n/e/r/a THE DESIRABILITY AND FEASIBILITY OF AN EFFICIENCY-IMPROVING LOCAL INTERCONNECTION POLICY A Report for OPTA Prepared by NERA

LOCAL INTERCONNECTION POLICY

A Report for OPTA

Prepared by NERA

December 2002 London Project Team: Nigel Attenborough Gideon Koch Derek Lyon (JRTC) Tim Miller

TABLE OF CONTENTS

1.

INTRODUCTION 3

1.1. Terms of Reference 3

1.2. Local Interconnection 3

1.3. Report Structure 5

2.

THE EFFICIENCY OF KPN’S EXISTING NETWORK

6

2.1. Introduction 6

2.2. What We Can Learn From BULRIC 6

2.3. Results 7

2.4. Conclusion 8

3.

FEASIBILITY OF AN EFFICIENCY-IMPROVING LOCAL

INTERCONNECTION POLICY

10

3.1. Introduction 10

3.2. Policy Options 10

3.3. Assessment Criteria 13

3.4. Performance of Different Policy Options 15

APPENDIX A.

SWITCH REBUILDING COSTS

20

A.1. Principle One - Cost Causality 20

A.2. Principle Two - Distribution of Benefits 22

A.3. Principle Three - Cost Minimisation 23

A.4. Principle Four - Practicability 23

A.5. Conclusion 24

APPENDIX B.

COUNTRY SURVEYS

25

B.1. Austria 25

B.2. The UK 27

B.3. Ireland 29

B.4. Germany 30

APPENDIX C.

MARKET PARTICIPANT INTERVIEWS

33

APPENDIX D.

ALTERNATIVE AUSTRIAN SOLUTIONS

34

D.1. Different Options 34

D.2. Assessment of Variants of the Austrian Solution 35

D.3. Analysis 38

EXECUTIVE SUMMARY

Desirability

1. Analysis using the BULRIC model suggests that there is substantial scope for increasing the efficiency of KPN’s network structure.

2. Modelling based on a “scorched node” assumption indicates that by employing fewer but larger switching units, KPN would need 369 fewer host local switch units in its network, and 781 fewer remote concentrator units. In addition, the number of host local switch sites in a fully efficient network would be 171 compared with around 400 in KPN’s existing network.

3. The existing BULRIC model does not allow for modelling under a ‘scorched earth’ approach, which would assume an optimised number and different mix of switching units together with optimum network structure diversity, and would not be constrained to use all the existing switch nodes. However, indications are that such a network might have fewer than 100 host local switch sites, compared with 400 local switch sites in KPN’s existing network.

4. The fact that there is an excessive number of switch units and an inefficient mix of local switch and remote concentrator sites means that KPN’s competitors will face a sub-optimal set of local points of interconnection. This will distort their choice their “make or buy” decision and lead to an inefficient level of investment in infrastructure. There are therefore strong indications that an efficiency-improving local interconnection policy would, in principle, be desirable.

Feasibility

5. Four possible policy responses have been identified and assessed against a variety of criteria. The policy responses considered are status quo (“do nothing”), virtual interconnection, the modified Austrian solution and fixed fee per local switch. The assessment criteria used were efficient infrastructure competition, incentives for KPN to increase its efficiency, switch rebuilding costs, regulatory burden, division of costs between KPN and other operators and discrimination between other operators. 6. It is difficult to draw categorical conclusions. However, it would appear that virtual

interconnection is not a desirable policy and is worse than the status quo. Both the modified Austrian solution and FFPLS have the potential to outperform the status quo.

7. The preferred option will depend on the weightings attached to the different assessment criteria. Concerns regarding the regulatory burden and possible discrimination between OLOs tend to point in favour of FFPLS. If switch rebuilding

costs are high, there is probably little to choose between FFPLS and the modified Austrian solution in terms of efficient infrastructure provision. If, on the other hand, switch rebuilding costs are low, the modified Austrian solution may be superior in terms of efficient infrastructure provision.

1. INTRODUCTION

1.1.

1.2.

Terms

of

Reference

National Economic Research Associates (NERA) has been commissioned by OPTA to investigate the desirability and feasibility of a local interconnection policy which would enable interconnecting operators to make a more cost-efficient “make-or-buy” decision for their local roll out to the local switch level of KPN’s network. This study results from OPTA’s Memorandum of Findings regarding ‘Integral Tariff Regulation for End-User and Interconnection Services’, ref. OPTA/EGM-IBT/2002/201084.1

Local

Interconnection

The provision of interconnection on fair and efficient terms is widely recognised as essential for the creation of a dynamically efficient and competitive telecommunications market. Interconnection charges can account for a substantial proportion of operators’ costs and it is therefore important that these charges are soundly derived and give proper economic signals to both the incumbent operator and interconnecting operators, to guide their investment decisions.

More specifically, it is important that competitors can make efficient and cost-minimising decisions about whether to “make” (i.e. invest in their own infrastructure) or “buy” (purchase interconnection services). This is implied by Article 8 of the new EU Framework Directive, which states that:

The national regulatory authorities shall promote competition in the provision of electronic communications networks, electronic communications services and associated facilities and services by inter alia…encouraging efficient investment in infrastructure, and promoting innovation.

This is likely to require a choice of interconnection points and that interconnection be available at cost-oriented prices.

In the Netherlands the picture is complicated by two interrelated factors. The first complication is KPN’s existing network structure, which has a large number of small local switches, each with relatively few customers attached. This means that widespread interconnection at local switch level is likely to require a particularly large amount of infrastructure investment by KPN’s competitors. If KPN’s current network topology were efficient then it might be argued that the existing situation is economically efficient, at least in a static sense. Competitors would choose between interconnection charges based on efficient interconnection costs or investment in their own network facilities. There is,

1 _{Paragraph 187.}

however, good reason to suggest that KPN’s existing network structure is not efficient, with too many small local switches instead of fewer, larger local switches and a greater number of remote concentrators. A network involving fewer, larger local switches would reduce unit costs and also make the cost of interconnection at the local level less prohibitive for KPN’s competitors.

The second complication is KPN’s existing local switching technology, which is relatively outdated. It is expected that many local switches will need to be substantially upgraded in order to offer interconnection with other operators. Whilst the precise cost is unknown at this stage, since there appears to be around 540 local switch units in KPN’s current network, the cost of upgrading the entire network could be very expensive. However, where switch upgrades would be required anyway for other reasons, the costs attributable to offering interconnection could be relatively small (for example software changes would need to be implemented irrespective of the requirement for local interconnection). An important question is therefore the extent to which switch upgrades would occur anyway and when they would occur (see Appendix A).

Because these factors create what may be a large gap between the costs that would have been incurred by interconnecting operators in the case of a fully efficient network and those given KPN’s current network, they offer significant challenges to the achievement of cost-efficient choices between interconnection and investment. This study examines the desirability and feasibility of implementing a policy (hereafter known as an “efficiency-improving” local interconnection policy) to compensate for these factors by attempting to answer the following questions:

•

To what extent is such a policy desirable?

•

What feasible alternative policy options exist? and

•

How effective are they at increasing efficiency, while keeping down the level of switch rebuilding costs, minimising the regulatory burden and ensuring fair treatment between KPN and its competitors and between the competitors themselves?

NERA has been informed that OPTA intends to employ engineering consultants to examine carefully the costs proposed by KPN for upgrading their local switches. Since the results of this assessment will not be known for some time, the report does not assume a particular level of switch rebuilding costs but examines what the implications would be of different levels of rebuilding costs.

In preparing the report, interviews were carried out with KPN, three alternative operators in the Netherlands2_{, and officials from the telecommunications regulatory authorities in the}

2 _{Versatel, WorldCom and Tele2.}

UK, Ireland, Germany and Austria. Relevant previous reports prepared for OPTA by external consultants have also been reviewed, in particular the report by Brunel University entitled, ‘The Relationship between Access Pricing Regulation and Infrastructure Competition’ (March 2001).

1.3. Report

Structure

This report is structured as follows:

•

Section 2 examines the question of whether an efficiency-improving local

interconnection policy is desirable, by using the BULRIC model to derive a number of insights in the efficiency of KPN’s existing network. Conclusions are provided at the end of the section;

•

Section 3 identifies and describes a number of policy options and rates then against various criteria and against the alternative of leaving things as they are now (“Status quo”). The analysis draws on lessons from the adoption of similar policies in other European countries. The findings of this analysis provide pointers to the preferred choice of policy option;

•

We discuss the recovery principles that are relevant when assessing the most appropriate cost recovery mechanism for the switch upgrading costs in Appendix A;

•

Appendix B summarises the local interconnection policies in the UK, Ireland,

Germany and Austria;

•

Appendix C summarises the key issues to arise from the interviews of KPN and the three Dutch alternative operators that were carried out as part of the investigation; and finally

•

Appendix D provides a relative assessment of different versions of the modified Austrian solution.

2.

THE EFFICIENCY OF KPN’S EXISTING NETWORK

2.1. Introduction

2.2.

In order to assess whether an efficiency-improving local interconnection policy is desirable, it is necessary to attempt to estimate the extent to which KPN’s existing network is efficient. This will also help ensure the policy response is proportionate, should there be indications that KPN’s network is inefficient, and may also offer indications of how the policy response might best be designed.

One relatively straightforward method for assessing KPN’s network efficiency is to use the bottom up forward looking long-run incremental cost (BULRIC) model that has been developed to set termination rates in the Netherlands. By modifying and flexing the model, it is possible to derive insights into the efficiency of KPN’s network.

What We Can Learn From BULRIC

The development of the BULRIC model involved modelling KPN’s network structure and variations of it based on different network optimisation assumptions. In this context, the term ‘network structure’ relates to the hierarchical levels of switches, the number and locations of switches, and the transmission technologies employed. For BULRIC, a network structure-modelling tool was used, involving highly complex algorithms, to model KPN’s network under different network optimisation assumptions.

For the purposes of this study, we were able to employ the network structure-modelling tool to provide insights into the extent to which KPN’s existing network structure differs from more optimised network structures. KPN’s existing network structure was compared with two alternative network structures, each of which had the same characteristics as the existing network in terms of traffic handling capability, number and distribution of subscriber access lines, quality criteria, and network diversity but with different levels of optimisation:

•

Modified Scorched Node I – assuming the same number (and location) of switch

nodes as in KPN’s existing network structure 3_{, but where the host local switch}

hierarchy is semi-optimised (the switch units are assumed to be capable of handling up to 50,000 lines); and

•

Modified Scorched Node II – assuming the same number (and location) of switch

nodes as in KPN’s existing network structure, but where the host local switch

3 _{The total number of switch and remote concentrator sites is identical to the existing KPN network, except that one}

site was excluded since there are no connections to the site.

hierarchy is fully-optimised (the switch units are assumed to be capable of handling up to 100,000 lines).

All equipment cost and output assumptions were kept constant in each of the three cases, in order that true like-for-like comparisons could be made.

2.3. Results

The results of the modelling exercise can be seen in Table 2.1 below, against KPN’s existing network switch hierarchy:

Table 2.1

BULRIC Modelling Results Network Structure Resulting number of tandem switch units Resulting number of local switch units Resulting number of remote concentrator units Total number of local and remote concentrator units (Total number of switch sites4₎ Existing KPN network

(with existing switch hierarchy) 18 540 1,963 2,503 1353

Modified Scorched Node I (with semi-optimised switch

hierarchy)

18 237 1,116 1353 1353

Modified Scorched Node II (with fully-optimised switch

hierarchy) 18 171 1,182 1353 1353

2.3.1. More efficient switch structure

These results suggest that if the switching hierarchy of KPN’s network was semi-optimised for its existing traffic and network diversity requirements, the number of local switch units required would fall from 540 in the current network to 237, and the number of remote concentrator units would fall to 1,116 from 1,963 in the current network. In other words, simply by using a smaller number of larger units and replacing some host switches by concentrators, there would be 303 fewer local switch units, and 847 fewer remote concentrator units. Overall, there would be 1,150 fewer units in the network. Because of economies of scale in switching and concentrators this would reduce costs considerably.

4 _{Under modified scorched node, all tandem switches are colocated with local switches.}

If the switching hierarchy of KPN’s network was fully-optimised for their existing traffic and network diversity requirements, the number of local switch units would be reduced from 540 in the current network to 171, and the number of remote concentrator units would fall to 1,182 from 1,963 in the current network. Therefore, overall there would still be 1,150 fewer units in the network, but network costs would be likely to fall even further, since switching costs would fall considerably.

2.3.2. Smaller number of host local switch sites

The results also suggest that by semi-optimising the switching hierarchy of KPN’s network, the actual number of host local switch sites would fall from around 400 in KPN’s existing network to only 237. And by fully-optimising the switching hierarchy of KPN’s network, the number of host local switch sites would fall even further, from around 400 in KPN’s existing network to 171 host local switch sites.

The implication for OPTA’s policy response is that not only would the cost of interconnection for competitors fall (as explained above) but since competitors would only have to roll out to a fewer number of sites, their roll out costs would also be significantly reduced.

In theory, it is also possible, to model a network using a Modified Scorched Earth approach, which would assume an optimised number and different mix of switching units together with optimum network structure diversity, and would not be constrained to use all the existing switch nodes.5 _{Under this approach, the number of host local switch sites in the}

network would be likely to fall even further, and therefore the make-or-buy decisions of competitors would be even more cost-efficient.

Unfortunately the existing BULRIC model does not have the capability to carry out this form of optimisation since this would require modelling the access network as well as the core network. However, indications from the model are that the total number of host local switch sites would be likely to fall from around 400 in KPN’s existing network to less than 100 host local switch sites.

2.4.

Conclusion

This modelling exercise indicates that there is substantial scope for increasing the efficiency of KPN’s network structure.

The modelling demonstrates that by efficiently employing digital host local switch units with capacity for up to 100,000 lines, KPN would require only 171 host local switch units

5 _{Theoretically, it might even be possible to vary the location of the transmission links (this would be a ‘Scorched}

Earth’ approach). However, this approach is rarely adopted since it requires modelling the access network in detail as well as the core network.

rather than the 540 present in their current network, and the number of remote concentrator units required would fall from 1,963 in the current network to 1,182 remote concentrator units. Together, this implies there would be 1150 fewer units in the network.

The results also suggest that by semi-optimising the switching hierarchy of KPN’s network, the actual number of host local switch sites would fall from around 400 in KPN’s existing network to only 237 and by fully-optimising the switching hierarchy of KPN’s network, the number of host local switch sites would fall even further, from around 400 in KPN’s existing network to 171.

The implication for OPTA’s policy response is thus even starker: not only would the cost of interconnection for competitors fall (as explained above) but since competitors would only have to roll out to a smaller number of sites, their roll out costs would also be significantly reduced.

In theory, it is also possible, to model a network using a Modified Scorched Earth approach, which would assume an optimised number and different mix of switching units together with optimum network structure diversity, and would not be constrained to use all the existing switch nodes. Under this approach, the number of host local switch sites in the network would be likely to fall even further, and therefore the make-or-buy decisions of competitors would be even more cost-efficient.

Unfortunately the existing BULRIC model does not have the capability to carry out this form of optimisation since this would require modelling the access network as well as the core network. However, indications from the model are that the total number of host local switch sites would be likely to fall from around 400 in KPN’s existing network to less than 100 host local switch sites.

The fact that there is an excessive number of switch units and an inefficient mix of local switch and remote concentrator sites means that KPN’s competitors will face a sub-optimal set of local points of interconnection. This will distort their choice of “make or buy” decision and lead to an inefficient level of investment in infrastructure. There are therefore strong indications that an efficiency-improving local interconnection policy would, in principle, be desirable.

3. FEASIBILITY OF AN EFFICIENCY-IMPROVING LOCAL

INTERCONNECTION POLICY

3.1. Introduction

3.2.

The previous section examined the extent to which KPN’s network is efficient (in terms of the number and type of switch unit s and sites).

In this section we explore the question of whether it is feasible for OPTA to redress the balance through a local interconnection policy so that the ‘make or buy’ decisions of competitors are more cost-efficient.

We begin by considering the main policy options that are available to OPTA and then go on to specify a set of criteria against which to assess the different options.

Following that, we assess the different policy options against these criteria, drawing on lessons from the local interconnection policies in the countries studied (see Appendix B). The results of this analysis provide pointers to the preferred choice of policy option.

It is important to note from the outset that the final choice of policy may be influenced by the weights attached to the different assessment criteria. In addition there is a need for the chosen policy option to be consistent with the policies adopted in related areas. Consideration of such policy consistency is outside the scope of this study. For these reasons, we believe that OPTA is in the best position to reach definitive conclusions about the preferred policy option.

Policy

Options

As part of this study we have considered four broad policy options, each of which is described below. These are:

•

Status quo;

•

Virtual local interconnection;

•

Modified Austrian approach;

•

Fixed fee per local switch.

3.2.1. Status quo

The status quo option involves leaving things as they are. Operators pay local interconnection charges where they have rolled out to KPN local switches and hence are obliged to roll out to all of KPN’s local switches if they wish to be charged local interconnection charges for all traffic. In order to interconnect at KPN local switches it

would be necessary for them to invest in their own network infrastructure or lease lines from KPN in order to roll out their networks to the local switches. They would also have to pay set up costs to KPN that would cover the costs of switch rebuilding. The term switch rebuilding is used here to refer to the costs of conditioning the switches and updating the software and hardware so that local interconnection is possible.

No specific measures would be taken to promote efficiency. However, other policies that have a bearing on efficiency, for example the proposed multiple year interconnection tariff control system and the fact that terminating tariffs are already based on a network with fewer local switches, will provide an independent incentive for KPN to improve the efficiency of its network and hence the efficiency of the make or buy decisions of those operators wishing to interconnect with KPN.

The purpose of including the status quo option is that it provides a reference point against which to judge alternative local interconnection policies.

3.2.2. Virtual local interconnection

Virtual local interconnection involves the payment of local interconnection charges even though interconnection takes place at the regional level. The simplest version would involve all traffic that interconnects at the regional level being charged for at local interconnection rates. However, it is also possible to conceive of variants where only a given volume or proportion of traffic that interconnects at the regional level would be eligible for local interconnection rates.

3.2.3. Modified Austrian approach

The third type of policy considered in this report involves the requirement for other operators to roll out their networks to interconnect with KPN at a specified number of local switches. Provided that they do so, all their interconnection traffic attracts local rates. One version of this type of policy has been applied in Austria (see Appendix B). In this case operators need to roll out to 43 out of a total of more than 200 Telekom Austria local exchanges in order to secure local interconnection charges for all their interconnecting traffic. Another variant occurs in Germany. There the network has 23 high level catchment areas and 475 local catchment areas (each of which contains several local switches). Provided that an interconnecting operator has a network presence in all 23 of the high level catchment areas, it qualifies for local interconnection charges in each of the local catchment areas in which it is present. It is therefore possible to obtain local interconnection charges for all interconnecting traffic without having to roll out to all of Deutsche Telekom’s local switches.

The performance of this type of policy will be assessed in a later section of this report. Suffice it to say at this stage that one of the objectives quoted by RegTP in Germany is to

replicate more closely the situation that would exist were the incumbent’s network structure more efficient.

It is possible to conceive of a large number of variants depending on:

•

whether the NRA specifies a number of zones or a number of switches to which operators must roll out;

•

who chooses the roll out switch locations (OPTA or the interconnecting operators or KPN);

•

whether all traffic that continues to interconnect at tandem switches attracts local interconnection rates or whether only a proportion does so.

In addition, arrangements have to be made for recovery of switch conditioning and rebuilding costs.

The relative merits of a selection of these possible variants is explored in Appendix D and what are considered to be the best variants are then assessed as part of the policy assessment later on this section.

3.2.4. Fixed fee per local switch

The final policy considered in this report is what we have called fixed fee per local switch. The idea originated within OPTA and it is intended to overcome the problem associated with switch rebuilding costs, while at the same time having desirable efficiency properties. Under this policy interconnecting operators would determine the KPN local switches (i.e. the subscriber numbers) to which they would like local interconnection.6 _{For the related}

interconnection traffic they would pay local interconnection charges. They would also have to pay a fixed fee per local switch (set of subscriber numbers) to which they want to interconnect.

This fixed fee is designed to cover switch rebuilding and related costs. It would be determined as follows:

•

The total cost of rebuilding all existing KPN local switches would be calculated.

•

OPTA would then determine whether particular types of rebuilding costs should be

excluded. For example, analogue switches are due to be phased out anyway and would be replaced either by concentrators or digital local switches. In the former

6 _{The term local switch as used here effectively means set of subscriber numbers. The relationship between switches}

and subscriber numbers is determined by the existing network structure.

case there would be no local interconnection, while in the latter case there would be a need to rebuild the switches totally irrespective of any requests for local interconnection.

•

A total figure for rebuilding net of exclusions would then be calculated and this would be divided by the number of existing local switches. This would be the fixed fee per local switch (set of subscriber numbers), which would be shared between the different operators seeking to interconnect at that switch.

KPN would be free to provide the local interconnection in the way that is most cost effective for it. For example, it might choose to opt for virtual interconnection or to provide the interconnection via another local switch, provided that it reimbursed other operators for any additional costs that were imposed on them.

3.3. Assessment

Criteria

In this section we describe the assessment criteria that are subsequently used to rate the different policy options.

3.3.1. Efficiency

Economic efficiency in the context of the operation of, and investment in, infrastructure has three (interdependent) components:

•

Technical efficiency — Where firms have the appropriate incentives to produce

services at least cost at a given point in time.

•

Allocative efficiency — Where the relationship between prices and costs is such that

resources are allocated in way that maximises economic welfare. This implies prices are equal to marginal cost plus an appropriate mark-up to recover fixed and common costs. It also implies that production activities are distributed between firms such that industry-wide costs are minimised.

•

Dynamic efficiency — Where firms have the appropriate incentives to invest,

innovate, improve the range and quality of services, increase productivity and lower costs over time.

For a given technology, technical efficiency will be achieved if there is pressure on KPN to reduce costs to the minimum level. This would happen if there were a competitive market for local interconnection services. In the absence of such competition, encouragement for KPN to remove excess costs will come from price control (i.e. the proposed multiple year tariff system) and from an appropriate local interconnection policy.

Meanwhile, allocative efficiency depends on prices being equal to LRIC plus an appropriate mark-up to recover common fixed costs and interconnecting operators being able to choose between efficient investment in their own infrastructure and efficiently determined interconnection charges. In these circumstances, KPN’s competitors will buy interconnection services from KPN if they are not able to achieve lower costs by building their own infrastructure. In this way, network services are always provided by the most efficient (i.e. lowest cost) operator.

It is important therefore that any local interconnection policy is consistent with the achievement of both technical and dynamic efficiency and the achievement of minimum overall network costs in the Netherlands (i.e. allocative efficiency).

The assessment criteria we have adopted therefore include:

•

Consistency with efficient infrastructure competition (i.e. allocative efficiency); and

•

Creation of incentives for KPN to increase the efficiency of its network (i.e.

productive and dynamic efficiency).

3.3.2. Switch rebuilding costs

As mentioned in Section 1.2, there are indications that a large number of KPN’s local switch units would need to be rebuilt in order to provide local interconnection at existing local switch sites. Whilst the precise incremental cost of upgrading these switches to provide local interconnection is currently unknown, since KPN has well in excess of 500 local switch units in its network, the total cost of upgrading the entire network could be significant.7

The impact of different policies on the size of switch rebuilding costs is therefore an important consideration when deciding between them.

3.3.3. Regulatory burden

The implementation of an efficiency-improving local interconnection policy is likely to impose a regulatory burden on both OPTA and operators in the Netherlands. The regulatory burden can be expected to take a variety of forms. OPTA will have to make a variety of decisions concerning the precise nature of the chosen option and will have to monitor and secure enforcement of the policy as well as deal with any disputes arising from it. Meanwhile operators will have to comply with the policy and may become involved in disputes. The size of the regulatory burden for each party will vary with each policy option, and therefore should be considered when assessing each policy.

7 _{Notwithstanding the fact that there are still over 250 PRX/A analogue switches and some analogue AXE switches}

in the network, most, if not all of which would be replaced by remote concentrators in an efficient network structure and would therefore not need to be rebuilt.

3.3.4. Division of costs between KPN and other operators

A further important consideration is whether the costs of any local interconnection policy are fairly distributed between KPN and other operators. Particular issues are whether other operators pay the true interconnection costs that they impose on KPN and how the cost of switch rebuilding should be recovered. A review of the principles that can be applied when addressing this latter question is contained in Appendix A.

3.3.5. Discrimination between other operators

It is also necessary to assess the extent, if any, that the different policies discriminate between different interconnecting operators.

3.4.

Performance of Different Policy Options

3.4.1. Status quo

The purpose of including the status quo option is that it is a reference point against which the alternative local interconnection policies can be judged. Its scores against each of the criteria have been arbitrarily set equal to zero. This means that if an alternative policy outperforms the status quo against a particular assessment criterion it will receive a positive mark, while, if it is inferior, it will receive a negative one. In this way the different alternative policies can be compared with each other and with the status quo.

The key characteristic of the status quo is that local interconnection is unlikely to occur on a significant scale. This is because of the high charges associated with switch rebuilding and the large number of local switches. As a result the status quo performs poorly in terms of encouraging efficient infrastructure competition. It also provides no additional incentives for KPN to improve the efficiency of its network.8 _{The associated switch rebuilding costs are}

likely to be minimal because no local interconnection is occurring.

The regulatory burden associated with the status quo is small mainly because nothing is happening. Meanwhile KPN benefits at the expense of other operators because the latter have to pay regional interconnection charges for both biba and buba calls, which means that the former are not profitable. There is no evidence that the present situation discriminates in favour of particular interconnecting operators (OLOs).

3.4.2. Virtual interconnection

Virtual interconnection performs badly in terms of encouraging efficient infrastructure competition. Indeed, given that regional interconnection is priced at local interconnection rates, there would be no incentive to roll out infrastructure. In contrast, virtual

8 _{Incentives are already provided via the proposed multi-year wholesale tariff system.}

interconnection would encourage KPN to improve the efficiency of its network as it operates like a sharp reduction in prices. It would also not be associated with any local switch rebuilding costs, as there would be no local interconnection.

The regulatory burden would probably be small, although one would expect KPN to be unhappy about virtual interconnection and this could lead to a prolonged regulatory process. Virtual interconnection clearly discriminates against KPN as it would be obliged to provide regional interconnection at local interconnection prices. There would not, however, be any discrimination between OLOs unless some had already partly rolled out their networks in anticipation of local interconnection while others had not.

3.4.3. Modified Austrian solution

As already mentioned, it is possible to envisage a substantial number of variants of the Austrian solution. Appendix D provides an attempt at a comparative assessment of some of the potentially better ones. The conclusion that comes out of this analysis is that the best variants would involve OPTA specifying a set of zones in the Netherlands (e.g. one for each tandem switch) and then specifying the number of local interconnection points within each zone in order for an operator to secure local interconnection tariffs for all its interconnection traffic within that zone.9 _{The precise nature of the best variant would also depend on who}

chooses the location of the switching points in each zone and the size of switch rebuilding costs. However, for the purposes of rating the modified Austrian solution against the assessment criteria, the different candidates for being the best variant are likely to have similar scores.

The modified Austrian solution seeks to secure efficient infrastructure competition by replicating the situation that would exist were KPN’s network efficient. If this could be achieved it would avoid the situation in which OLOs base their roll out decisions on the existing inefficient KPN network structure and end up either spending more or rolling out less than would have been the case had the network structure been efficient.10 _{In reality it}

may be difficult to secure perfect replication of the efficient network situation either because OPTA lacks complete information (where it specifies the switch locations) or the preferences of OLOs do not correspond to the efficient network structure (where they specify the location).

The modified Austrian solution provide incentives for improvements in the efficiency of KPN’s network as it will focus traffic and local switch rebuilding at the correct points in the network. It should also limit switch rebuilding costs as it will avoid the need to rebuild

9 _{The idea here is that the minimum number of local interconnection points corresponds to the situation that would}

exist if KPN’s network structure were already efficient.

10 _{Many telecommunications assets are sunk costs and hence infrastructure investment cannot be recovered in the}

event of incorrect decisions.

analogue switches and limit the number of switches at which switch rebuilding is required as a result of the need to provide local interconnection.

The modified Austrian solution is likely to impose quite a substantial burden on OPTA in terms of defining zones, specifying the number of points of interconnection in each zone and (in the event that OLOs cannot agree on the specific points of local interconnection) specifying the switches at which local interconnection should take place. There is also a potential problem in that specifying the points of interconnection might favour some OLOs against others. If that happens, there might in turn be disputes and challenges to the decision, which would involve an additional regulatory burden. In contrast the division of costs between KPN and the OLOs should be appropriate as interconnecting operators will pay the true efficient costs that they impose on KPN.

3.4.4. Fixed fee per local switch

Fixed fee per local switch (FFPLS) seeks to secure efficient infrastructure investment by providing the right incentives to KPN. Decisions about where local interconnection should take place are left to KPN. OPTA decides on the reasonable level of switch rebuilding costs and a figure is derived for the average reasonable rebuilding cost per switch (set of subscribers).11 _{This is the fee that KPN is allowed to charge for switch rebuilding costs.}

KPN is therefore provided with an incentive to specify interconnection for a set of subscribers at a point where the actual rebuilding costs per switch are less than or equal to the average cost. In that way KPN is encouraged to choose switches that are cheap to rebuild and to minimise the extent of switch rebuilding costs by focussing local interconnection at particular locations. As it is also required to pay for any costs imposed on interconnecting operators as a result of subsequent changes in the points of interconnection it can be expected to take the future more efficient network structure into account when making its decisions.12

KPN’s decision making process is complicated by the fact that its choice of local points of interconnection will also affect the initial roll out costs of interconnecting operators.13 _In

calculating its best strategy KPN can be expected to take account of its impact on other operators’ costs as well as the impact on its own costs. The impact of this is likely to depend on the size of switch rebuilding costs. If these are large, then KPN’s desire to reduce its own costs is likely to predominate. If, however, switch rebuilding costs are small, the impact on other operators’ costs could feature more strongly in KPN’s decision making.14 _{The rating of}

11 _{The fee could also be expressed as a charge per number block.}

12 _{KPN will of course also need to take account of the cost of call conveyance to and from particular points of}

interconnection and compare this with the local interconnection charge.

13 _{It is important to note that we are talking here about the choice of the initial points of interconnection. The cost of}

any subsequent changes has to be borne by KPN under fixed fee per switch. However, the interconnecting operators bear the initial roll out costs.

14 _{For example, KPN could potentially decide to provide local interconnection at the switch via which the chosen}

subscriber numbers are currently accessed, even though that point of interconnection is not optimal from its own

FFPLS in terms of consistency with efficient infrastructure competition therefore depends on the size of switch rebuilding costs.

For the reasons given above, FFPLS provides KPN with incentives to increase the efficiency of its network. It also provides KPN with incentives to reduce switch rebuilding costs, particularly if these are large.

The regulatory burden associated with FFPLS is likely to be significantly smaller than for the Austrian solution. In addition FFPLS does not appear to give rise to the possibility of discrimination between OLOs as any decision by KPN to move the point of interconnection for a set of subscriber numbers away from the switch via which they are currently accessed requires KPN to pay for the costs imposed on the operators that are affected. As with the modified Austrian solution, it should be possible to achieve a fair balance of the costs between KPN and the OLOs.

3.4.5. Summary and conclusions

The table below provides a summary of how the different policy options rate against the various assessment criteria. A number of points need to be stressed:

•

all scores are relative to those for the status quo option which have been arbitrarily set equal to zero;

•

these ratings provide only a broad indication and rely on judgements regarding the relative importance of different factors;

•

for an overall assessment of the different policies it would be necessary to weight the scores on the different criteria together. We do not possess all the information that is required in order to do this and believe that OPTA is in the best position to make such a judgement;

•

the ratings for FFPLS will depend on the size of switch rebuilding costs. We have provided a broad indication of the situation given medium switch rebuilding costs. High switch rebuilding costs would tend to improve its efficiency ratings, while low switch rebuilding costs would tend to reduce them;

•

there is also a need to assess the consistency of the different policy options with policies adopted elsewhere. This has not been part of the remit of this study and is something that is best left to OPTA.

network point of view because the increase in the interconnecting operators’ costs exceeds the increase in its own costs (even after allowing for higher switch rebuilding costs and any subsequent point of interconnection relocation costs when the network structure becomes more efficient).

Table 3.1

Policy Option Assessment Efficient Infrastructure Provision Efficiency Incentives for KPN Switch Rebuilding Costs Regulatory Burden Fair division of costs between KPN and OLOs Discrimina-tion between OLOs Status quo 0 0 0 0 0 0 Virtual interconnection - - ++ 0 0 - - 0 Modified Austrian solution ++ + - - + -

Fixed fee per

local switch + + - + + 0

It is difficult to draw categorical conclusions. However, it would appear that virtual interconnection is not a desirable policy and is worse than the status quo. Both the modified Austrian solution and FFPLS have the potential to outperform the status quo.

The preferred option will depend on the weightings attached to the different assessment criteria. Concerns regarding the regulatory burden and possible discrimination between OLOs tend to point in favour of FFPLS. If switch rebuilding costs are high, there is probably little to choose between FFPLS and the modified Austrian solution in terms of efficient infrastructure provision. If, on the other hand, switch rebuilding costs are low, the modified Austrian solution may be superior in terms of efficient infrastructure provision.

APPENDIX A. SWITCH REBUILDING COSTS

OPTA has requested that NERA examine the economic principles that should be followed in assessing who should pay for any rebuilding costs (or system set up costs) associated with implementing local interconnection in KPN’s network.

The following analysis applies four cost recovery principles that NERA believe should be applied. These principles are similar to the six principles used by OFTEL, the UK regulator, in the context of number portability system set up costs, and were subsequently endorsed by the UK Monopolies and Mergers Commission.15 _{They were also used by OFTEL in the}

context of carrier pre-selection costs.

The analysis also draws, in places, on lessons from the experiences of the other countries under investigation.

A.1. Principle One - Cost Causality

OPTA has indicated that it intends to appoint a cost expert to determine the true cost per switch to KPN for providing the functionality required for local interconnection.

The key cost recovery principle for any fixed set up costs related to networks is ‘cost causation’: that the costs should be recovered from those whose actions caused the cost at the margin. In cost recovery exercises, this is usually the guiding principle since economic efficiency is enhanced by requiring parties to pay for the costs which they directly cause to be incurred.

Thus OPTA needs to determine the incremental cost to KPN of providing local interconnection functionality, examining whether the ‘rebuilding’ costs are solely caused by the need to provide interconnection, or whether KPN has included software upgrades and other investments that would have occurred anyway as part of the process of innovation and upgrading.

For example, their existing switches have a limited asset life 16_{. Therefore, in any case, KPN}

would be expected to upgrade a significant proportion, particularly of the older PRX/A and analogue AXE switches, in the near future.

The expected asset life and upgrade programme of KPN must therefore also be considered by the costing experts appointed by OPTA, in order that OPTA obtains a clear picture of the scale of ‘true’ incremental upgrading costs.

15 _{The other two principles are not applicable in this context.}

16 _{The switches in KPN’s network are assumed to have an asset life of 8 years in the BULRIC model.}

Under cost-causality, a further important question is whether OPTA believes the costs are caused by the fact that KPN operates in the fixed telephony market and a necessary consequence of this is having to provide interconnection to other operators, or whether the costs are caused by the competitors, since, without them, there would be no need to provide interconnection. In the former case, rebuilding costs attributable to interconnection would typically be spread between operators (including KPN) on the basis of their use of the switches concerned. Examples of this type of approach include recovery of the cost of separating accounts, and providing number portability and CPS etc, where the set up costs are recovered across all traffic minutes. In the latter case, the interconnecting operators would have to bear the costs.

OPTA will therefore need to decide how far the switch rebuilding costs should be attributed to local interconnection and what share of these costs it is appropriate for KPN to recover from other operators.

A.1.1. Past constraints outside KPN’s control

Since KPN’s network was, to some extent, developed under non-competitive conditions (it was a state monopoly until 1989 when it became a public limited liability company, and the Dutch market was only completely liberalized in 1998), there is, in OPTA’s view, a possible argument that KPN should not bear the full cost of “legacy inefficiencies” brought about by past decisions outside its control.

The argument is as follows. KPN’s existing network structure is, at least partially, the result of investment decisions made when it was a state monopoly and when there was no competition, and hence not all the costs associated with improving the efficiency of KPN’s network structure should be borne by KPN. NERA recommends that OPTA carry out an investigation into the extent of the impact of “legacy inefficiencies” on KPN’s network development and current network structure, so that it is in a position to make an informed decision on whether this issue is relevant in this context.

A.1.2. Irish case study

An interesting case study in this respect is provided by Ireland, where, as part of the recent consultations on Eircom’s latest Reference Interconnection Offer, the regulator consulted on whether a Near End Handover solution for number translation codes (‘NTC’s) would be beneficial for interconnecting operators. Most alternative operators supported Near End Handover for NTCs, citing benefits (similar to the ones in the current context) such as that it represented the best use of infrastructure rollout and it encouraged more efficient network based routing as it would enable the operators to benefit from points of interconnection at the local level.

provide Near End Handover for these services, it would be necessary to implement a double IN query at all switches. This implementation would require significant investment and rollout time. Eircom argued that as IN costs were then recovered across all calls and proportionally impacted on all operators, a cost benefit analysis would indicate no net benefits. Therefore Near End Handover should be confined to the sub-set of switches that supported it, which represented approximately 60% of all interconnect switches.

The regulator, ODTR, ruled that Near End Handover for all NTCs was a legitimate requirement, since operators should not be forced to incur additional conveyance charges which were associated with hand-over at the tandem level only. Furthermore, ODTR said that operators should not be forced to accept tandem level interconnection indefinitely solely as a result of the legacy of a decision by Eircom which was made on the basis of servicing the requirements of Eircom retail. Eircom could not penalise alternative operators for this, and therefore there was a requirement for Eircom to provide Near End Handover across the totality of their network. In the absence of a technical solution at their E10 switches (not capable of supporting Near End Handover) Eircom was directed to set their charges for “Near End” NTCs as if they were in fact “Near End” where alternative operators had a point of interconnection. Therefore Eircom was required to implement the capability for Near End Handover in all AXE switches within three months.

Furthermore, to avoid the requirement for all alternative operators to duplicate capacity at an E10 switch (at which Eircom does not physically hand over traffic) and the AXE switch at which the traffic is handed over, when Eircom receives a request for interconnection for NTC traffic at an E10 switch they must advise the alternative operator of the location at which the interconnection traffic is being physically provided.

A.2. Principle Two - Distribution of Benefits

According to this principle, costs should be recovered from all the beneficiaries. This is particular important since the investment concerned has a variety of impacts not all related to the facilitation of local interconnection.

This is likely to be case with the cost of the switching equipment. The new switching equipment that will be installed to provide local interconnection will have a level of functionality that is superior to that of the existing technology. It is possible, and indeed probable that KPN will be able to offer new services and applications on the back of the new investment. For example, KPN Broadband Unit is very likely to want to interconnect at local level following the installation of the new switches.

In addition, KPN is likely to benefit from the incremental traffic that is generated on its network. Such incremental traffic would generate new earnings, which should, according to this principle, be offset against the cost of upgrading switches.

Therefore it is important that KPN does not receive compensation from other operators (in the form of higher interconnection charges) for that part of the switch upgrade costs that relates to new services, better functionality etc, which benefits KPN’s customers and has nothing to do with the provision of local interconnection. Further cost benefit analysis would be required to determine the possible scale of other such benefits.

It is also the case that, given that local interconnection will increase competitive pressure in the market place, all consumers including those of KPN will benefit. The Association for Competitor Operators in the Netherlands, ACT, in its Response to PTA Consultation Document on Integral Tariff Regulation for End-users and Interconnection Services, January 2002, argues that the costs of the CPS system set up costs should be distributed equally across all industry players including the incumbent, since the incumbent’s customers will also benefit from the additional competitive pressures in the market:

Additionally, there are some costs that could be deemed to be of general benefit to the entire industry (and hence to all customer groups, whether of the incumbent or of competing operators). An example would be Carrier Pre-Selection (CPS) system set-up costs. The introduction of CPS benefits the competitive process, and thereby, potentially, benefits all customers in the market. To the extent that CPS creates general competitive pressures in the market, this will benefit the customers of the incumbent network operator on whom CPS obligations are imposed. Therefore, such costs should be shared equitably across all industry players, rather than only those carriers that “order” CPS from the incumbent operator. This approach has been taken in both the UK and Ireland.

A.3. Principle Three - Cost Minimisation

OPTA needs to ensure that the mechanism chosen for cost recovery provides strong incentives to minimise costs. If KPN were responsible for none of the costs, there would be no such incentives. This would suggest that (even after ignoring the fact that switch upgrades benefit the customers of KPN as well those of other operators) local switch rebuilding charges should be shared in some way by KPN and other operators. Given the need to bear a proportion of the costs, KPN would explore all technically viable and commercially viable avenues to ensure the approach taken for local interconnection was indeed the one with the minimum cost.

Various cost recovery mechanisms can be thought of which provide such incentives. For example, KPN is only able to recover the efficient Modern Equivalent Asset (MEA) cost for the local switch, regardless of the actual switch type implemented.

A.4. Principle Four - Practicability

interconnecting operators to bear all the upgrading costs up-front, since this would encourage potential interconnecting operators to temporarily stay out of the market in order to avoid paying the costs.

A.5. Conclusion

OPTA needs to determine the precise ‘incremental’ cost that is associated with local interconnection service, taking into account the costs that would have been incurred by KPN in the course of its usual network development. Cost causality is usually the guiding principle followed by regulators when allocating system set up costs, and therefore OPTA will need to make a judgment (consistent with previous judgments) as to the extent to which KPN and the competitors are the ‘cause’ of the incremental costs.

It is clear that KPN will benefit significantly from the upgrading of its local switches and subsequent local interconnection traffic, and these benefits should be taken into account when deriving what interconnecting operators should pay for the upgrading of the local switches. Furthermore, it could be argued that, as with CPS and other services, KPN’s customers will benefit from the increase in competition in the local market as much as the customers (or potential customers) of the alternative operators, and therefore KPN should shoulder part of the costs for this reason as well.

One serious issue is that potential interconnecting operators may stay out of the market to avoid initial rebuilding costs.

OPTA should assess whether these potential ‘free rider’ problems can be alleviated through a mechanism whereby the interconnecting operators that contribute initially to the rebuilding costs are able to charge a portion to other interconnecting operators who enter the market later.

It is likely that, once a full assessment of the benefits and costs to KPN of upgrading its local switches has been carried out, the cost attributable to interconnecting operators will be found to be smaller than currently expected.

APPENDIX B. COUNTRY SURVEYS

B.1. Austria

B.1.1. Interconnection regime

Until 1994, the Austrian Postal and Telegraph Administration (sterreichische Post- und Telegraphenverwaltung) operated a state monopoly offering telecommunications networks and telecommunications services in Austria except satellite services. Market liberalization in Austria has taken place in stages. First, text and data transmission services and the remaining monopoly segments of the terminal equipment sector were opened to competition in 1993. The provision of voice services for corporate networks and within predetermined user groups was liberalized in 1994. In 1996, the operation of transmission paths for telecommunications services other than public voice telephony via fixed network was fully opened to competition.

The liberalization of the Austrian telecommunications market was completed on 1st January 1998, with the elimination of Telekom Austria’s last exclusive right, which was to provide domestic and international voice telephony services to the public by means of a self-operated fixed network in Austria.

The Telecommunications Act allows unrestricted market access to all entrants who qualify under the Telecommunications Act. The principal objective of the Telecommunications Act is the promotion of competition in the Austrian telecommunications sector through regulatory measures that promote reliable, high quality and innovative telecommunication services at a reasonable price. The Telecommunications Act also aims to create a modern telecommunications infrastructure that leads to high-quality services and sustainable competition.

Other objectives include the protection of customers and operators against the abuse of significant market power. Operators having significant market power in particular telecommunications markets are subject to a special regulatory framework, based on EU Directives. The regulatory authority is required to ensure that operators with significant market power provide access to their networks, internally used services and facilities to all competitors on a non-discriminatory basis, particularly in the areas of interconnection, pre-selection of a carrier, number portability and opening of the access to local subscriber lines. Access may only be refused on objective grounds, for example based on reasons of network-security or interoperability of services.

Since the enactment of the Telecommunications Act, the minister has issued several important ordinances, including the Interconnection Ordinance which deals with interconnection under the Telecommunications Act, special network access and opening of the local network. It subjects operators with significant market power to a cost accounting

system based on the methodology of forward-looking long-run average incremental costs (FLRIC).

All operators of public telecommunications networks are obligated upon request to offer network interconnection to other operators. If the parties fail to reach an agreement within six weeks, one of the parties can refer the matter to the regulatory authority, who will then decide on the conditions for interconnection. The regulatory authority is also entitled to determine fees for interconnection if providers with significant market power are involved. The regulatory authority must publish reference interconnection offers. They also have the power to impose amendments to those reference interconnection offers at any time.

Operators must file agreements on interconnection and special network access to the regulatory authority. The terms of such agreements must be based on objective criteria and be comprehensible and interconnection and special network access must be provided on a non discriminatory equal basis.

In various decisions since March 1998, the regulatory authority has set out additional principles for interconnection. With these decisions, the regulatory authority set the tariffs for interconnection from fixed and mobile-to-fixed line and also from fixed line to the mobile networks. The regulatory authority decide on the tariffs for carrier selection, the type and scope of carrier selection to be implemented, local interconnection, mutual access to freephone numbers, value added services, shared cost services, private networks, personal numbers and access to on-line services.

Fees for call termination are element-based. Origination and termination fees are identical at the local and single tandem level, but not at the double—tandem level. The interconnection charges are based on peak and offpeak times, and the principle of reciprocity applies.

At the end of 2001, Telekom Austria had entered into interconnection agreements with 49 other fixed operators.

In order to avoid bottlenecks with point-of-interconnection links in the future, e.g. too little usage, too many orders by network operators or long delivery times, the Telekom-Control Commission ordered in 2000 that extensive arrangements be made with regard to planning, provision and usage of these scarce resources. Contractual penalties in case of late delivery or too many orders are intended to ensure that these capacities are allocated more efficiently in future.

B.1.2. Network structure efficiency

Telekom Austria is the leading provider of fixed line telecommunications services in Austria. In 2001, it was estimated by the EU to have a 85% share of the local call market by retail revenues, and a 67% local call market share by outgoing minutes.

Telekom Austria appears to be undertaking an investment programme to improve the efficiency of its network: In 2001 they reduced the switching network by 7 switches in 2001, and in 2002 intend to reduce it by more than 11 switches.

B.1.3. Local interconnection policy

The Interconnection Ordinance addresses policy for the opening of the local network through interconnection. As with other interconnection tariffs, local interconnect charges must be based on forward-looking long-run average incremental costs, as determined by the regulatory authority. These fees have been subject to several recent appeals before the Austrian High Courts.

Regarding eligibility for local termination tariffs, at the end of 2001, Telekom Austria required that interconnecting operators only roll out to 39 local exchanges out of over 200 in order to be able to use the local tariff for all their traffic. This is what OPTA refers to as the ‘Austrian Solution’. Recently, this number has been raised to 43, although the reasoning behind the change is not clear.

According to an analysis of the interconnection traffic made by Telekom Austria in 2000, a large part of the quantity growth was due to local interconnection, whereas the quantity of regional interconnections clearly went down. This may be considered as an indication that operators are substituting local for regional interconnection, or (less likely) that traffic is shifting increasingly towards those operators who push ahead with local interconnection and thus gradually expand their infrastructure network.

B.2. The

UK

B.2.1. Interconnection regime

Oftel, the UK Regulator, takes the clear view that regulation should only be imposed where it is justified considering the level of competition in the market. Too much regulation could reduce the incentives to invest and innovate but on the other side lack of regulation could harm consumers’ interests.

Interconnection Regulation from 1997 determines that operators with significant market power shall follow the principles of transparency and cost orientation for the interconnection charges. In order to ensure that the charges are cost oriented a cost accounting system needs to adopted and approved by Oftel. The Interconnection regulation does not specify the cost allocation system to be used.

system were to bring charges for interconnection services in line with what would prevail in a competitive market; and to improve incentives on BT to increase efficiencies.

Interconnection charges are regulated on the basis of baskets subject to a price cap regime. Baskets are distinguished by whether the markets are considered competitive, prospectively competitive or non competitive. Fixed call termination and originating access belong to the non-competitive interconnection services basket.

On the basis of market analysis, Oftel has concluded that BT has a dominant market position for both terminating access and originating access and that price control of these services continues to be justified.

Since the 1st _{of October 1997 the charges for fixed call termination on other operators’}

networks have been set according to an agreement between BT and the other operators which reflects the principle of reciprocity. It should be noted however that reciprocity of call termination only applies to voice and is not applicable to Internet (NTS) services.

This network charge control regime appears at first to favour the ‘buy’ decision over the ‘make’ decision since all service providers are subject to the same costs, regardless of their investment in the network. However, this has been overcome by ensuring that only those operators with Relevant Connectable System (RCS) status are eligible for these tariffs. Indeed, OFTEL’s February 1997’s statement ‘Promoting Competition in Services over Telecommunications Networks’ made clear that operators could not assume they would keep RCS status unless they were making a significant contribution to the construction of competing networks, or making a significant contribution to the development of competition in international telecommunications.

BT’s interconnection charges are widely found in surveys to be among the lowest in the world, and in particular to be the lowest within the European Union.

B.2.2. Network structure efficiency

As in Austria, British Telecom (BT) also appears to be improving its network efficiency off its own back, with a strategy to migrate an increasing number of local switches to remote concentrators, and to deploy Next-Generation Switches (NGS) in the core network. This programme is designed to accommodate forecast growth in traditional fixed and mobile voice traffic. A total of 70 (NGS) have already been deployed. In addition, more than 700 digital local exchanges have been enabled to route to the IP network all internet-type calls originating from BT customers. These calls are grouped together and routed directly to BT's Dial IP platform and other operators' networks. This enhanced functionality in the switched network is enabling the evolution from circuit switch and voice to packetised data and IP.