer experience; and

• lower unit data prices.

1.12 In addition, physical access will help to ensure that the technical capability of backhaul networks can keep pace with the technical advances made in mobile radio networks and that, crucially, the specific requirements of mobile networks (such as synchronisation) can be met in a timely way.

1.13 There is some urgency: mobile operators are planning and building their backhaul networks. Architecture choices will be locked‐in for years to come. Physical access remedies are required as soon as possible for the potential of 4G to be released. This matches the political imperative for the growth benefits of 4G to be harnessed quickly.

1 See Ofcom’s international communications market report, 13 December 2012: “Among the five largest European economies, the UK had the highest proportion (23.1%) of smartphone users visiting retailers’ websites on the mobile web… In the UK, 36% of respondents who owned internet‐capable devices used smartphones to access the internet in 2012, second in Europe only to Spain, where 43% did so…. Smartphone ownership is high in the UK, at 58%.”


1.14 We therefore consider that Ofcom should act as quickly as possible – ideally through the ongoing Business Connectivity Market Review – to implement appropriate PIA remedies for mobile backhaul.

1.15 In the rest of this paper we demonstrate why physical access is needed; how it would work in practice; we deal with some of the arguments against it; and show how a targeted remedy can produce the best possible outcome with the least possible cost.

1.16 Our conclusions are set out in section 5. We duplicate them here for ease of reference:

i. The government and Ofcom have highlighted the need for a quick 4G roll‐out to maximise economic growth opportunities. To this end, considerable amounts of spectrum are being auctioned in 2013 and efforts are being made to free up more spectrum still.

ii. There is clear evidence of a likely backhaul bottleneck (our phrase) or “capacity crunch” (Ofcom’s phrase) in mobile backhaul in the foreseeable future. This will put the success of 4G at risk;

iii. More spectrum will be to no avail if the actual service available to end users is constrained by backhaul.

iv. There is clear evidence that, where mobile operators can provide their own backhaul over self‐managed fibre, the service available to end users is significantly better.

v. BT’s existing backhaul products are unlikely to suffice because they will not allow mobile operators to have sufficient control and will not offer the right bandwidth to deal with the demands of 4G.

vi. PIA would provide an excellent resolution to the backhaul bottleneck.

vii. PIA has already been productised in the UK and the extra cost involved in making it available for mobile backhaul (and other uses such as serving enterprise customers) would be low.

viii. A modest amount of regulatory and quasi‐regulatory effort would be likely to deliver a usable, cost effective PIA product which could be used for mobile backhaul and other uses.

ix. We have, in addition, presented viable alternative remedies which would be focussed solely on the mobile backhaul bottleneck, including a dark fibre remedy.


x. The time to act is now. Mobile network operators are making technology and architecture choices which will be in place for some years. The sooner action can be taken, the more benefits will flow.


A note on terminology

The electronic communications sector is well‐known for its use of confusing acronyms and overlapping (or even contradictory) definitions of jargon terms.

To avoid confusion, here is what we mean by some of the main technical terms used in this report:

Mobile backhaul (or, in context, just backhaul) – high bandwidth connectivity which brings signals from a mobile base station to an operator’s core network. Sometimes this will include connectivity to and from an intermediate “point of concentration” which might also be a mobile mast.

Passive access – the ability for competing provider to use “passive” network elements in the incumbent fixed telco’s network. These include unbundled copper loops, ducts, poles, dark fibre or even trenches. Active elements, though, (electric or electronic) are not provided.

PIA – acronym for physical or passive infrastructure access.

Physical access – the same as passive access.

Active access – the ability for competing providers to buy wholesale access products from the incumbent fixed telco’s to the network with active electric and/or electronic elements provided by the incumbent fixed telco.


2. Background to regulation of physical access ‐ the backhaul bottleneck

The backhaul bottleneck

2.1 It is a truism that mobile data traffic is set to expand exponentially over the coming years. But the sheer extent of that growth and its wider implications for communications networks are often not appreciated.

2.2 Total UK internet traffic is expected to increase by an average of 37% every year between 2010 and 2015. Although the majority of traffic is currently on fixed networks, traffic on mobile networks is growing 84% year over year and is expected to account for 11% of total traffic by 2015, up from 2% in 2010.2

Figure 1 – forecasts for UK fixed and mobile traffic

2 http://www.atkearney.com/paper/‐/asset_publisher/dVxv4Hz2h8bS/content/the‐internet‐economy‐in‐the‐united‐kingdom/10192


2.3 Clearly any broadband solution to keep pace with the way consumers and businesses access the internet must involve a mix of both fixed‐line and mobile broadband.3 The House of Lords Select Committee on communications recently commented that:

“There is also a role for wireless technologies which can be used as an alternative to copper or fibre for the final link to the premises.”

2.4 In December 2012 Ofcom noted that mobile data usage had doubled in the previous 12 months and spoke of the “need to avert a capacity crunch”4. To this end Ofcom has embarked on a number of regulatory measures such as releasing of 700MHz frequencies for mobile use towards the latter part of this decade.

2.5 The anticipated growth rate in mobile data is, of course, based on a projection of uptake using our current understanding of 3G technologies and device use. It is not yet possible to take into account the effect of increased proliferation of smartphones and tablets, or the pace that mobile data use could grow at if 4G speeds become a realistic substitute for fixed broadband (as it is, for example, in the US5). One thing that is obvious is that given the growth in mobile data has already been outstripping fixed growth by a long shot, and that technological innovation in mobile sphere remains relatively untapped, growth in mobile broadband is potentially very significant indeed.

2.6 It is impossible to say exactly how fast demand for mobile data will increase but there is widespread agreement it will increase rapidly and growth will be fostered and amplified by innovations in mobile technology. Previous attempts to forecast bandwidth demand over the long term in fixed broadband have tended grossly to underestimate future demands6.

3http://www.parliament.uk/business/committees/committees‐a‐z/lords‐select/communications‐committee/news/governments‐broadband‐strategy‐risks‐leaving‐communities‐behind/ at para 24 4 Source – Ofcom Infrastructure report update and UHF Strategy, December 2012 5 See for example http://www.clear.com/plans/home/?intcmp=home:t1‐a:plans‐home 6 As recently as 2003 Ofcom classed 128kb/s as broadband, while recognising most households would want 256kb/s. A report of 2002 by the US firm Strategic Policy Research, whose author group included the former Chief Economist of the FCC, was able to suggest that “broadband (as with “Picturephone” before it!) may simply be a “niche” service that cost‐effectively addresses a relatively limited (although not as insubstantial as Picturephone) domain of economic wants. Given what it costs to supply (including the opportunity costs

2.7

2.8

2.9

of rispurchlimite7 See

9

So this tagrowth – extremely

Figure 2 – Pro

The risk isradio segmIt is this massive dslow backwill be fut

This is notbeing feltalready bethe way hbackhaul conferenc

sky venture cahase. The limited set of comm

e, for example, E

able – whichmay be too y rapid.

ojected growth in

s that the cament but wiwhich will idata throughkhaul connetile because

t a theoretic. The practieing markethome”) whiccan be pr

cing and sec

apital), only a ed size of the

mercial and gove

EE’s UK market

Non-c

h shows Ofcconservative

n UK mobile data

apacity of mthout a conn practice ghput capabilictions; goveend‐to‐end s

al problem acal effects oed as enablih, while it soovided7. 4Gurity are like

limited set of set of potentiaernmental appl

ting site for 4G:

confidential v

com’s range e; even thou

a demands

mobile netwocomitant incgive rise to aities will be ernment andspeeds will b

and the impaon users will ing mobile vounds good, G benefits iely to have

subscribers mal subscribers mlications—end

: http://explore

version

of estimateugh it already

orks will incrcrease in thea “capacity will be techd Ofcom effobe no better.

act on plannibe significa

video streamwill only won relation tvast impacts

ay find the semay permit ecoof story.”

e.ee.co.uk/4gee

es for mobiley suggests gr

rease dramae availabilitycrunch”. Cenologically horts to free‐.

ing engineernt. The 4G e

ming and HD ork effectivelto call relias on the stru

rvice sufficientonomic develop

e

e data traffirowth will b

atically in thy of backhauell sites withamstrung b‐up spectrum

ring is alreadexperience i gaming (“oly if sufficienability, videucture of th

tly attractive tpment of only

ic e

e l. h y m

y is n nt o e

to a


business of the way people do business, in particular in relation to time maximisation and freeing people from their desks and offices. Flow‐on effects to industries such as property, retail and payment systems are similarly uncharted but practically infinite.

2.10 For these reasons, Vodafone’s CEO has identified 1Gb/s backhaul per base station as a priority in the short term. Given the need to hub mast sites at intermediate points of concentration where the traffic of 3 to 5 base stations is aggregated, this in itself creates the need for passive remedies8.

The regulatory background9

2.11 Ofcom’s telecoms strategic review of 2004/5 resulted in the implementation of a policy which favoured competition at the deepest possible level –in other words, through passive or physical network access.

2.12 As a policy this has remained in place since then and has been spectacularly successful in relation to local loop unbundling (LLU).

2.13 It has been markedly less successful elsewhere. There are two reasons for this. First, it has not been seriously applied in most other product markets. Second, the passive access policy is not particularly suitable for the one area where it has been applied – fixed next generation access (NGA). Fixed NGA suffers from very high capital expenditure, relatively low intensity of use over that physical infrastructure, uncertainty over future demand and an uncertain premium given the existence of a lower cost copper alternative.

2.14 In contrast, PIA is well suited to mobile NGA. The point‐to‐point links needed for mobile NGA need to use comparatively less passive infrastructure and the demand (if not the timing of that demand) is clear. While fixed networks do not (apparently) support meaningful competition at the sub‐loop (the local network for fixed NGA purposes), mobile NGA networks, on the other hand either do support local NGA competition already, or will as a consequence of the rollout obligations in the 4G auction.

8 See, for example: http://www.lightreading.com/document.asp?doc_id=227068&f_src=lightreading_sitedefault

9 This section summarises arguments made in depth (with appropriate references) in Annex 1.


2.15 A regulatory intervention in mandating PIA for mobile backhaul is much more likely to be effective, and is much less intrusive, than the regulatory intervention which mandates PIA for fixed NGA. We have modelled the likely extent of the remedy required for mobile backhaul (using the proposals set out in Chapter 3) and compared it with the requirement for fixed NGA using today’s regulated “PIA for FTTC” product. We have expressed this as a cubic duct requirement in centimetres per addressable household. .

2.16 This exercise is inevitably crude but the results are so striking that we do not believe they can be wrong in directional terms. A fixed NGA roll‐out based on PIA is far more onerous (on BT) than our proposed mobile backhaul solution:

Technology Cubic duct required per addressable household

FTTC (using current products) 2,119 cm3

4G (using 10mm duct proposal) 53cm3

The result is that FTTC requires vastly more duct space than 4G.

2.17 If anything this very stark picture understates the extent to which our proposals represent a better regulatory intervention than the existing fixed PIA remedy, for a number of reasons. For example, our calculation assumes that mobile customers are static; it does not reflect the extra utility from mobility or the fact that additional customers (over and above the addressable households) can access the services in a given cell.

2.18 For all these reasons, MNOs are particularly well‐placed to use PIA remedies to backhaul its NGA networks. It would a highly‐targeted, non‐intrusive regulatory intervention with very significant benefits. In this context it is anomalous that the favoured PIA policy has been applied to fixed NGA networks (where it is not apparently suitable) but not to mobile NGA networks, where it is.

2.19 Our comparison schematic in Annex 2 maps the remedies we propose in Chapter 3 to the PIA remedies currently available for fixed NGA. It demonstrates that the two are very similar in terms of their place in the network.

2.20 In our view, therefore, Ofcom should apply its favoured regulatory policy to this area. This is necessary having regard to Ofcom’s duties to promote competition and act in a manner which is non‐discriminatory and technology‐neutral.


2.21 While we think the arguments for opening up physical access to a wide variety of uses is strong, we are sensitive to the need of regulators to be minimally interventionist where possible. In addition, we are not wedded to passive access for its own sake.

2.22 In Chapter 3.1v below (Product Definition and Downstream Uses) we demonstrate how physical access remedies are in practice less interventionist and set out a physical access remedy which would deal with the core problem we identify in this paper; included in that analysis are some targeted options which could still be effective.

Passive access is the effective remedy for mobile backhaul

2.23 Passive access has been a successful reality for mobile operators including Vodafone in several other European countries with extremely beneficial results for network performance and customer experience. As discussed above, passive access is not currently available in the UK for use with mobile networks.

2.24 Vodafone’s deployment of self‐managed fibre to cell sites in the UK is .

Figure 3 ‐ Fibre deployment

Figure 4 –

2.25 Figure 5 –

2.26 Figure 6 ‐

2.27 .

2.28

Figure 7 –

2.29 This phenomenon also has significant implications for future mobile demand. In order to meet customers’ increasing data demands Vodafone needs to reduce its opex per mb/s for base station backhaul.


2.30 . PIA (including potentially dark fibre) is an elegant and efficient way of giving customers a self‐supply option at the deepest level of competition10.

2.31 We appreciate that duct and dark fibre exhibits some natural monopoly characteristics; indeed some countries have explicitly recognised this such as Singapore, Australia and Sweden. Therefore, arguably, in a state of perfect knowledge (and perfect regulation of active products) there may be no need for PIA as BT’s cost structure would be known. However, there will always be an information asymmetry in favour of BT ‐ it can (and does) create many different flavours of different fixed access products which are optimised for its own network topology and needs. This is only a small (defensive) part of the story of the need for deep infrastructure access. The competitive process is underpinned by the ability of competitors to create new services. PIA enables access seekers to make numerous and significant improvements to efficiency including innovative and allocative improvements both at the service level and for product offerings. This means that customers buying from BT for other purposes, and ultimately consumers, are still disadvantaged vis a vis the PIA regulatory counterfactual.

Active access

2.32 Access to BT’s existing active products is not sufficient and ultimately much less effective at promoting competition than PIA. Active product tariff gradients will tend to be set in such a way as to favour BT’s downstream business and Ofcom’s basket structure allows them to do just that11. BT has no downstream mobile business and therefore equivalence / non‐discrimination obligations do not assist in any meaningful way. (BT does not have to “eat its own dogfood”, in the colourful characterisation used by an interested party in the Telecoms Strategic Review before the Undertakings were put in place). Ofcom assumes that BT will load common costs where it is most efficient to do so; but this ignores the strategic incentive to load costs against its largest enterprise competitors12.

10 See also the BCMR responses by O2 and EE/MBNL, available here: http://stakeholders.ofcom.org.uk/consultations/business‐connectivity‐mr/?showResponses=true&pageNum=1#responses 11 See Annex 3 for UK precedent on Ramsey Pricing 12 One example of BT’s pricing in relation to products not directly used by its own downstream business was in relation to PPC Points of Handover (CC case 1112/3/3/09) – see especially paragraphs 5.9 and 5.36


2.33 These incentives mean that BT products are often not suitable for use for mobile backhaul13. One instructive example is the requirement for timing information on backhaul links for synchronisation purposes. MBNL summarised the position well in their BCMR response of 3 September 2012:

One example here though is in the provision of timing information, which BT has taken since the launch of MEAS in 2008 to deliver. While BT states this will be delivered very shortly, at the time of this response, it has yet to deliver a truly synchronous Ethernet product. In a much shorter time period, MBNL has been able to work with VM in order to provide such a product, which it has now delivered.

2.34 BT’s active products are also comparatively expensive. Even based on today’s pricing, for practical purposes BT’s active products are more expensive than its PIA offering . This graph shows Vodafone’s current comparison (the (winning) PIA option being marked with a gold star)

2.35 With the measures we recommend in section 3 below, we would expect the effective price for PIA to reduce significantly, which would make the case even more compelling.

2.36 As will be clear, part of the problem here is with tariff gradients. BT’s active products are priced at traditional bandwidths (10mb/s, 100mb/s, 1Gb/s, 10Gb/s). This is of limited use for mobile backhaul where something more incremental is required between 1 and 10 Gb/s. This problem is of course exacerbated when the requirement for backhaul from concentration points, which hub signals from multiple base stations (and may themselves also be base stations) is taken into account.

2.37 Charge controls, by their nature, are an intrusive regulatory measure and therefore, as a matter of principle, should be used sparingly. There has been

13 Particularly for bandwidths above 1Gbit/s where Openreach’s OSA product is over engineered for the provision of last mile connectivity to mobile base stations

Figure 8 – backhaul cost by various solutions


significant criticism of Ofcom’s approach to regulating BT’s prices of charge control baskets, in particular in relation to BT’s ability to load common costs in leased line and ethernet service markets (both historic and as proposed in the BCMR consultation). The philosophical assumption underpinning Ofcom’s approach to setting caps and sub‐caps over ‘baskets’ of services is that, by allowing BT to decide how to recover its common costs by reducing prices for elastic services and raising prices for inelastic services, effectively cross‐subsidising between services within a basket, BT will choose a Ramsey efficient outcome.14 The impact on efficiency depends on the degree of pricing flexibility allowed and the extent that price rebalancing will improve allocative efficiency.

2.38 However, BT will have an incentive to focus price reductions on its own competitive services and not those services most used by its closest competitors. In addition the regulations in no way reflect the fact that prices are set for different bandwidths and services in a way that bears no relationship to cost. This has led to allocative inefficiencies, distorted demand in wholesale markets and subsequent downstream supply and prices for consumers in retail markets and a raft of regulatory disputes.

2.39 Furthermore, Ofcom has failed to recognise that services such as Alternative Interface (AI) access common costs, which consist primarily of access duct and fibre, are not shared with AI backhaul common costs, which consist primarily of duct and fibre between exchanges.

2.40 Charge control baskets also provide BT with the ability and incentive to distort competition in horizontal and vertical supply markets. In horizontal supply it can do this by under‐recovering on services where competition or a potential for competition on that service exists. Sub‐caps provide no protection against anti‐competitive below cost pricing. In vertical markets, despite the best intentions of equivalence of input obligations, raising the price for regulated wholesale products it supplies internally and externally has no net effect on BT at the group level yet will distort the cost structure of less integrated rivals. BT also has the ability and

14 See Ofcom’s leased line charge control consultation, 5 July 2012: “A broad basket would give BT some pricing freedom to determine the structure of prices which meet the charge control. This pricing freedom may be more likely to result in charges which recover costs, particularly fixed and common costs, in an efficient way.”


incentive to distort the pricing of services which are either not used by BT retail or are purchased in different proportions by external customers. BT can do this by reducing prices for services it purchases in relatively high proportion (giving it basket ‘headroom’) and increasing the price of those purchased in higher proportion to its competitors.

2.41 Without doubt, monitoring and enforcing charge controls and compliance with baskets is complicated and resource intensive. Access to physical infrastructure, which allows the communications provider to intercept the network further up the network supply chain significantly reduces BT’s ability to cross‐subsidise inefficiently. This is because upstream access enables communications providers (CPs) to determine which services within the ‘basket’ to self‐supply based on their own cost structures and therefore face none of these perverse incentives faced by BT. This provides CPs with greater freedom to develop new products and greater efficiencies. It will simultaneously eradicate the constant, numerous and costly operational arguments that currently arise with BT over service provisioning.

2.42 This makes it particularly difficult to understand why Ofcom considers PIA radical15. Ofcom also seems to contradict itself when on one hand referring to its preference for infrastructure competition at the deepest level possible16 then on the other voicing concern over PIA (in form of dark fibre) that it might have a chilling effect on existing investment in areas where there is infrastructure competition (in which it must inter alia be referring to LLU).17

2.43 BT is also incentivised to choke off the capabilities (at a given cost) of mobile internet access ‐ as it has most to lose from substitution within the retail market of mobile for fixed. Technological developments in the Radio Access Network and the release of more spectrum (800 MHz, 2600 MHz, 700 MHz and more) mean that it will be the cost of backhaul transmission which could prevent this real explosion in mobile data. Therefore this policy is contrary to Ofcom’s overriding duty to promote competition, not discriminate and act in a way which is technologically neutral.

15 BCMR, 8.133 16 BCMR A10.27 17 BCMR, A10.49


2.44 Ofcom specifically said in the consultation that it was worried PIA would mean BT loading more common cost on LLU and WLR‐ but that is exactly the point. It is favouring the interests of a particular business model over those of competition and fixed broadband over mobile broadband; the loser is the consumer.

2.45 The result is the further entrenching of BT dominance and a favouring of BT’s interests over those of consumers and, especially, the prospects for mobile networks to bring further benefits to customers.

2.46 It is important to note that in all of this we do not intend to accuse BT of any wrongdoing. Our point is simply that where incentives exist and the rules allow BT to act on them, it is unrealistic to expect them to do anything else; indeed, BT would not be acting in the interest of their shareholders if they did not act on them.

2.47 For mobile operators such as Vodafone to unlock competition in the provision of high speed data that the technology promises, self‐managed fibre backhaul is necessary. This will allow Vodafone to respond quickly to the demands LTE and LTE‐Advanced will impose. Self‐build fibre provides the platform that is essential to move away from leased wholesale providers that capitalise on offering middle mile support for commonly leased site access products e.g. BT Openreach. Access to fibre is key to a self‐build strategy and in the right geographical places will unlock the potential of high speed 4G networks and beyond.

Conclusion

2.48 In summary:

i. There is a clear and present danger of a backhaul capacity crunch.

ii. Active access is highly unlikely to meet the problem effectively.

iii. .

iv. PIA has been mandated in fixed NGA markets but, thus far, effectively overlooked by Ofcom as a solution in mobile NGA markets. This is a mistake because PIA is actually better suited to mobile markets.

v. In short, PIA is an obvious, simple and welfare enhancing solution to the mobile backhaul capacity crunch.


3. How to make it work 3.1 In this chapter we focus on the key factors required to make PIA work in practice.

We deal with six elements:

i. The regulatory rules in outline

ii. Price regulation

iii. Functional separation

iv. The practicalities of productising PIA for mobile backhaul

v. Targeting, part 1: product definition and downstream users

vi. Targeting, part 2: geographic aspects

3.2 Taken together, the right action in these six areas would deliver a highly‐useable PIA product set which would unleash data growth in mobile markets. We deal with them in detail below. In essence, the aim is a properly functioning regulatory regime, properly targeted and proportionate. There is no point in implementing a wide‐ranging, intrusive regulatory policy in the mere hope that it will prove successful. Rather, effort should be targeted at measures which are the most effective from a cost‐benefit perspective. There are three key elements to this which need to be balanced carefully:

i. Maximising efficiency by sharing fixed and common costs across a variety of downstream uses;

ii. Sharing the benefits of economies of scope and scale to ensure that no one player enjoys unfair advantages over another, thereby maximising dynamic efficiency;

iii. Ensuring that there are no perverse incentives which encourage short term or overly‐selective market entry (essentially, cherry‐picking) which may actually hinder sustainable competition and undermine other regulatory remedies.

3.3 This actually presents a reasonably complex balancing exercise which we address in section (e) below by proposing a series of regulatory remedies.

3.4 As an integral part of this activity, it is worth giving considerable attention to the practicalities of the chosen remedies. Inevitably there is some complexity in implementing PIA remedies. Unless this practical complexity can be overcome, there is a danger that perfectly valid remedies may go unused. We deal with this under section (d) below.

3.5 We turn now to our six considerations in more detail.


The right rules in outline

3.6 These paragraphs deal with the network elements which would need to be subject to regulation and the type of rules which ought to apply. They do not deal with price regulation (which is covered at (b) below), with arguments about limiting downstream use, or touch on geographic market definition. Again, those are dealt with below.

3.7 The precise network elements to be made available are discussed in section (e) below. Associated Facilities would need to be made available. These would include collocation (if necessary), handover (at the BT local exchange or via a footway box or street cabinet), break in, pull‐through, and so on.

3.8 The rules themselves would follow the standard approach for SMP conditions in fixed networks. In outline they would consist of:

i. An obligation on BT to provide the product;

ii. Obligation to offer fair and reasonable terms;

iii. Cost oriented pricing; and

iv. Prohibition on undue discrimination.

Price regulation

3.9 Benchmarking BT’s PIA pricing against the pricing available to Vodafone in other European markets strongly suggests that it is more expensive here in the UK than in countries where it is a more mature remedy .

3.10 Although BT’s existing PIA products are subject to a cost‐orientation term, they are not subject to explicit price controls. Experience in these markets suggests that sustainable entry is only viable where the regulator actually sets the prices.

3.11 In our view, therefore, it is essential that the starting prices are set by Ofcom with a clear eye on both actual costs and efficient engineering considerations. In other words, Ofcom should set a price based on the most efficient engineering solution rather than necessarily following BT’s actual practice. We would anticipate that this point would be most relevant to installation charges since the vast majority of rental charges will be accounted for by return on capital employed.

3.12 The availability of narrower sub‐duct categories than the 25mm which has been productised thus far will be essential for mobile backhaul users and will affect the price. In fact there should be a very‐nearly linear relationship between reductions in sub‐duct diameter and reductions in rental price.

3.13 For mobile backhaul, a sub‐duct of 10mm is appropriate. Clearly this is significantly smaller than the 25mm which is currently available from BT. A


subduct of 25mm may well be appropriate for (say) Gigabit Passive Optical Network GPON or other fixed NGA applications. For point‐to‐point mobile backhaul, employing two fibre pairs, 25mm is inefficiently wide.

3.14 Although some older engineering solutions require a 25mm subduct, there are perfectly good modern alternatives at lower diameters – for example, the Innerduct approach.

Figure 9 – comparison of sub‐duct types

Functional separation

3.15 Although the place of PIA in the structure set up by the Undertakings does need to be considered, it is our current view that there should be no equivalence of inputs requirement on PIA (beyond what is already set out for NGA).

The practicalities of productising PIA

3.16 When LLU was first productised – in the late 1990s, in advance of formal obligations being put in place – it was done in a way which made the product virtually unusable at scale. Leaving aside the price – which was also too high – the practicalities of using the product were so complicated that they acted as a strong disincentive to take‐up.

3.17 One of Ofcom’s first tasks was to correct this through the creation of the Office of the Telecoms Adjudicator (“OTA1”). This was a different creature from its successor, “OTA2”. OTA1 had binding dispute resolution powers – effectively as an arbitrator. This was given effect through multilateral contractual arrangements. The result was that OTA1 had real teeth and, while some stakeholders felt that more powers would have been useful, on balance this contractual remit struck the proper balance between (on the one hand) being significantly more than a talking shop and (on the other), impinging on Ofcom’s role.


3.18 OTA1’s remit covered LLU only and it had a set of clearly defined objectives. Although this limited role was doubtless a contributor to its success, it is noteworthy than a major criticism of OTA1 was that it had no power over backhaul. Ofcom reported as follows in its 2005 review of the scheme:

“Almost everyone argued that the absence of backhaul from the Scheme was a very serious drawback that impacted on the effectiveness of the LLU roll‐out”18.

3.19 This obviously reflects the importance of backhaul to broadband networks generally, and the importance of getting the practicalities of broadband right.

3.20 Something similar may be needed to enable PIA for mobile backhaul.

3.21 However, one of the great attractions of liberalising usages restrictions on the existing PIA product set is that it has already been productised. The fixed costs of productisation have already been incurred. Accordingly, it is to be hoped that there would be not much work to do; and little further cost. One exception to this would be the need to establish a productised variant of the produce for narrower diameters of subduct. In addition, in the context of mobile backhaul, there may be a need to deal specifically with issues arising in delivering services to non‐domestic sites (i.e. masts and towers).

3.22 This area will need to be examined in detailed field trials, which we would nevertheless hope could be completed quite quickly.

Targeting, part 1: product definition and downstream users

3.23 We are alive to the need for regulators to act proportionately. We have therefore given careful consideration to three separate kinds of remedy.

i. Remedy 1 – unrestricted PIA: in this package, the existing usage restrictions on PIA remedies would be removed; duct and pole access would be made available for any use, including fixed and mobile backhaul and also enterprise applications such as leased lines.

ii. Remedy 2 – restricted dark fibre: this involves a much more targeted approach. A new dark fibre PIA remedy would be created. However, it would only be made available for use as mobile backhaul.

18 Review of Telecoms Adjudicator Scheme for LLU, 25 October 2005


iii. Remedy 3 – virtual passive backhaul: in this package a virtual passive remedy would be created. As with remedy 2, it would be restricted to mobile backhaul. However, unlike the other remedies it would actually be an active remedy, consisting of unrestricted access to a single wavelength (λ) for an IRU period of 20 years but priced as a passive remedy19.

3.24 Each of these remedy options falls within SMP Markets and could be implemented comparatively easily in legal terms. In fact, there is likely to be a degree of freedom for Ofcom to implement these remedies through different market reviews – bearing in mind that a market review will very often cover multiple product markets and multiple remedies. The relevant designations currently in force will include:

i. Wholesale Local Access (WLA): market power designation in force by virtue of Ofcom’s statement dated 7 October 2010. This is the decision which currently mandates PIA. That mandate could be extend to cover the remedies proposed in this paper.

ii. Business Connectivity (BCMR): market power designation in force by virtue of Ofcom’s statement dated 8 December 2008. This is the decision which obliges BT to provide backhaul. That remedy could be expanded to include the remedies proposed here.

3.25 Both WLA and BCMR reviews are ongoing, with the BCMR work somewhat more advanced. Either of these workstreams could be used to progress the remedies we propose here.

3.26 We deal with these remedy options in turn.

3.27 Remedy 1 – unrestricted PIA: the great advantage of this remedy is that it maximises economies of scale and scope. Larger providers with multiple service offerings would be able to generate their own economies of scale and scope. However, smaller players would benefit too because BT’s fixed and common costs borne by PIA would be spread over a much wider set of downstream users.

19 For information, it is possible that this remedy could be implemented in a way which is consistent with the mooted WDM PON remedy for implementing unbundling over GPON networks. In the current context, we are not likely to be talking about areas where GPON networks have been rolled out. So we have not considered this remedy directly. This could change.)


3.28 Assuming reasonable take‐up by the two main mobile network‐sharing entities (MBNL and Beacon (including C&WW)) and the two main mass‐market LLU‐based competitors (TalkTalk and Sky) there is the genuine possibility that this solution would achieve significant takeup. It is a very important part of this picture that UK communications markets are now relatively concentrated; there are at least 4 players with very significant economies of scope and scale. In addition, it is clear from other BCMR responses (e.g. EE, MBNL) that many of the scale players would be keen to use passive remedies in this way. This makes Remedy 1 viable in a way which would not have been the case even a couple of years ago.

3.29 One apparent difficulty with this remedy is that it relies on at least some competitors right across downstream markets using it at scale. Otherwise BT would be the only player to benefit from economies of scope and scale. Users of Remedy 1 would suffer from their ongoing scale disadvantage – they would be in a less good position than BT. In other words, self‐supply using Remedy 1 could be more expensive than buying a regulated product from BT downstream. Although this sounds worrying, in fact it is not a real practical problem. The costs of productising PIA have already been incurred. So if Ofcom were to mandate Remedy 1 and no‐one used it because they felt they could not generate the necessary economies, there would be at most minimal loss.

3.30 Thus the dynamic efficiency benefits of Remedy 1 are potentially very significant indeed. Conversely, this remedy is likely to be of little use if downstream uses are restricted because economies of scope and scale would be lost.

3.31 The relevant product definition for Remedy 1 is likely to look something like this:

Network Access comprising predominantly of the provision of space, anchorage, attachment facilities and/or such other facilities as may be reasonably necessary to permit a Third Party to occupy parts of the Dominant Provider’s Physical Infrastructure located between Network Termination Points and Local Access Nodes serving those Network Termination Points, and/or from the Dominant Provider’s local access Nodes to the Third Party’s own network sufficient to facilitate the establishment, installation, operation and maintenance of the Electronic Communications Network of a Third Party at that location. The uses to which such Network Access may be put by the Third Party are not to be restricted provided it is used for the provision of any electronic communications service.

3.32 Diagrammatically, the configuration of unrestricted PIA for mobile backhaul would look like this:


Blue lines – own network Green lines – microwave / traditional backhaul Red lines – Vodafone self‐managed fibre using unrestricted PIA product

Figure 10 – unrestricted PIA

3.33 . It is important to keep this in context – the need here is for comparatively short links in a fairly narrowly defined part of the network.

3.34 Remedy 2 – restricted dark fibre: this option is more restricted than with Remedy 1. However it does have some obvious attractions. First, it is ideal for a regulator who likes the idea of freeing up mobile backhaul markets but remains unconvinced by the benefits of Remedy 1. Unlike in Remedy 1, BT itself benefits from economies of scope and scale in use of the duct infrastructure and would share those benefits through the pricing of access to dark fibre. The access‐seeker would only need to have sufficient economies of scope and scale to make efficient use of the fibre for this option to be valuable. Limiting downstream use to mobile backhaul, therefore, though still sub‐optimal is less likely to undermine the remedy. The relevant product is likely to be within the wholesale local access market. However it could also be brought within the scope of the BCMR and could therefore be implemented quickly.

3.35 The product could be described thus:

Access to Dark Fibre and such Associated Facilities as are necessary to enable the mobile network provider to self‐provide:

i. backhaul from mobile base stations to the mobile network provider’s core network; and/ or

ii. backhaul from a local point of concentration (aggregating traffic from multiple base stations) to a mobile network provider’s core network.


3.36 Represented diagrammatically, the configuration would look like this:

Blue lines – own network Green lines – microwave / traditional backhaul Red lines – restricted dark fibre product

Figure 11 – restricted dark fibre

3.37 Remedy 3 – virtual passive backhaul: virtual passive remedies were first described in our paper of March 2011, “Wholesale Pricing for Next Generation Access Networks – A New Approach”. The approach in outline is simple: if access pricing for active products involves low (or zero) marginal cost for incremental customer (or data) volumes, and access‐seekers enjoy a high level of control over the service, an active product can be made to generate exactly the same dynamic efficiency benefits as a passive product. Backhaul lends itself particularly well to this solution because:

i. A lambda is very nearly a passive product in the sense that it allows the user a very high degree of control over the service; and

ii. The IRU (or indefeasible right of use) is a well‐known and accepted commercial construct in the industry already.

iii. Thus all that would be required for this solution to be productised would be a new price structure and remote control of the multiplexer to be given to the MNO. The new price structure would, like PIA, consist of a significant upfront payment and little or no ongoing charges.

3.38 The product definition for such a product would look something like this:

Access to optical spectrum wavelengths and such Associated Facilities as are necessary to enable the mobile network provider to self‐provide:


backhaul from mobile base stations to the mobile network provider’s core network; and/ or

backhaul from a local point of concentration (aggregating traffic from multiple base stations) to a mobile network provider’s core network.

3.39 The pricing could be set separately by direction in accordance with the principles outlined above.

3.40 The product is likely to be within the BCMR and could therefore be implemented quickly.

Targeting, part 2: geographic aspects

3.41 It will be important to target the remedy where it is needed in a geographic sense as well. There is some evidence that there are varying degrees of competition in the very high bandwidth market across the country. BT has argued (according to Ofcom) that “some important users of very high‐bandwidth services choose to locate in areas where competing networks are already present.”

3.42 We remain to be convinced by this argument but for the purposes of the rest of this section we assume that there is some prima facie evidence which merits further investigation.

3.43 There are various implications of this line of argument.

i. First, BT’s argument may (or may not) hold true for the enterprise market. It axiomatically does not hold true for backhaul markets, which display entirely different characteristics. BT claims that enterprise users may cluster in areas which already have dense fibre networks. This is, of course, not physically possible for backhaul, because demand is driven by the requirements of mobile network topology20: it is not possible simply to locate all the cell sites of a mobile network in areas where there happens to be plenty of fibre. One of the blocks of 800 MHz spectrum in the forthcoming spectrum auction carries obligations of 98% mobile coverage with a minimum of 95% in each UK nation. Separately, Vodafone and Telefonica have pledged to meet 98% indoor mobile coverage and EE is

20 (Or, in the case of fixed backhaul, the location of BT’s own local exchanges)


targeting this level of outdoor coverage. Taken together, these plans are clearly going to result in cell sites very widely distributed across the UK.

If there is anything in BT’s argument, it strongly implies there is a separate product market for backhaul.

ii. Secondly, there may be a case either for identifying separate geographic markets or

iii. Thirdly, there may be a case for applying geographically‐targeted remedies within a national market.

3.44 Options (ii) and (iii) would be likely to look fairly similar in practice and a choice between them would probably depend on the degree of geographical variation in competition. We return to this below. Both options were specifically recommended for PIA in the EC’s NGA Recommendation (2010/572/EU, Recital 9)21:

NRAs should define sub‐national geographic markets in accordance with Commission Recommendation 2007/879/EC of 17 December 2007 on relevant product and service markets within the electronic communications sector susceptible to ex ante regulation in accordance with Directive 2002/21/EC of the European Parliament and of the Council on a common regulatory framework for electronic communications networks and services ( 1 ) if they can clearly identify substantially and objectively different conditions of competition which are stable over time. In situations where it cannot be concluded that the different competition conditions would justify the definition of sub‐national geographic markets, it could nevertheless be appropriate for NRAs to respond to diverging competitive conditions between different areas within a geographically defined market, for instance due to the presence of several alternative infrastructures or infrastructure‐based operators, by imposing differentiated remedies and access products.

21 The NGA recommendation applies to physical infrastructure access; but in a slightly different context. Nevertheless, the principles apply by analogy.


Taking the NGA recommendation as a starting point, it is worth considering how it might be applied in practice in this context. One possible approach would be to follow the type of analysis undertaken by Ofcom in the WBA market review22.

3.45 Applied here, this kind of approach might involve the following stages:

i. Identify total number of mobile sites where backhaul is required

ii. Identify sites with competitors (“Type A sites”) – how competitors many at each?

iii. Identify sites without competitors (“Type B sites”)

iv. Argue by analogy with WBA that 1 or 2 competitors does not equal competition – therefore there is probably either a national market or two separate markets – one with competitors and one without;

v. In either event, apply the PIA remedy to Type B sites only. This approach maximises competition downstream and does not preclude entry

3.46 It is not entirely clear what the outcome of this analysis would be – and indeed we are not entirely convinced that it would be necessary – but if there were prima facie evidence that such an approach was warranted, it would not be unduly burdensome.

Conclusions

3.47 In this section we have shown that the challenges to implementing PIA for mobile backhaul are, in practice, quite limited. Targeted regulatory and quasi‐regulatory intervention would serve to fix these comparatively limited price and product issues. Although there are good welfare‐enhancing reasons for liberalising the use of PIA across the board, a more limited focussed remedy would still be an effective way to fix the mobile broadband backhaul bottleneck.

22 Review of the wholesale broadband access markets ‐ Statement on market definition, market power determinations and remedies, 3 December 2010


4. The benefits of passive access

4.1 Broadband is central to the lives of consumers and success of UK businesses23 and ‘unlocking its potential’ is fundamental to achieving the Government’s Broadband Policy and Digital Agenda targets. For a consumer, high speed fixed broadband (Fibre to the Cabinet or Fibre to the Home) and mobile broadband enabled by 4G, though not necessarily perfect substitutes in a technical sense, may well look and feel very similar24. Supporting that growth and making it work in practice is a very significant challenge.

4.2 At the retail level, in other words, mobile broadband is both complementary to, and may ultimately be a substitute for, fixed‐line broadband and competes for customers with LLU and fixed‐lines at current speeds. In the early stages of regulation in the UK, faced with the need to simply ‘deliver broadband’ and meet its obligations and objectives, government and regulators required an approach that accounted for market conditions at the time, in particular market structure and investment choice. This reflects the ‘lumpiness’ of investment patterns which are characteristic of communications investment due to economies of scale and scope.

4.3 It is widely accepted that in communications markets parts of the incumbent’s copper and fibre networks exhibit natural monopoly characteristics – such as duct and fibre. That is, that there are great economies of scope and scale, excess capacity and, in the absence of regulation, the incumbent has the ability and incentive to raise access prices above the competitive level which may cause bottlenecks and inefficient supply to occur.

4.4 In the early days of denationalisation of communications networks and competition policy regulators faced a trade‐off: the full benefits of competition are only achieved by facilities based competition, yet at that point service based competition was the only viable option for would be competitors. However service based competition has a foreclosing effect on facilities based competition as it entrenches reliance on the incumbents infrastructure.

23 WBA review – statement, paragraph 1.1 24 See, for example, Capital Economics report of April 2012, Mobile Broadband and the UK Economy, which equates superfast fixed with 4G, and positions 3G as sub‐ADSL performance


4.5 The “ladder of investment” (“LOI”) theory was an early and widely‐embraced policy designed to stimulate entry and competition in communications markets.25 This approach dictates that regulators encourage entry by granting access to progressive “rungs” of the incumbent’s facilities by setting access prices at a level that enables them to offer products similar to those of the incumbent without having to reproduce that level of the facility, akin to simple resale. Once access seekers gain sufficient footing they are then be encouraged to move up the “ladder” by investing more in the network elements necessary to bypass the previous level of access, and offering products that are increasingly differentiated and service that is more efficient than those offered by the incumbent.

4.6 This migration up the LOI can be facilitated by lowering the increment between the prices of the lowest‐rung access products and those higher up the ladder. Ultimately, once competitors have gained sufficient revenue base, scale and technological sophistication, they may even invest in their own access infrastructures.26 Tracing back the ladder theory to its logical presumption is that facilities/inter‐platform based competition results in better outcomes than service/intra‐platform based competition.

4.7 Thus the debate surrounding communications access regulation has three aspects: whether to unbundle networks, on what terms (e.g., pricing) to offer the unbundled product, and where in the network to unbundle (e.g., at the exchange or the street cabinet).27 Accepting that deeper levels of access result in deeper competition between rivals, it follows that the choice between such passive and active access remedies should reflect these outcomes.

4.8 That is, regulatory preference, based on LOI, between passive and active access regulation is dictated by the ultimate net benefits that will follow from each; competition between supply methods and technologies (inter‐platform enabled by passive access) will always provide superior results for competition, in particular in relation to transformative technologies and price, when compared with competitors using the same supply methods and technology (intra‐platform

25 Cave, Martin, “Encouraging Infrastructure Competition through the Ladder of Investment”, Telecommunications Policy, Volume 30, April/May 2006. 26 Access Regulation and Infrastructure Investment in the Telecommunications Sector: An Empirical Investigation, LECG report for ETNO, 2007, paragraph 3.2 27 LECG report


enabled by active access). In the current regulatory debate, this is reflected in the distinction between fibre and mobile broadband delivery and LLU, using copper wire or even further down the ladder, LLU and wholesale broadband access.

4.9 However given the topology of the network, a degree of active access regulation offering entrants access to incumbent networks at regulated prices utilising the incumbent’s last mile infrastructure is likely to always be needed.28 This has clearly been a driver behind the unprecedented success of LLU competition in the UK which has resulted in better quality and lower priced broadband speeds for consumers using ADSL technologies up to the theoretical ceiling of 24 Mbit/s.

4.10 However, whilst active regulation of copper wire (LLU) broadband has been integral to delivering an early competitive broadband market in the UK, ADSL/LLU is simply not capable of providing access across all of the UK to data at the speeds required to meet the Governments broadband targets and objectives or the demands of end‐users consuming multiple HD TV channels. Notwithstanding this, lower LLU prices may cause consumers to substitute away from fibre, which can, in turn, stymie investment.29 Continuing along this glide path, the UK may end up in a position where it has a very cheap but relatively slow broadband network but has failed to deliver the BDUK’s next generation broadband speeds and targets.

4.11 Somewhat curiously though, in the context of dark fibre, Ofcom considers that any additional remedy should not undermine existing infrastructure competition.30 This position is difficult to reconcile with practically every other policy. If passive access happens to leave LLU infrastructure equipment “stranded” due to consumers substituting away from it, how on earth can this be thought of as anything but a policy success, and meeting the regulatory objectives?

4.12 The solution is obvious: communications providers must be enabled to gain better access and greater access to fibre, for both fixed and mobile broadband.

28 Access Regulation and Infrastructure Investment in the Telecommunications Sector: An Empirical Investigation, LECG report for ETNO, 2007, paragraph 1.6 29 Ibid, paragraph 1.18 30 BCMR, A10.56


The options for fibre

4.13 For mobile operators the solution is obviously a downstream or ‘backhaul’ story. However, as mentioned above in Section 3, in this context a solution that is currently focussed at fixed networks has also the potential to resolve the issues for mobile. Accepting ‘fibre’ is the answer; the question is therefore on what rung of the ladder should it be unbundled by the regulator? Above we have established that passive access is the economically sound default intervention preference to promote competition and is most likely to achieve regulatory and Government objectives. What is more, the drivers are self‐evident – much of the work has already been done to productise PIA, and in light of international evidence, market maturity, technological advancement and exponential growth in data use the risk of not introducing PIA far outweighs the risks involved in Ofcom doing the work to introduce it and no‐one using it.

The arguments against passive access infrastructure

4.14 The elephant left in this room is therefore – why, in this country, is the approach to the regulation of broadband restricting the ability of access seekers to decide whether they consider it is economically viable to invest in fibre and whether to gain access to the incumbent’s network at the deepest level possible? Below we set out Ofcom’s most recent rationale for not adopting passive access remedies, as set out in the BCMR, and our response to each in turn.

Duplication of expenditure

4.15 Ofcom states: “Introducing passive remedies would also potentially add costs of competition. While passive remedies would avoid or reduce the need for BT’s competitors to invest in building physical infrastructure, those competitors choosing a passive remedy would nevertheless incur additional costs in network infrastructure. The investments would include the costs of purchasing, installing and managing active equipment and, in the case of PIA, the costs of purchasing, installing and managing fibre in BT’s ducts. The investments would, to some extent, duplicate BT’s, and would therefore add to the cumulative costs of the industry. Models developed as part of our review of the WLA market suggest that these additional costs could be significant. In the case of NGA investment using PIA,


the cost per end‐user with four competing networks was modelled at more than double that with just one network.”31

4.16 This might be so if the costs to the end user are in fact the underlying costs of the provision – but where the costs to the end user are BT’s scaled wholesale costs, then the argument is entirely vitiated. Vodafone and other potential users of PIA have requested the provision of a PIA remedy, confident that the actual costs of provision of high capacity but scalable bandwidth are far below those charged by BT. We have stated that PIA would enable us to invest in further network roll out ensuring wider and richer data coverage for UK consumers. Given this evidence it is not clear how Ofcom can so easily claim that such investment would not be efficient nor is it evident that this is a relevant consideration. It is not Ofcom’s role to prevent access seekers from making bad investment decisions and, notwithstanding this, it is not in a position to make this judgment.

4.17 Similarly, Ofcom appears to put the onus of proof upon respondents, stating:

“We remain open to any evidence that shows that NGA investment could be unlocked by being able to use PIA for leased lines services, which could help us to formulate our policy in this area.”32

4.18 In our view, the proper approach is for Ofcom to create the conditions for investment by putting in place the appropriate regulatory environment:

“…the NRA will need to be careful not to second‐guess the market place but rather should provide a coherent background against which market developments take place.”33

4.19 Only then, and after a suitable period, can it judge the impact of unlocking of NGA investment. Vodafone is convinced from its experience that a favourable impact for customers is very likely, particularly given the known need for high capacity mobile backhaul. Ofcom is too quick to dismiss the possibility of benefits to the national economy from the additional investment.

31 BCMR 8.60 32 BCMR 8.89 33 Review of the Wholesale Local Access market, Ofcom, 2010, at 9.42 http://stakeholders.ofcom.org.uk/binaries/consultations/wla/statement/WLA_statement.pdf


4.20 Moreover, this argument can only be properly assessed if one also considers the beneficial impact on competition and services. The principle of additional cost of duplicative investment can be asserted in relation to any non‐monopoly, for example in the case of the mobile industry, but that is only half of the story, given the potential benefit at the wholesale and retail levels of competitive investment. Ofcom does not appear to make any effort to consider the consumer benefits that could be unlocked by PIA.

Economies of scale and scope

4.21 Again, Ofcom presumes the outcome and then relies upon that supposition to decide it does not need to encourage competition in this area. It states:

“Introducing passive remedies would also carry significant risks. Investment in fibre‐based networks is subject to strong economies of scale, and, while passive remedies could reduce barriers to competition based on infrastructure, any such additional competition they stimulate may not be sustainable outside some dense geographic clusters of businesses, such as major urban centres.”34

4.22 Ofcom virtually guarantees this outcome by restricting PIA to fixed NGA access thereby preventing all operators other than BT from achieving the natural economies of scope and scale available. This point was recognised by the House of Lords:

“Lifting the restrictions on PIA could help provide other infrastructure providers with that breadth [of different revenue streams] and thereby, in our view, play a significant role in introducing more competition at the level of the access network.”35

4.23 .

Recovery of common costs

4.24 Ofcom states in this connection:

“…introducing passive remedies in business connectivity markets could have wider implications on the recovery of common costs that underpins the current pricing of

34 BCMR 8.61 35 House of Lords Broadband Report


all of BT’s regulated products. Extending the example above, BT may respond to competitive entry based on passive remedies by reducing its charges for higher bandwidth services. This may, in turn, require rebalancing of the recovery of BT’s common costs, which may lead eventually to an increase in its charges for other regulated services, not only for those used in business connectivity markets but potentially also for others, such as local loop unbundling and wholesale line rental.”36

4.25 “One possible factor in MNOs’ concerns is that BT currently recovers

proportionately more of its common costs from its higher‐bandwidth regulated products, which tend to be purchased by large organisations, by MNOs and by operators of fixed broadband services, including BT itself. In allowing BT to determine, within appropriate constraints, how to allocate its common costs between the regulated services subject to a charge control, we have previously considered that BT has access to better information than we have to do so, and incentives to achieve outcomes consistent with economic efficiency.”37

4.26 This is a curious argument as Ofcom is perfectly content in other contexts to prevent the recovery of any common costs from particular regulated services (e.g. Mobile Termination Rates). In this case, it uses the threat to BT’s recovery of a “disproportionate” amount of its common costs from other regulated services as a reason not to intervene.

4.27 It is equally likely that BT’s disproportionate recovery of common costs from high bandwidth services is not a sign of economic efficiency, but rather a manifestation of its market power. Faced with increasing customer demand for data and a need to have higher bandwidth transmission further out in their networks, mobile operators in particular have few other choices but to contract with BT, given their inability to use PIA.

4.28 Furthermore Ofcom has failed to address the issue of pricing for routes where the traffic demand is greater than 1Gbit/s – it is here that the contrast in pricing between two or more 1Gbit/s Ethernet links and a single PIA link is most acute,

36 BCMR 8.63 37 BCMR 8.82


and where the loading of operating costs (and to a lesser extent capital costs) purely by scale is most obviously incorrect.

Alternatives to BT Ethernet and PIA for high bandwidth mobile sites

4.29 There are no obvious alternatives open to the mobile operators for backhaul supply to high capacity mobile sites other than BT Ethernet and PIA. There are limited alternative suppliers of fibre other than BT, particularly for the last mile, and outside a few urban areas.

4.30 It is feasible today for microwave technology to reach up to 400Mbit/s throughput. A significant problem however is the lack of spectrum availability in some frequency bands below 20 GHz. Microwave links that can provide more than 400 Mbit/s of transmission capacity have appeared recently in the market. They are placed in the upper part of potential microwave spectrum bands (> 60 GHz): a clear example is the E Band (at 80 GHz). As soon as several spectrum management issues are resolved by Ofcom it is likely that E Band microwave rollout will occur38. However the biggest limitation of this band is the maximum hop length, which is less than 3 kilometres.

4.31 Higher capacity microwave is thus no panacea: it will have its place in a mobile operator’s backhaul portfolio, for example for the last mile for perhaps suburban or semi‐rural cell sites where traffic loads are relatively high, but for longer or higher capacity routes it will be an ineffective and impractical substitute for fibre. This considerably reduces its overall attractiveness in comparison with fibre.

Technical development of MEAS

4.32 Ofcom states:

“We note that MNOs’ broader concern that their future backhaul costs could escalate unduly may be addressed to some extent over the next few years by a combination of technical development, which should enable the effective bandwidth delivered by BT’s MEAS product through Openreach’s 1Gbit/s Ethernet access tails to increase substantially, and by the operation of any

38 See, for example, the Ofcom Spectrum Review update of 13.12.12 http://stakeholders.ofcom.org.uk/consultations/spectrum‐review/update/?utm_source=updates&utm_medium=email&utm_campaign=point‐to‐point


charge controls agreed in relation to leased lines following conclusion of this review.”39

4.33 As welcome as this development is, it does not address the fundamental concern regarding the cost/bandwidth structure of BT’s prices and the fact that BT’s customers are less able to innovate as a result of being dependent upon BT40. Complete end‐to‐end control of the backhaul network increases the flexibility and autonomy of operations. For the mobile network operator it means better network performance, faster network deployment, the absence of dependencies with third parties, flat growth costs etc. because the network and the topology is aligned with its fundamental purpose, i.e. to efficiently link mobile cell sites with the rest of the network and to support mobile services in the circumstances of sustained high traffic growth.

4.34 At 8.62 of the BCMR Ofcom makes the point that:

“While current regulations continue to apply to existing services, the opportunity for BT’s competitors to use a passive remedy could be most attractive when delivering high‐bandwidth services, not because those competitors could necessarily do so more efficiently than BT, but because of the way that BT had decided to recover its common costs.”

4.35 This is entirely the point. The current pricing policy of BT with a low initial charge

but a high on‐going annual charge, and prices scaled to bandwidth bears no relation to the underlying costs of provision, and is an inhibitor to the future deployment of mobile broadband.

4.36 The necessary flat cost curve could be achieved by changing the structure of

Ethernet prices, the use of long lease dark fibre, and by PIA. These solutions are complementary, rather than mutually exclusive since this will add to the level of

39 BCMR 8.84 40 See, for example, MBNLs’ BCMR consultation response of 3 September 2012: “One example here though is in the provision of timing information, which BT has taken since the launch of MEAS in 2008 to deliver. While BT states this will be delivered very shortly, at the time of this response, it has yet to deliver a truly synchronous Ethernet product. In a much shorter time period, MBNL has been able to work with VM in order to provide such a product, which it has now delivered.”


competition for such links, enabling the development of a more cost reflective pricing structure for each solution, and thus removing the potential bottleneck to the supply of high capacity backhaul services for mobile operators.

Section conclusion

4.37 The government’s Digital Agenda target to deploy a faster fibre network to rural areas by 2015 has recently been described as "a challenging target". A key issue with BDUK may be traced back to the fact that there is an absence of alternative options to BT as it is the only preferred bidder to receive any grants to roll out fibre. The absence of a good PIA product has been suggested as a conspicuous reason for the absence of alternative providers in a position to deploy fibre and win BDUK contracts.

4.38 In order to meet the increasing data demand of UK customers, mobile operators need to reduce their operating expenditure costs associated with base station backhaul. Improvements are needed in both middle mile and last mile to achieve that. There is a clear and significant risk that that the provision of backhaul links between the cell sites and the core network of a mobile operator could become a bottleneck in the delivery of mobile services, particularly of data, in the next few years.

4.39 There is clear empirical evidence available that duct sharing has proved successful for Vodafone Group .

4.40 Technology is driving demand for data in communications markets where regulations are simultaneously stifling innovative and efficient supply of products and services. This creates substantial knock on effects to the wider economy and consumer welfare. Aside from getting left behind other European countries that are in a position to capitalise on the clear opportunities that higher bandwidth technology creates, failing to act now will also severely risk meeting the Government’s broadband targets. Oddly, the regulations Vodafone and other communications providers are seeking are permissive in nature and the costs of adapting the PIA product set to use for mobile backhaul would be minimal; even if no‐one uses it, nothing is lost. The likely set of users will be small so any remaining practical issues should be soluble with consensus (especially if assisted by OTA). The costs if no‐one bought the product (which is unlikely), conversely, would be very low. In all there is little to lose but a lot to gain.

4.41 There are no valid reasons why Ofcom should be ‘second‐guessing the marketplace’. The theoretical and empirical background to access regulation shows that the time is right to open up downstream markets to access seekers and allow the market to supply through PIA in response to consumer demand for


bandwidth through a PIA remedy if it chooses; and Vodafone for one will. This will simultaneously and effortlessly drive growth and deliver the UK’s targets and obligations in relation to its broadband policy agenda.


5. Conclusions

Our conclusions in outline are as follows:

i. The government and Ofcom have highlighted the need for a quick 4G roll‐out to maximise economic growth opportunities. To this end, considerable amounts of spectrum are being auctioned in 2013 and efforts are being made to free up more spectrum still.

ii. There is clear evidence of a likely backhaul bottleneck (our phrase) or “capacity crunch” (Ofcom’s phrase) in mobile backhaul in the foreseeable future. This will put the success of 4G at risk.

iii. More spectrum will be to no avail if the actual service available to end users is constrained by backhaul.

iv. There is clear evidence that, where mobile operators can provide their own backhaul over self‐managed fibre, the service available to end users is significantly better.

v. BT’s existing backhaul products are unlikely to suffice because they will not allow mobile operators to have sufficient control and will not offer the right bandwidth to deal with the demands of 4G.

vi. PIA would provide an excellent resolution to the backhaul bottleneck.

vii. PIA has already been productised in the UK and the extra cost involved in making it available for mobile backhaul (and other uses such as serving enterprise customers) would be low.

viii. A modest amount of regulatory and quasi‐regulatory effort would be likely to deliver a usable, cost effective PIA product which could be used for mobile backhaul and other uses.

ix. We have, in addition, presented viable alternative remedies which would be focussed solely on the mobile backhaul bottleneck, including a dark fibre remedy.

x. The time to act is now. Mobile network operators are making technology and architecture choices which will be in place for some years. The sooner action can be taken, the more benefits will flow.


Annex 1– regulation of PIA in today’s policy environment Background to the regulatory scene

A1.1 In this section we elaborate on the arguments in chapter 2.

A1.2 Passive and active access policies are the primary regulatory interventions used in the communications sector to promote competition and economic growth. Communications providers are given access to the incumbent’s network to unlock the potential of network infrastructure where it is uneconomic to reproduce and there is excess capacity.

A1.3 Both options have formed part of Ofcom’s regulatory policy toolkit since its inception in 2003. In 2009, in Ofcom’s Statement on “Delivering super‐fast broadband in the UK” it recognised:

“Passive products form the main basis for competition in current broadband and their presence has arguably been an essential ingredient in the success not just of competition but of the market as a whole.

One of our core principles in regulating telecoms is to establish competition at the deepest level that is effective and sustainable. It is clear that passive products represent the deepest level…”41

A1.4 Consumer access to internet data at higher speeds underpins user demand and drives innovation and growth across the sector and the UK economy. It was estimated that in 2009 the internet accounted for over 5% of the UK’s GDP, and a far larger proportion of total economic growth (23%) over the past five years42.

A1.5 We do not presume that the regulatory choice of passive over active access is welfare optimising in all circumstances, and so we have come to this paper with no pre‐conceptions. Our starting point in principle is that the regulatory choice between passive and active access should be driven by the net benefits that will accrue to consumers from each in the particular circumstances. Ofcom, however, has expressed a policy preference for passive access.

41http://stakeholders.ofcom.org.uk/binaries/consultations/nga_future_broadband/statement/statement.pdf 42 “Internet matters: the Net’s sweeping impact on growth, jobs and prosperity”, McKinsey Global Institute May 2011 at pages 15 and 16


“Ofcom’s competition policy aims to promote infrastructure competition at the deepest possible level. This descends from the EC Framework’s view of introducing infrastructure‐based competition as the main remedy to market dominance by the incumbents, and follows the belief that real innovation can only come when different infrastructures compete with each other”.43

A1.6 We agree that if markets operate consistently with accepted regulatory theory, this leads to the conclusion that passive access should result in better outcomes for competition and consumers and therefore agree with Ofcom’s preference for passive access.

A1.7 As Ofcom reflects:

“…each level of intervention determines the extent of competition: the deeper the level of regulatory intervention, the deeper the depth of competition…”.44

A1.8 When faced with natural monopolies, regulatory theory dictates that inter‐platform access(passive access), which promotes facilities‐based competition and investment, will always provide superior results for consumers in relation to transformative technologies, efficiencies and price competition compared with competition between suppliers buying wholesale products from an incumbent utilise a single supply technology (intra‐platform/active access). Where inter‐platform competition is not viable, passive access, if viable, is likely to be the next best thing; where physical access, in turn, is not viable, active access regulation is more likely to be appropriate.

43 BCMR A10.27 44 BCMR, A10.23

A1.9

A1.1

45 Natreas

43

Figure 12: sou

9 RecognisiNational the UK’s million torequires network iconsumer

10 To this en

“ ‐ hasducts a

However,fibre is noNGA. It h

ational Infrastrusury.gov.uk/nat

urce BCMR, figur

ng this, on Infrastructurtransport ano achieving tremoving uinfrastructurrs benefit frod the Plan st

s supported and poles…”

this stated powhere to behas all but be

ucture Plan, pagtional_infrastru

Non-c

re 108

29 Novembere Plan. Thend broadbanthe best supnnecessary re across theom diversity tates the Gov

significant

policy prefere seen in theeen effective

ge 70, paragrapucture_plan291

confidential v

er 2011 thee Plan contand network perfast broabarriers to,e UK which in the commvernment:

progress on

rence for phe regulatory fely abandone

ph 3.107 http://1111.pdf

version

e Governmenains major cinfrastructudband in Eu and suppowill promot

munications m

n opening up

ysical accessframework eed by Ofcom

/cdn.hm‐

nt publishedommitmentsure and comurope by 20orting the ete competitiomarket.45

p access to

s to ducts, poexcept in relam for other p

d an updates to improv

mmitted £83015. The Plaexpansion ofon to ensur

BT’s

oles and daration to fixeproduct area

d e 0 n f, e

rk d as


such as mobile backhaul, leased lines, in which BT still enjoys very nearly ubiquitous market power in wholesale markets, and a very significant market share in most retail markets, nearly 30 years after liberalisation began46.

A1.11 That BT enjoys such entrenched dominance 30 years after liberalisation sounds astonishing on the face of it however this can be explained in part by the significant economies of scope and scale involved in high capex network markets.

A1.12 This is also extremely disconcerting given that realising the potential from emergent and convergent technologies in next generation access and Long Term Evolution (LTE)/4G (and beyond) fundamentally depends upon CPs gaining access to sufficient backhaul bandwidth; and is critical for achieving Broadband Delivery UK.

A1.13 Tackling the dominance of the former incumbent is a function of applying regulatory remedies (in any market, not just the UK) that reflects the current and likely state of competition and maturity of market at the right times. Physical access remedies we propose in Chapter 3.1v are now appropriate and timely.

A1.14 Failure to adopt physical access in markets beyond LLU (in which market it has unquestionably promoted economic growth and enhanced the welfare of consumers47), is having a significant impact on the glide path to delivering the benefits of communications technology and spectrum bandwidths which will be available soon or are available now. In essence, while the potential for physical access to benefit fixed NGA is finally being recognised at the European level, regulatory remedies in mobile have lagged behind. The problem is not the answer, but the question; instead of asking “how can we ensure sufficient homes are connected to fibre?” Government should instead be asking “What is the most efficient and future‐proof way to deliver broadband?” This is a big missed opportunity.

46 For example, in Ofcom’s quarterly telecoms market data tables, Q2 2012, it was recorded that BT’s share of revenues in fixed telephony was 46.3%; and, while this shows a gradual but slight decline, in some segments it is still remarkably stable (e.g. UK geographic calls). In business markets, BT’s share has actually increased since 2010 (from 50.3% to 50.6%) 47 Residential broadband is a rare market in which BT’s retail share is consistently below 30%


A1.15 Importantly and tellingly, within the foreseeable future, consumers are unlikely to make a practical distinction between fixed broadband and 4G wireless (even if they may still be in separate economic product markets).

A1.16 Fixed NGA requires an extensive tree of connections across a wide area in a given DLE area. Many of the branches of this tree may serve a limited number of premises and thus be lightly loaded and potentially with low returns – this makes fixed PIA investment difficult (apart from all the practical issues). By contrast mobile has already achieved the area concentration of multiple users’ traffic at the cell site – the need is for a limited number of point to point links that will be heavily loaded. Arguably therefore mobile is a much more suitable case for PIA than fixed.

A1.17 However, the requirement for access to BT’s backhaul infrastructure in the middle mile is fundamentally similar for both. This paper concerns the need for physical infrastructure access to unlock the capabilities of 3G (now) and 4G mobile going forwards. But it is likely to be beneficial for the successful delivery of both new mobile and new fixed broadband technologies. On that basis it is important to note that the policy rationale for PIA applies across both platforms – dots which seem yet to be clearly joined in UK and European communications policy.

A1.18 Our key point is that opening physical access to BT’s network is likely to be essential to taking the UK communications sector forward, meeting the targets for the Digital Agenda and restoring the UK’s position as a leader in world mobile.

The state of play in physical access regulation

A1.19 BT has been subject to requirements to offer passive access products in relation to copper local loops since 8 August 2000 when Oftel brought into force Condition 83 of BT’s licence. Since then, copper local loops (as well as shorter “sub loops”) have been available to BT’s competitors on regulated terms. The Oftel approach was broadly reaffirmed by Ofcom in 2004.48

A1.20 More recently Ofcom’s 2010 Wholesale Local Access decision49 continued obligations on BT to provide access to its physical copper infrastructure in the form

48 Regulation (EC) 2887/2000; www.ofcom.org.uk/consult/condocs/rwlam/statement/rwlam161204.pdf 49 http://stakeholders.ofcom.org.uk/binaries/consultations/wla/statement/WLA_statement.pdf


of Local Loop Unbundling Services (“LLU”) and introduced, for the first time, rights to access BT’s ducts and poles.

A1.21 Demand for BT’s existing duct and pole access products has been limited. It is beyond the scope of this paper to assess the reason for this. As already discussed, though PIA take‐up on fixed has been very low, assuming this reflects likely take up of PIA for mobile is not a valid assumption. However, it is worth emphasising in this context that the economics of fixed and mobile networks look quite different. It is currently reasonably clear that competition amongst fixed networks over the “NGA last mile” (in fixed networks, the “sub‐loop”) is neither desirable, necessary nor economic. In mobile, the NGA last mile (essentially the radio segment) has been shown already to be susceptible to competition and roll‐out into it will be a condition of market entry through the 4G auction50. So while under the regulations as currently drafted there is apparently little demand from fixed network operators for passive access between the street cabinet and the local exchange (or even to the street cabinet itself), this is likely to be a simple function of the fact that there is limited (or no) potential for competition based on sub‐loop unbundling. The opposite is true at the equivalent point in mobile networks.

A1.22 Given this, it is particularly startling that physical access has been made available for fixed broadband networks – where standalone demand is still questionable because of the structure of competition ‐ but not for mobile networks (where the structure of competition makes it much more likely to be valuable, even on a standalone basis).

A1.23 On 20 September 2010 the European Commission published its Recommendations on regulated access to Next Generation Access Networks (NGA) (the “NGA Recommendation”). The NGA Recommendation was published in the context of meeting objectives of the Digital Agenda, one of the flagship initiatives for Europe, which sets targets for the deployment and take up of broadband and support investment based on optical fibre.

A1.24 Based on Article 19 of the Better Regulation Directive, the NGA Recommendation requires that NRAs “take utmost account” of it when determining appropriate remedies as part of their analysis of the ‘Wholesale (physical) network

50 One of the mooted 4G licences at 800Mhz will have roll‐out obligations – see Annex 2B to the auction IM


infrastructure access market’ and ‘Wholesale broadband access market’, markets 4 and 5 respectively. The NGA Recommendation identifies access to civil engineering infrastructure as a remedy which should be considered by the NRA when SMP is found in either of these two markets.

A1.25 Though the NGA Recommendation supports obligations to share ducts and poles for the purpose of mobile backhaul, it is less helpful in relation to dark fibre as it only relates to Fibre‐To‐The‐Home (“FTTH”) but not to other termination points such as cell site aggregation points.

A1.26 Ofcom’s 2010 WLA charge controls51 appear designed to meet the strict terms of the NGA Recommendation by imposing conditions on BT to provide access to, inter alia, ducts and poles in a way which is limited in scope and unequal in focus. Ofcom has cited insufficient demand and cost issues as being likely to be behind the slow uptake of the physical access products on offer.

A1.27 According to communications providers in a position to purchase these products, there is actually now, a significant opportunity here52. Vodafone and various other communications providers (both mobile and non‐mobile providers) have made numerous submissions to Ofcom in response to consultations for example in the BCMR Call for inputs April 2011 and BCMR June 2012 pleading reasons for a robust physical access regime to be introduced to open up the spare capacity of BT’s underused infrastructure. Most recently, in its draft response to the BCMR, Ofcom set out a further explanation for why it does not support passive access to ducts and poles in that Ofcom believes it will result in duplication of costs to industry and consumers. We are alive to this concern – and we touch on it elsewhere in this paper – but Ofcom does not back this up with any sound economic reasoning or evidence to support why this should occur.

A1.28 Uptake of the current physical access offer has not occurred because of a lack of demand but rather because of failure in the regulatory design and product. In particular the offer is limited to:

51 http://stakeholders.ofcom.org.uk/binaries/consultations/wla/statement/WLA_statement.pdf 52 See the responses to the BCMR by Vodafone, EE / MBNL, O2. Available here: http://stakeholders.ofcom.org.uk/consultations/business‐connectivity‐mr/?showResponses=true&pageNum=2#responses


“…broadband access networks serving multiple premises…”.

A1.29 This approach to regulation is akin to supply‐driven‐demand and amounts to

second‐guessing what is effective for the market. It is worth noting that in the 2010 WLA Statement Ofcom referred to the approach of the European Regulators Group (“ERG”) that:

“…the NRA will need to be careful not to second‐guess the market place but rather should provide a coherent background against which market developments take place.”53

A1.30 Excluding provision of leased lines and fixed, mobile and wireless backhaul by

‘second‐guessing’ demand is a significant flaw in the regulations with no justifiable reasoning or robust analysis behind it. The limitations imposed on the physical access offer by Ofcom reduce the revenues communications providers are able to achieve from deployments, in particular, compared with BT which does not face such restrictions. We understand that today, only one operator is seriously using Openreach’s duct sharing solution in the UK54.

A1.31 Notwithstanding this, passive access regulation in relation to LLU/last generation broadband has achieved unprecedented and unquestionable success in promoting intra‐platform competition. This is the only real example of a policy which has significantly undermined BT’s market power (and even there, in downstream wholesale broadband markets, BT enjoys SMP across 22.5% of the country). This policy is a clear example what can be achieved with effective regulatory design, in particular, which does not limit the use of the access seeker.

A1.32 In the same paragraph quoted by Ofcom, referred to above, the ERG also recognised:

“If the level of uncertainty as to replicability is low (i.e. replication that appears efficient has happened elsewhere), then there may be a case for believing replication is feasible in the particular context under consideration.”

53 Review of the Wholesale Local Access market, Ofcom, 2010, at 9.42 http://stakeholders.ofcom.org.uk/binaries/consultations/wla/statement/WLA_statement.pdf 54 Source: THC own research


A1.33 Ofcom has expressed concern that passive access will undermine active remedies.

We do not consider this to be a serious concern.

A1.34 Not only does this contradict Ofcom’s argument that there will be no take up of physical access, it is wrong headed. This argument implies that, because one regulatory remedy appears to be working adequately, a less interventionist regulatory remedy should not be permitted to compete with it. This is tantamount to protecting the structure of the market and businesses that currently use LLU.

A1.35 As set out above, there are simple and straightforward explanations for why the uptake of products offered as part of physical access regulation in the UK has been chequered, and it has nothing to do with market demand, cost or duplication – these arguments hold no water. What the UK experience in LLU and PIA regulation does provide is neat empirical evidence of the difference between good and bad regulatory design on separate supply markets that offer products and services that are indistinguishable to consumers for practical purposes. In 2010 a study by Deloitte noted55:

A further development of mobile broadband services with high potential, such as the fourth generation mobiles (4G), and other wireless technologies including satellite communications, is expected to result in a paradigm shift that will enable the emergence of ubiquitous data services, and thus combine the benefits of broadband with mobility.

A1.36 It is therefore unclear why passive access as a policy preference has been used as a remedy for fixed superfast broadband but abandoned for provision of the mobile equivalent. Ofcom must act to develop PIA regulation that provides access for communications providers to enter infrastructure markets to maximise the potential of communications technologies whose potential lies latent and untapped in BT’s physical infrastructure.

55 Deloitte, “Background document in support of the Digital Agenda for Europe”, 2010

Ann

Netw

Sub‐

Mob

Figure

Note

The

The mob

56 http:aring

50

nex 2 ‐ Co

work Diagram

‐loop PIA fro

bile network

e 13 – compariso

es:

configuratio

NGA local bile network

//www.openreg/downloads/PI

omparison

ms comparin

om BT IRO56

k from this p

on of PIA for FTTC

on is essentia

loop is the .

each.co.uk/orpgIA%20IRO%20F

Non-c

n of NGA t

ng sub‐loop P

paper (includ

C and mobile bac

ally the same

sub‐loop in

g/home/producFinal%20V01.do

CPhaOp

confidential v

topologies

PIA with Rem

ding NGA loc

ckhaul

e in each diag

the fixed n

cts/ductandpoloc

P PoP – point of ndover from penreach to CP

version

s

medies 1, 2 a

cal loop adde

gram.

network and

lesharing/ducta

nd 3 from th

ed)

d the air seg

andpolesharing

his paper

gment in th

g/ductandpoles

e

h


The NGA backhaul requirement is in the subloop for fixed networks and between the base station and the CP‐POP in the mobile network

The red lines in this diagram represent THC remedies 1, 2 and 3. .


Annex 3 ‐ Appellate rulings on Ramsey‐efficient pricing

Ramsey Pricing in Regulation

View from the MMC in its 2003 mobile termination inquiry

1. This note looks at the views on Ramsey pricing expressed by the MMC (the predecessor to the Competition Commission) in their 2003 inquiry into mobile termination.

2. This case contains the best appellate discussion on Ramsey pricing in relation to telecoms price controls (in the sense of being in‐depth and of wider application)57.

3. The MMC was essentially considering whether the regulator was in a position to set a Ramsey‐based mark‐up. This is obviously different from mandating baskets of prices in the hope that an SMP player itself will set Ramsey‐efficient prices. However, much of their thinking applies to both situations.

4. The Commission’s views can be summarised as follows:

a. It is very difficult for anyone to specify the own‐price and cross‐price elasticities in electronic communications markets with any degree of accuracy at all;

b. At the time (according to the Commission), the MNOs themselves had at most a pretty perfunctory understanding of the elasticities of their own product set, even at a firm‐specific level, let alone at the market level which would be required to set Ramsey prices;

c. The economic consultancies who tried to identify Ramsey prices didn’t do any better;

d. Absent perfect competition, the MNOs had no real incentive to set Ramsey prices anyway and, where they had the opportunity to do so

57 Ramsey pricing was also considered in Carphone Warehouse (1149/3/3/09) and also in the combined MTR cases (1180‐1183/3/3/11, CC decision of 9 February 2012); the discussion in each case was less in‐depth, more specific to its facts and in any event broadly consistent with the MMC’s view.


(e.g. in the difference between on‐ and off‐net pricing) they demonstrably did not.

6. So the MMC’s decision provides some useful pointers for analysing Ofcom’s approach to baskets in the LLCC. These would include:

a. Is there any actual evidence that BT’s prices within multi‐product baskets are set at a level which enhances efficiency?

b. Is there even any evidence that BT understands the super‐elasticities at a level of sophistication which enables it to set Ramsey prices?

c. Does BT have insight into elasticities at a market level (rather than just a firm‐specific level) in product areas which are more competitive?

d. Does competition in these markets act in such a way as to make Ramsey pricing likely to work in practice?

e. Is there evidence that BT might set prices at levels which a non‐Ramsey‐efficient?

f. Has pricing within baskets been analysed by Ofcom to ascertain whether those prices have been set at efficient levels under previous charge controls?

Extracts from MMC decision

2.127. We suggested to the MNOs that the price difference might relate to cost-allocation made by reference to different price-elasticities, as envisaged under demand-led pricing (for example, Ramsey pricing (see paragraphs 2.212 and 2.213)). Vodafone told us that the differential between on-net and off-net pricing did not reflect Ramsey principles. Orange said that it had no direct evidence on the relative magnitudes of demand elasticities between on- and off-net calls on which to base an assessment of whether on-net prices were consistent with Ramsey principles.

2.131(b) …the Ramsey price models presented by the MNOs, none of which differentiates between on-net and off-net calls in terms of price elasticities, could not be relied upon to give an accurate depiction of optimum relative prices even in their own terms,

2.201. We discussed the difference between the prices of on- and off-net calls earlier, in paragraphs 2.122 to 2.131, and expressed our view that this difference threw doubt on the MNOs’ claim that their pricing approximated to a Ramsey structure of prices.

2.431. First, we were not satisfied that there was any way of establishing reliable estimates of elasticities of demand in the mobiles sector with enough precision to


inform pricing decisions. Hence, we believe that there are problems in calculating reliable Ramsey prices—we discuss this issue in more detail in Chapter 8.

2.434 Taking the considerations in paragraphs 2.431 to 2.433 into account, we conclude that the MNOs do not currently set Ramsey prices, except possibly for the time-of-day variations around the regulated (or quasi-regulated) average termination charge. We accept that the MNOs’ decisions on the day, evening and weekend rates to set, given the target average charge, are based on at least an informal understanding of the respective price-elasticities and are there-fore fairly similar to Ramsey prices

2.515. We have set out above our conclusion on the correct approach to establishing the costs of call termination. In our view, a Ramsey approach is not consistent with the cost-causation principle, that callers to mobile phones should pay only those costs that they cause by virtue of making calls. But, in addition, we identified three further objections to the Ramsey approach, which we set out in the following paragraphs.

2.516. First, we believe that if we were to set a Ramsey-based termination charge, we would have to be confident that other relevant prices would also be set at Ramsey levels. In other words, we would have to be confident either that it would be appropriate for us to intervene to set not just termination charges but all other relevant prices, or that if we set termination charges at Ramsey levels those other relevant prices would be set at Ramsey levels by the MNOs. This is because the Ramsey approach is concerned with the relativities between different prices, for example, subscription, origination and termination. In other words, Ramsey pricing is about a structure of pricing. However, we do not believe it would be appropriate for us to try to set levels for all the relevant prices. Further, if termination charges were set at the level commensurate with Ramsey pricing, there would, in our view, be no guarantee that the MNOs would set their different retail prices at Ramsey levels which maximized overall consumer surplus, subject to the recovery of fixed and common costs. A regulated Ramsey-based termination charge would be likely to ensure an efficient, Ramsey-type outcome in the mobile sector as a whole only if competition at the retail level was sufficient to constrain the MNOs to set their overall structure of prices at Ramsey levels. We have, however, already concluded that the retail market is not fully competitive and, as a result, the MNOs would both wish, and be able, to set some prices at levels different from those of Ramsey in that market. Moreover, it is clear that at least some prices at the retail level (for example, on-net and off-net calls) are even now not set at levels consistent with a Ramsey pattern of prices. Furthermore, as profit-maximizing firms, the MNOs will set prices using firm-specific elasticities of demand and their relative levels. We think these will clearly differ from the analogous market elasticities, not least because the retail market is not fully competitive. Consequently, even if (which is not the case) we thought that Ramsey pricing was in principle superior to the ‘fair charge’ approach which we advocate, we would nevertheless find it impossible to recommend Ramsey pricing on this occasion.


2.517. Our second objection to Ramsey pricing arises from the formidable problems associated with obtaining reliable estimates of the elasticities of demand which are the basis of the Ramsey approach. It is clear from the evidence submitted to us during our inquiry (dis-cussed in detail in Chapter 8) that no unanimity—indeed, considerable disagreement—exists among the main parties and their economic advisers as to the correct values for the various different elasticities that are involved. Moreover, a large number of elasticity estimates are required to establish Ramsey prices in the mobile sector as a whole (including those for text messaging and mobile internet), which compounds the problems inherent in obtaining a reliable set of estimates. This means that, if Ramsey were to be used as the basis of a regulated price, we believe that the DGT would experience considerable difficulty in achieving consensus on reliable estimates of the relevant elasticities forming the basis of the Ramsey price structure. No consensus was reached among the parties submitting evidence on this matter during our inquiry. There would in our view be considerable scope for disagreement if these exercises were to become part of the regulatory process, giving rise to prolonged disputes between the DGT and the MNOs and resulting in an increase in the overall costs of regulation and delay in the realization of the regulatory benefits to consumers. Oftel, indeed, told us that the informational requirements were too onerous for Ramsey pricing to provide a reliable basis for regulated termination charges.

2.518. Third, we believe that a regulated price based on Ramsey principles would lead to the sort of distributional inequities which we have discussed earlier in this chapter (see paragraphs 2.388 to 2.400) and which result in some consumers unfairly subsidizing other consumers.

2.519. The considerations set out in paragraphs 2.510 to 2.518, together with evidence dis-cussed earlier in this chapter and in Chapters 6 and 8, lead us to believe that the objections to using Ramsey principles to set an average regulated mobile termination charge are conclusive. In summary, they are:

(a) the absence of consensus regarding absolute and relative price-elasticities, determination of which is necessary for the computation of Ramsey prices (see Chapters 8 and 9);

(b) the extreme disagreement among the parties as to the nature and extent of fixed and common costs, determination of which is also necessary for the computation of Ramsey prices (see Chapter 7);

(c) the implausible outputs of some of the models claiming to compute Ramsey prices (see Chapter 9 and Appendix 9.1);

(d) the inconsistency between actual retail price patterns observed and the retail price patterns claimed as naturally resulting, in accordance with Ramsey principles, from competition in the absence of regulatory constraint (see paragraphs 2.122 to 2.131);

(e) the uncertainty as to the actual level of prices, having regard to the difficulty of allocating subscription charges between handsets and calls (see paragraphs 2.177 to 2.187);


(f) our doubts as to the complete effectiveness of retail competition necessary for the emergence of Ramsey pricing patterns even in theory (see paragraphs 2.210 and 2.211); and

(g) our inability, in any event, to recommend Ramsey prices in the retail market, if our proposals for regulation relate to one element only, namely termination charges (see paragraph 2.516).


ANNEX 4 – PIA REQUIREMENTS EXPRESSED IN cm3 PER ADDRESSABLE HOUSEHOLD

Non confidential version Contents - Ofcom · 7 Non-confidential version 2. Background to regulation of physical access ‐ the backhaul bottleneck The backhaul bottleneck 2.1 It is

Documents