3.1 Historically, the main basis of costs used to regulate BT has been fully allocated costs using the historic cost accounting convention. This has been the cost base used to set interconnection charges for use of BT's network. In future, Oftel proposes to use the most appropriate costing methodology for individual regulatory decisions, but expects that this will usually be forward looking long run incremental costs. For the new regime of network controls Oftel proposes that the cost base for interconnection charges will, as far as possible, be based upon long run incremental costs.
3.2 Long run incremental cost is a forward looking concept and so is more accurately reflected by the use of current cost accounting (CCA) rather than historic cost accounting (HCA). In the HCA approach the basis for the valuation of an asset is its historic cost, its cost at the time that it was purchased. In the CCA approach the basis for valuation is the replacement cost of an asset, how much it would cost today if it were to be replaced.
3.3 Forward looking costs constitute the appropriate cost base for interconnection charges because they reflect resource costs and are consistent with the workings of a contestable market. Ideally for economic efficiency, the prices of retail services should be set in a way which encourages consumers to take account of the resource costs of their purchasing decisions. Operators would be encouraged to set efficient retail prices if they could purchase a major input (interconnection) at a charge that was set by reference to the cost of the resources consumed by its provision. Since replacement costs would be the costs faced by a new entrant, signals would be given to encourage efficient entry into and exit from interconnection services, if the incumbent's interconnection charges were set on the basis of forward looking costs. An entrant into provision of interconnection services that was more efficient than the incumbent could make a profit by setting a charge below the incumbent's charge, whereas an inefficient firm would be unprofitable if it were to match the incumbent's charge.
3.4 In its purest form, the concept of forward looking costs requires that assets are valued using the cost of replacement with the modern equivalent asset (MEA). The MEA is the lowest cost asset which serves the same function as the asset being valued. It will generally incorporate the latest available and proven technology and is the asset which a new entrant might be expected to employ. In a world in which technology is changing rapidly, it is quite likely that, for some assets, the MEA will differ from the asset that an incumbent currently has in place. For example, in recent years BT has continued to use analogue local switches to serve a proportion of customers, but the MEA for a local exchange in those years was a digital switch. In broad terms, this is the basis of BT's CCA asset valuation approach, which has been used in the incremental cost models.
3.5 Oftel and the industry have been developing a methodology to calculate incremental costs for more than two years. A robust methodology for incremental (and common costs) has now been established through the development of the top-down model, the bottom-up model and through the reconciliation work (see below). Long run incremental costs may be defined in general as the costs that are caused in the long run by the provision of a defined increment f output. The network assumed for the purpose of developing the incremental cost methodology is a standalone network of inland Public Switched Telephone Network (PSTN) services and inland private circuits. The network assumed in the methodology therefore excludes the international network and the specific equipment required for advanced services such as the Digital Derived Services Network (DDSN), the Integrated Services Digital Network (ISDN) and Virtual Private Networks.
3.6 In order to establish a robust methodology to calculate long run incremental costs, two modelling approaches have been adopted. BT has developed a top-down model, which starts from its financial accounts and removes costs that are not incremental. For a description of the top-down approach see BT's Top Down Incremental Cost Methodology, BT 1996, which is available from BT. The Incremental Cost Working Group has constructed a bottom-up model of the network components using economic engineering models that identify the engineering elements required to build a network, and has sought to populate the models with the generic costs of a fully efficient operator. For a documentation of the bottom-up model, see Long Run Incremental Costs: The Bottom-Up Network Model, Oftel, 1996, which is available from Oftel.
3.7 Both approaches have had a role to play in deriving the robust methodology and the estimates of incremental cost through the third essential element in establishing the methodology: the reconciliation and production of hybrid figures. Each model has its own strengths and weaknesses and each therefore provides a useful cross-check on the other. The results from both approaches have been reconciled and hybrid incremental cost figures derived for 1993/94 and 1994/95 (see paragraphs 3.9-3.12 below).
3.8 In the methodology to calculate incremental costs, the increments have been defined as the whole of the output of two services: conveyance and access. The long run incremental cost of conveyance is the cost that would be saved in the long run if no traffic were provided over the network, but access were to continue to be provided. The long run incremental cost of access is the cost that would be saved in the long run if no final links to customers were provided (but, hypothetically, conveyance continued to be provided). If there are economies of scope between conveyance and access, it will follow that there are common costs of the network, i.e. costs of the standalone network that are incurred when conveyance and access are provided simultaneously, but would not be saved if either were no longer provided.
3.9 The output of the incremental cost work is the production of hybrid figures, which provide the best available measure of BT's relevant, incurred incremental and common costs. The incremental cost models are static in that they are designed to measure costs in a particular year. Neither the top-down nor the bottom-up model is designed as a forecasting tool. The hybrid results will feed into Oftel's financial model (described in Chapter 5), which is a dynamic model that provides forecasts of the costs and revenues of BT Network, to enable Oftel to produce the initial charges (i.e. as at 1 August 1997) for all of the interconnection services in the network baskets. The intention of the hybrid incremental cost figures is to include all relevant costs, measured using the appropriate methodology and reflecting BT's incurred costs, not the costs of a fully efficient operator.
3.10 The hybrid figures should include only those costs that are relevant to the network. The principle of the relevance of certain costs has been raised in the context of the determinations of charges for BT's Standard Services. Oftel has not yet taken a final view on such exclusions in that context or in this review. But for incremental costs, Oftel has approached these issues afresh. The hybrid figures will include only those costs that are, in Oftel's view, relevant to the network. As well as obvious exclusions, such as any retail costs, Oftel may exclude some costs that BT might wish to include, for example elements of redundancy costs, and some of the financing costs of short term investments and cash balances. The treatment of these costs in the context of the reconciliation of the bottom-up and top-down models is also referred to in the NERA report which is to be published shortly.
3.11 The hybrid figures should relate to BT's incurred level of relevant operating costs, not the costs of an efficient operator. This is because Oftel's proposal is that the initial charges in the network baskets will reflect incurred costs and that inefficiencies will be driven out of the system over the period of the price control, i.e. the value of X will take account of the estimated 'efficiency gap' as at 1 August 1997 (see Chapter 5). However, for the valuation of assets, the hybrid figures should be forward looking using the modern equivalent asset approach, because this provides the appropriate economic measure of incurred resource costs. The hybrid figures represent the best available measurement of incremental costs, because they have been derived by drawing upon the different strengths of each of the top-down and bottom-up models.
3.12 The latest report by NERA (to be published shortly) provides advice to Oftel on the hybrid figures for incremental costs, common costs, and floors and ceilings for 1994/5. It also sets out in detail how these hybrid figures can be derived from the top-down and bottom-up models. Copies of this report will soon be available from Oftel's library.
Table 3.1 Hybrid figures for 1994/95: cost of interconnection services (before markups)
Pence per minute
Local exchange segment 0.404
Local-tandem conveyance 0.234
Inter-tandem conveyance 0.446
Single tandem segment 0.638
Double tandem segment 1.084
Notes: The cost of a single tandem segment is given by the sum of the costs of the local exchange segment and local-tandem conveyance. The cost of a double tandem segment is given by the sum of the costs of a single tandem segment and inter-tandem conveyance.
Costs exclude interconnection specific costs (known as Product management, Policy and Planning).
3.13 During the course of the incremental cost work Oftel has tried to ensure that the process has been as transparent and widely understood as possible. The Incremental Cost Steering Group was set up more than two years ago and all operators were invited to join and attend meetings. The Group has typically met on a monthly basis to discuss the general principles of the methodology, to comment upon the workings of both the top-down and bottom-up models and to advise on the reconciliation process.
3.14 The Incremental Cost Working Group, which is chaired by Oftel and includes representatives from six operators, was charged with the task of developing a bottom-up engineering model. The Group has met frequently, often on a weekly basis, to work through the detailed issues.
3.15 The top-down model has been developed by BT. One of the key problems has been that other operators are largely unsighted on the workings of the model, because it contains commercially sensitive information and so has not been able to be inspected by other operators. However, the economic consultants, NERA, have been employed to carry out a thorough investigation of the top-down and bottom-up models on behalf of the Steering Group and report upon the strengths and weaknesses of each model. NERA's report (The Methodology to Calculate Long Run Incremental Costs) was published by Oftel in March 1996. Oftel intends also that BT's detailed documentation of the top-down model will be published early next year, to allow other operators to understand and to comment upon the workings of the top-down model in much greater detail.
3.16 Oftel has also published two reports by NERA on the reconciliation work: for incremental costs in 1993/4 (Reconciliation and Integration of Top Down and Bottom Up Models of Incremental Cost), published in June 1996, and for incremental costs in 1994/5 (Reconciliation of Top Down and Bottom Up Models), to be published shortly. These set out in detail the further improvements that have been made to the two models and how the model results have been reconciled and hybrid figures produced. Copies of all three reports will be available from Oftel's library.
3.17 In addition to the published reports and the industry involvement in the Steering and Working Groups, Oftel has held two workshops in April and October of this year to discuss the incremental cost work. Both workshops were well attended by a wide cross-section of industry participants. A further workshop for the industry on 9 December will also include discussion of the incremental cost work.
Oftel will continue to seek to ensure that all operators have an understanding of incremental costs and would welcome suggestions about the ways in which transparency could be further improved.
3.18 The reconciliations between top-down and bottom-up models for 1993/4 and 1994/5 have established not only hybrid figures, but also the methodological changes and adjustments required to either model to derive those figures. Therefore, by incorporating these 'hybrid adjustments' in the top-down model it will be capable in future years of deriving hybrid figures, on which Oftel will base the initial charges for the services in the network baskets. To provide operators with confidence and to ensure that the nature of the top-down model has not been changed between years, the model results will be audited.
3.19 To produce the necessary incremental cost information to allow Oftel to set the initial charges (at 1 August 1997) for interconnection services in the baskets, the top-down model will be updated to 1995/6 and also to interim 1996/7 (i.e. the first six months of the financial year, April-September). Hybrid results will be available by incorporating in the model the hybrid adjustments specified by Oftel. The model results will be audited to a robust standard in accordance with the detailed documentation of the top-down model. As an additional check, the model results will also be scrutinised by NERA. The detailed documentation, excluding commercially sensitive information, will be published early next year.
3.20 Oftel recognises that a key requirement of this process is that operators have confidence in the validity of the resulting hybrid incremental cost figures. Table 3.2 lists the information and reports that are to be provided to Oftel by BT. Oftel is currently considering what further work or additional requirements would contribute further to the objective of providing confidence.
Oftel invites comments on how confidence in the incremental cost work and the hybrid figures could be further improved.
Table 3.2 Provision of information by BT to Oftel
Information/Report Deadline Update of the top-down model to 1995/96 29 November 1996 (Received) Audit of the 1995/96 top-down model results 31 January 1997 Update of the top-down model to interim 1996/97 7 April 1997 Audit of the interim 1996/97 top-down model results 6 May 1997 Final DAM and detailed documentation of the top-down 6 May 1997 model on which audited results are based
3.21 In the previous consultative document Oftel had proposed that the initial charges for individual interconnection services in the two baskets at 1 August 1997 would be set by BT, subject to an allowable revenue set by Oftel. However, Oftel's proposal is now different: Oftel proposes to specify the initial charge for each interconnection service in the two baskets. The reasons for this change are the importance of having the most up-to-date information possible when setting the initial charges and allowing interconnecting operators fair warning of the initial set of charges.
3.22 As shown in Table 3.2, audited incremental cost information for interim 1996/7 will not be available until May 1997. Given the period required for statutory consultation, the licence modifications could not be agreed until sometime in July. If Oftel were to set initial allowable revenues, leaving BT to set the initial charges for individual services, the set of charges would be known to interconnecting operators at most four weeks before its introduction and very probably with a significantly shorter notice period. However, if Oftel were to set the initial charges for individual services, this set of charges would be known at the time of the statutory consultation, since it would be contained in, or annexed to, the licence modifications.
3.23 The earliest date that BT would be permitted to alter the initial set of charges would be 90 days after the agreement of the licence modifications (if notification were given to interconnecting operators of the proposed changes on the day that the licence modifications were agreed).
3.24 The change in the average charge for services in the basket during the first year of the network controls will, as in subsequent years, be governed by the RPI-X constraint and the constraint on the weighted average date. In the first year, the RPI-X constraint will require that by 31 July 1998 the reduction in the average charge in real terms must be at least as large as X compared to the average charge on 1 August 1997, with the weights to be applied to the individual services given by their respective shares of revenue in the prior financial year, 1996/97. There will also be a second charge control constraint concerning the timing of changes in charges, the so-called 'weighted average date'. This requires, in addition to the obligation to meet the RPI-X constraint, that the effect of all changes in charges in the control year, whenever they occur, is at least as large as if the charge reductions (in real terms) had occurred on 1 February, the midpoint of the network control year. This constraint, which mirrors a provision for the retail price cap, will prevent BT from complying with the network caps, for example, by keeping charges high for the whole of the control year and only reducing them to the level required by RPI-X on 30 July.
3.25 For each of the services in the call termination and general network baskets Oftel will set an initial charge for 1 August 1997. The charge for each service will be given by Oftel's forecast of the initial unit cost of that service. The unit cost of each interconnection service will be calculated as a forecast of BT's relevant incurred average cost of the service, derived from the incremental cost of conveyance, plus an equal proportionate markup for the recovery of the common costs between conveyance and access. The forecasts will be generated by integrating the static measure of incremental cost into Oftel's dynamic financial model (see Chapter 5).
Equal Proportionate Mark-up
3.26 The equal proportionate markup is to determine how much of the burden of recovery of the common costs between conveyance and access should fall upon conveyance, and how much upon access. The equal-mark approach is a practical, if broad-brush, way in which to determine the split of common costs. Using the figures derived for 1994/95 it would operate as shown in Table 3.3 (below).
3.27 Oftel is aware that the common costs have so far been subject to less scrutiny than the incremental cost of conveyance. Work has already commenced to examine the source and relevance of the common costs in the Incremental Cost Working Group, although this examination is at an early stage. Oftel hopes that this work will be completed early next year.
3.28 By defining the network in the incremental cost methodology as a standalone inland network of PSTN services and private circuits, all of the costs that are common between services not included in the model (e.g. IDD conveyance) and the inland network are included in the inland network costs shown in Table 3.3. The services outside of the model are treated, therefore, as 'marginal' and are assumed to make no contribution to any common costs that they share with the inland network (see Figure 3.4, below). There is an argument that the services outside of the model should bear some proportion of the common costs that they give rise to. This could be implemented by using the available evidence to estimate the likely size of such common costs and then applying an equal proportionate markup to split them between the relevant service and the inland network.
3.29 Oftel intends to carry out some further work on this issue. However, the reduction in the cost base for inland conveyance that will result is unlikely to be substantial. This is because the cost of the inland network is large relative to the cost of the services outside the model and, therefore, the vast majority of the common costs would, under equal proportionate markups, fall to be recovered from the inland network. Unless there are substantial common costs between services outside the model and the inland network, the effect on the cost base for inland conveyance would be small (perhaps a reduction of 1% or less).
Table 3.3 Equal proportionate markups: indicative figures for 1994/95
£ million Incremental Cost Common Cost Cost including
before mark-up apportioned markup
Inland Conveyance 1,265 208 1,473
Inland Private Circuits
(conveyance related costs) 292 48 340
Access 3,541 583 4,124
Common Costs between
Conveyance and Access 839 n/a n/a
Total 5,937 839 5,937
Note: The equal proportionate markup rate is 16.5% [=839 / (1265 + 292 + 3541)]
Figure 3.4 Illustrative example of common costs between inland conveyance and other services (outside of the incremental cost models)
The Markup on Network Components
3.30 Having determined the cost base for inland conveyance, i.e. incremental cost plus equal markup, the next task is to decide whether all network components or interconnection services should contribute equally to the recovery of the common costs that have been apportioned for recovery from conveyance. If only a subset of the network components gives rise to the common costs between conveyance and access, Oftel would consider that only these components should bear a mark-up. The figures in Table 3.5 reflect an initial consideration of the common costs between conveyance and access, in which no common costs were associated with the tandem switch, the junction transmission link and the trunk transmission link. It is possible that this view may need to be modified following further work on the source, nature and relevance of the common costs.
3.31 The costs of interconnection services, including mark-up, for 1994/95 are derived by applying the 1994/95 routing factors to the component costs in Table 3.5 - this is set out in Table 3.6.
3.32 The initial charge in each safeguard cap of RPI+0% will be the Interim Charge that has been determined by the Director General for 1997/98 under the current interconnection regime (fully allocated costs using the historic cost accounting convention), which will be superseded by the new regime of network controls on 1 August 1997. It should be recalled that the safeguard caps are neither intended nor expected to provide the binding constraints on the charges for the services covered - competitive pressures should force BT to move its charges for such services towards forward looking incremental costs after August 1997.
3.33 Oftel proposes that the interim charges determination for 1997/98 should be the last determination of interconnection charges. It is minded, therefore, not to make a final determination of 1997/98 charges, which might involve retrospective adjustment of the interim charges for the four month period, April-July 1997. Oftel considers that it would be appropriate for there to be a clean break between the current regime, based on determinations, and the new network control regime commencing on 1 August 1997. This approach would avoid Oftel carrying out a retrospective determination at the end of 1998, well over a year into the new regime. Moreover, that determination would only apply to charges in a four month period.
Oftel would welcome comments on its proposal that the interim charges determination for 1997/98 should be the last under the current interconnection regime, i.e. that there should not be a final determination of interconnection charges for 1997/98.
Table 3.5 Applying the markup to the inland conveyance network components: indicative figures for 1994/95
Pence per minute Component cost Component cost
before mark-up including mark-up
Local switch 0.275 0.334
Tandem switch 0.084 0.084
Junction transmission link 0.083 0.083
Junction transmission length 0.0082 0.0100
Trunk transmission link 0.079 0.079
Trunk transmission length 0.0016 0.0019
Notes: The markup rate on local switch, junction length and trunk length is 21.8% to allow the recovery of common costs of £208m.
All cost figures exclude interconnection specific costs (known as Product management, Policy and Planning).
Interconnect specific services
3.34 As discussed in Chapter 2, Oftel proposes that the charges for two data management amendment services and the charges for interconnection circuits should be indexed by RPI-X. For each of these services, Oftel proposes that the starting charge should be an estimate of the incremental cost derived from dis-aggregated accounting information. The DMA services involve specific tasks with a high labour content; many of the interconnection circuit charges are made up from the capital and maintenance costs of specific pieces of equipment. By examining in greater detail the nature and relevance of the indirect and overhead costs included in the fully allocated costs of these services, it will be possible to construct reasonably reliable estimates of their incremental costs.
3.35 From 1997/8 Oftel has proposed that BT should change from the current set of network inland conveyance components to a more transparent and further unbundled set of components (see Annex E for a list of the current and new components). This would involve two changes:
Table 3.6 Cost of interconnection services, including markups: indicative figures for 1994/95
Local exchange Local-tandem
segment conveyance
Routing factors
Local switch 1.016 0.211
Tandem switch 0.789
Junction transmission link 0.580 0.483
Junction transmission length 9.394 4.449
Trunk transmission link 0.205
Trunk transmission length 11.111
Price per minute
Cost of interconnection service including mark-up 0.481 0.259
Fully allocated costs under the historic cost accounting convention 0.517* 0.290*
Notes: * To allow direct comparison with the incremental cost based figures, the fully allocated cost figures have been recalculated using a 12.5% nominal cost of capital.
Cost figures exclude interconnection specific costs (known as Product management, Policy and Planning).
3.36 The current approach, in which use of the local switch (but not the concentrator) is counted, leads to inaccuracies in measuring the cost of, and charging for, calls. There are two main sources of error:
BT own exchange calls
3.37 A BT end to end own exchange call uses one local switch and two concentrators. However, with the current set of components, only use of the local switch is counted. BT Retail pays for use of one local switch and, the built-in use of only one concentrator, even though two concentrators are used. With total cost recovery overall, this implies that the charge for a local switch and a concentrator, e.g. in a local exchange segment, is slightly higher than it otherwise would be.
3.38 This point has been recognised in the determinations in the current regime. OLO payments for interconnection are reduced by a small amount (about 0.7% in the 1996/97 interim determination) to attempt to correct for this error. For the new network control regime Oftel is currently considering the appropriate form of the adjustment.
3.39 One possibility would be that, for as long as charges under the network caps are to be set using the current components, by Oftel for the initial charges and by BT for subsequent charges, the same percentage reduction in the component rates as in the latest determination (interim 1997/98) should be retained. However, this adjustment will have been derived using information on fully allocated costs and will not be derived from information on forward looking incremental costs. Other possibilities would be to:
Oftel would welcome comments on the appropriate adjustment to correct for the underpayment by BT Retail on own exchange calls, which will arise during the early period of the new regime, in which the current set of components will continue to be used to set interconnection charges.
DLE as a Tandem Switch
3.40 In some parts of the country there is a shortage of tandem switching capacity - often because of the level of demand for tandem interconnect from OLOs and so DLEs are sometimes used to perform the tandem function (though only for BT-BT calls). DLE as a tandem switch is the middle DLE in a routing of DLE-DLE-DLE. As a result, the routing factor for the single tandem segment includes some usage of DLE as well as DMSU (in the 1995/96 routing factors the single tandem segment contains DLE as tandem for 12.5% of call minutes and use of a DMSU for 87.5% of call minutes).
3.41 In the current set of components, there is no separate component for the concentrator, so the cost of the local exchange includes the cost of the concentrator. Currently the DLE when used as a tandem switch is charged for in the same way as any other use of a DLE, even though there is no usage of a concentrator when the local switch performs a tandem function. Since a DLE costs more per call minute than a DMSU, this inflates the cost of a single tandem segment. With total cost recovery overall, if DLE as a tandem were to carry a lower charge, the price of normal use of a DLE, e.g. in the local exchange segment, would be increased. Overall the same amount of cost would be recovered, but the current regime does not appropriately reflect the pattern of costs and particular operators may be paying more or less than they should. This issue has recently been under discussion in the Interconnect Policy Forum in the context of the determination for 1995/6.
3.42 With the change in the set of components from 1997/8, when the concentrator will be separately identified from the local switch (see Chapter 3), this problem will be considerably reduced, but it will not disappear. When the concentrator is split off from the local switch, the extent to which the local switch costs more than the tandem switch will reduce substantially, although it is likely still to be more costly than a tandem switch minute.
3.43 Oftel proposes that, since a tandem function is being served, the charge for the DLE when used as a tandem switch should be the same as the charge for a tandem switch. This approach is consistent both with the forward looking long run basis of costs in the new regime and the structure of the proposed network controls, in which services are identified by the function that they perform in the network (e.g. distinction between call origination and call termination). Since the cost base of the new network controls will be quite different from the cost base currently used for interconnection charges, a different approach to charging for DLE as a tandem switch might be appropriate in the context of determinations in the current regime.
Oftel invites comments on its proposal that the DLE when performing a tandem function should carry the same component rate as a tandem switch.
3.44 As for the unbundling of the concentrator from the local switch, transparency and accuracy are the key reasons why a reclassification of the transmission components is proposed. The current distinction between junction and trunk transmission is historically derived and does not map neatly onto the current network architecture. For example, transmission between a DLE and a DMSU (local-tandem) sometimes involves equipment classified by BT as part of junction transmission and sometimes equipment classified as part of trunk transmission. Transmission between two tandem switches (inter-tandem) typically uses trunk transmission, but uses junction transmission for 9% of call minutes (in the 1995/6 routing factors). The categorisation of the new components will ensure that data is collected on a basis that is much more consistent with the classification of interconnection services (by function in the network) in the new network controls.
3.45 Remote-local transmission is transmission between the concentrator, when remotely located, and the local switch (some concentrators are located in the same place as their host local switches). In the current set of components, remote-local and local-tandem transmission are essentially bundled together as junction transmission (although, as noted above, local-tandem will also include transmission currently classified as trunk). Consequently, remote-local transmission is shown as having a very similar cost per link and per kilometre as local-tandem transmission. However, the bottom-up model, which has explicitly modelled remote-local and local-tandem separately, suggests that remote-local transmission may be substantially more expensive than local-tandem transmission. Unfortunately, because the top-down model does not address this issue, it has not been possible in the reconciliation between top-down and bottom-up models to evaluate the robustness of this cost difference. It appears likely that if BT were to measure the costs of remote-local and local-tandem transmission separately, remote-local would be more expensive, because remote-local transmission involves thinner routes, less sharing with private circuit traffic and less duct sharing than local-tandem transmission. However, quantifying the difference robustly has not been possible in the absence of BT measuring costs in a different way (i.e. reclassifying the transmission components).