Annex 1: Specific comments on Draft Model Reference Paper. Part A. Common guidelines

Annex 1: Specific comments on Draft Model Reference Paper Part A. Common guidelines Criterion CG 2 The models should be based on forward-looking long run average incremental costs. No migration costs should be included. The models should allow coverage of common costs. These costs should be shown separately. In the explanatory text preceding this criterion it should be emphasized that it is consistent with forward looking costing to utilize depreciation methods (e.g. economic depreciation) which partly rely on historical levels of expenditure or cost recovery. The deciding factor should be whether the historical periods are consistent with forward looking costing over the entire duration considered. Criterion CG 3 For the core network, the increment should include all services that use the core network. For the access network, the increment should include all services that use the access network. The LRAIC of co-location is the cost incurred in providing co-location services. These definitions include the services that the SMP operator s network division provides to its own retail division as well as services to other operators. While the Forum agrees with this criterion, it should be emphasized that all former versions of NITA s fixed network costing models has failed to be in compliance herewith. A range of TDC s fixed line products and services has at any given time been outside any real regulatory control and every time NITA tries to patch one hole, TDC digs out new ones by inventing new service fees. One of the most glaring examples of this seemingly never ending patchwork game is TDC s introduction of a new set of terms for access line fault repair in the reference offers. Prior to this change, a relatively high fault repair service level was included in the LRAIC regulated LLU prices. This standard service has, however, been downgraded to a best effort fault repair standard service which in practice means that TDC does not guarantee any specific service level. At the same time TDC introduced a new set of higher service level agreements that can be chosen optionally but at an additional cost. Arguably the most significant part of operating expenses in the access network is related to fault repair and new (expensed) installations at customers premises. Therefore, it seems reasonable to expect that by lowering the standard service levels, TDC will experience either significant cost savings or revenue gains (if the retail and wholesale customers choose to purchase the supplementary (higher) service level products). In fact this expectation is very much supported by the fact that TDC has given notice to wholesale customers that in 2009, a surcharge of DKK 20 per quarter for every full copper line will be applied for keeping the original so-called 5/12 fault repair service level. Unfortunately NITA has so far failed to acknowledge that this type of tampering represent a fundamental change of the underlying premises for the LRAIC costing and in fact a clear-cut breach of criterion CG3 above. Clearly, not all of TDC s fixed products do necessarily have to be explicitly price regulated by the LRAIC model. However, to enable any reasonable precise and consistent LRAIC costing, it is a prerequisite to

comply with criterion CG3 and hence to include updated and even forward looking costing/revenue data on all significant services that share costs with the LRAIC regulated ones. Given that it is not possible in practice to update the model continuously or to foresee all TDC s future tampering plans, it is very important that NITA at least takes the necessary steps to limit the possible costing impact. That could for instance be done by demanding that changes in reference offers which impact the costing of LRAIC regulated service can only take place in some pre-specified windows and only after a proper public hearing. Furthermore, as the fault repair example shows, there is clearly a need for a more comprehensive and explicit product descriptions underlying the costing of the LRAIC regulated services than is currently the case. Criterion CG 6 The models should include all standard PSTN/ISDN and Broadband services. Criterion CG 7 When dimensioning the network, the leased-lines traffic volume should include leased lines provided to retail customers, to other operators and to the network operator. Leased lines used by the network operator should not be double counted as other services. The models will not need to calculate the costs of leased lines explicitly. Leased lines should only be included for dimensioning of the network and for ensuring that a fair amount of shared costs are allocated to leased line services as well. The model should distinguish between leased lines used in the core network and leased lines in the access network. Furthermore, to enable allocation of shared trenching and ducting costs, distinctions should be made between wired leased lines (copper, coax and fibre) and wireless leased lines. Allocation of other shared and common costs would clearly also benefit from an explicit costing of leased lines in the model. Criterion CG 11 The models should model the costs for 2007. In subsequent years, the base year will be adjusted accordingly. Criterion CG 12 The models should use a replacement-cost concept to estimate current costs. Replacement cost in the revised hybrid model should correspond to the cost of buying new equipment in the base year. For the top-down model, replacement cost should be based on CCA values calculated using the MEA. For properties in general, the building costs should be valued in accordance with the method previously developed by NITA. With this method, the public property values are adjusted by a factor representing the difference between the market price and the public valuation in each geographical area In the LRAIC Forums view copper based access will continue to play a significant role for at least the next 10-15 years. However, there is very little doubt indeed that the decline in the demand for copper based access will continue and probably even accelerate during this period. There is of course nothing unusual in the fact that an old technology will eventually be phased out and substituted with a modern wireless and fibre based access that better meets future demands. This development does, however, pose some challenges for LRAIC based pricing. In particular, the current single year cost modelling practice of the access network needs to be replaced with a multiyear approach that can handle technology transitions consistently.

The existing copper in TDC s access network dates largely back to the 1950s, 1960s and early 1970s. If it is assumed that copper access is completely phased out within the next 15-20 years it seems reasonable to assume that the average economic lifecycle for copper access assets (including trenching and duct) will turn out to have been at least 50 years. A consistent treatment of the copper access lifecycle - that also at least partially takes the future transition from copper to fibre into account - therefore requires a modelling of cash flows and demand for perhaps a 30 year historic period and 20 years into the future. This stands in contrast to the continued use of the current static single year model where the crucial underlying assumption is that the economic environment is relatively stable in perpetuity. If this assumption is not met - which is clearly not the case here - the single year approach seizes to produce valid costing results. In particular, if a static single year model is used in the latter stages of a technology cycle with relatively low demand, the model will obviously produce very high unit costs. To get an idea of just how incapable the current static model is in handling technology changes, the LRAIC Forum has done some costing simulations for variations in the demand for copper lines. The observation that stands out is that the total capital expenditures allocated to the access network are not very sensitive to variations in output and that this reflects that trenching and digging usage are currently treated as a constant in the model. In particular, a 10 per cent decrease in the demand for copper lines reduces capital expenditure by less that 2 per cent and an 80 per demand decrease reduces capex by approx. 15 per cent. Clearly, these cost volume relationships are very discomforting and considering that the demand for copper lines has already decreased by 23 per cent since 2003 this is not only representing a problem for future pricing. This model error is so significant that it should be corrected immediately. The challenges in modelling a fixed access network is conceptually very similar to the challenges entailed in modelling the transition from a 2G to a 3G network in a mobile LRAIC model. In the mobile LRAIC model, NITA decided to solve this challenge by explicitly modelling the full lifecycle for both technologies including assumptions on the future migration profile. NITA is recommended to be consistent and take a similar approach to the modelling of the fixed access network. Regarding the concept of replacement costs, it should also be noticed that it is inconsistent on the one hand to argue that asset replacement costs in the revised hybrid model should correspond to the costs of buying new equipment and then on the other hand to calculate the corresponding operating expenditures based on input from TDC s separated account from 2007. First, it is not correct to assume in the first place that the modern equivalent asset for a, say, 50 year old copper access network is a new copper access network. The stream of services that can be supplied with a sufficient quality would clearly be much improved with a new copper network. Second, if the copper access network was indeed brand new, the expenses spent on operation and maintenance would be of a completely different magnitude than TDC s current experience with an old worn down copper access network. Criterion CG 13 Both models should only include efficiently incurred operating costs. Operating costs related to retail activities should either not be included or should be clearly marked such that the retail cost element of every product is very apparent.

NITA should pay more attention to the identification and correction for operating network costs (e.g. fault repair or customer installations) that are recovered through separate fees (wholesale or retail) or third party damage compensation. Criterion CG 14 For the top-down model, the annualisation approach used in the statutory accounts should be adopted (straight line depreciation), modified to incorporate CCA adjustments. To facilitate comparisons with the revised hybrid model, the top-down model should also include the ability to use tilted annuities. The revised hybrid model should have the flexibility to use both straight-line depreciation and (tilted) annuities, with the latter being the default approach. In addition, to the extent practical, the revised hybrid model should also include the facility to use economic depreciation. The LRAIC Forum strongly objects to the continued use of tilted annuities to annualize costs in the hybrid model. In the model specification paper for building the hybrid mobile LRAIC model published by NITA in May 2007, it was explicitly stated on page 54 that Economic depreciation is the recommended approach for regulatory costing and further that only economic depreciation considers all potentially relevant depreciation factors. NITA correctly concluded that the primary factor in the choice of depreciation method is whether network output is changing over time. If network output in relatively constant over time, tilted annuities can be chosen as a reasonable proxy for economic depreciation. However, where demand is varying over time, results using tilted annuities will differ significantly from economic depreciation. Since 2003 the demand for copper access has declined by 23 per cent, a development that is expected to continue and probably even accelerate further over the next 10-15 years. Thus, there is no reason to expect a relatively constant network output over time, which disqualifies the use of tilted annuities as a proxy for economic depreciation. Whether the practical implementation of economic depreciation is straightforward and convenient to NITA or not, is in this respect irrelevant and a model criterion is no place for making such reservations. It should also be stressed that a consistent use of economic depreciations imply estimating cash flows and demands for both future and past periods that belong to the same technological lifecycle. In particular, it will be very incorrect to only apply the economic depreciation framework to the remaining part of the copper access lifecycle. In the mobile LRAIC framework the analogy would have been to base the termination charges on a 2G lifecycle from 2008 until 2013 instead of on the full expected lifecycle from 1992 until 2013. Criterion CG 15 Both models should initially use 9.5% as the cost of capital. NITA will inform on the final cost of capital that will be used in the revised hybrid model at that time. The LRAIC Forum does not object to the initial use of a dummy value for the cost of capital. However, the Forum does expect eventually to be given an opportunity to comment in detail on the underlying principals to be used instead of simply being informed on the final cost of capital that will be applied.

Criterion CG 19 The total annualised cost of the actual (raw) copper pair should as a starting point be the same whether it is used for providing PSTN services, full access, shared access or Bitstream access plus PSTN services. If it is decided to cost Bitstream capable copper pairs separately from non-bitstream capable copper pairs then this must be justified and documented. The total annualized average cost of a copper pair does in practice depend on the service provided. PSTN coverage is available nationwide whereas xdsl coverage by means of full or shared access is not possible in certain remote areas where the cable length exceeds 5-7 km. It is clearly not reasonable that a broadband provider should subsidize TDC s PSTN service obligations in remote areas where it is not possible to supply broadband in the first place. Hence, a differentiated costing of copper pairs that take this into account should be preferred. Criterion CG 22 Co-location products should be costed individually, and the costs should be disaggregated to provide as much information on the different types of costs incurred as possible. At a minimum, the costs of accommodation, the costs of installation, the costs of power, the costs of security, the cost of equipment and other costs should be shown separately. Similarly as for the core and access increments, the co-location modelling should include all significant services that are included in this increment. Compared with the current practice there is a need for a more transparent and true modelling of existing LRAIC regulated service such as in-span cabling and power supply to e.g. take account of reuse of existing facilities. Furthermore, the inclusion of additional services such as 48V 30A and 48V 60A power supply is long over due. Part B. Specific guidelines for top-down model From prior experience the LRAIC Forum is seriously concerned that the transparency and hence outcome of the process will be seriously hampered by the fact that the starting point for the LRAIC modelling apparently again will be a TDC top-down model that TDC undoubtedly will insist should be black boxed for other parties. Given that fully depreciated asset are not supposed to be attributed any value in a top-down model, one would normally expect that a top-down model should produce results quite similar to FAC models. This was, however, by far not the case in the process during 2000-2002 where TDC somehow succeeded in demonstrating that the top-down based LRAIC cost for e.g. full access should be several times higher than the highest FAC they had ever postulated prior to that. The LRAIC Forum has noted that NITA has made a formal decision regarding the submission of a top-down model from TDC. However, this potential overload of unverifiable and irrelevant information will hopefully not divert NITA s attention from what should be primary focal points, e.g.: Collection and validation of categorized historical capex and opex cash flow estimates for use in a consistent application of the economic depreciation methodology auditing of TDC s separated accounts with particular focus on allocation principles employed and elimination of operating network costs that are recovered from separate fees like fault repair and installation fees

collection of relevant network information, e.g. long range cable management plans to enable a proper modelling of the future changes in the length of and boundaries between the core and access network. Criterion TD 3 The gross asset valuation in the top-down model should reflect the replacement cost of the modern equivalent assets. The estimation of replacement costs of MEA will in practice have so much built in leeway that it is likely to be of no significant value. This criterion should therefore at least be supplemented with a criterion asking for a full time series of historical capital and operating expenses covering, say, 1980-2007 including a mapping to different relevant network elements. Focusing on actual cash flows would also in effect render superfluous most of the subsequent criteria regarding asset prices, MEA adjustments etc and hereby limiting the complexity of the process. Criterion TD 6 The MEA definition, that the top-down model should use, is that of an asset that can produce the same services produced by the existing asset at lowest cost, adjusting where possible to reflect differences in operating costs, quality, asset lives and space requirements. Adjustments to take into account differences in service quality and functionality, operating costs and space requirements should be undertaken. MEA adjustments to take into account differences in service quality and functionality, operating costs and space requirements have historically been more or less ignored to the detriment of alternative operators and competition in general. By nature it is difficult to establish hard facts for a highly hypothetical scenario. This should, however, not be used as an excuse for not taking into proper account that there will indeed be significant differences in service quality and functionality between an old worn out network and a brand new one. Similarly, operating costs in a new network would in all likelihood have to be measured on a completely different scale than TDC s current operating costs. A LRAIC costing based on capital expenditures from a hypothetical new access network and operating expenses from an old access network is a cocktail that will clearly not support any future competition in these fixed markets. Criterion TD 9 An SMP operator should show the utilisation level for exchanges, transmission equipment, optical fibre, copper cable, buildings and other significant asset categories and justify why this is efficient, for example by drawing comparisons with current deployment practices. Criterion TD 10 SMP operators should justify their ratio of actual number of pairs to subscriber lines in different parts of their access network and in different geo-types. Where the telecom network and cable TV network share duct and trench costs should be split between pro rata to their respective number of customers. If a single year approach is taken and demand is expected to decrease, it is inconsistent to include any inactive lines in the top-down modelling. For instance, it is clearly of no value and hence inefficient to include trenching, ducting and cable costs to households that are no longer active customers.

Criterion TD 14 The top-down model should use SDH equipment as the MEA for any remaining PDH equipment, except possibly for low capacity routes. If the MEA concept is to be taken seriously, both SDH and PHD equipment will have to be replaced with modern transmission equipment in the top-down model. Criterion TD 16 Top-down models should identify trenching costs for different terrain types, and for ducted compared to direct buried cable. The model documentation should provide supporting evidence to justify the relative proportions of the different terrain types and also explain the rationale for any differences in costs in different parts of the network for the same terrain type. If trenching, ducting and cable costs should only reflect the costs of a modern day network, no value should be attached to inactive lines. In particular it is important that costs associated with the last mile from street cabinet to the customer s premise is either identified separately for both inactive and active customers or more correctly be assumed recovered in full directly through individual connection fees. Criterion TD 23 The top-down model should only include efficiently incurred operating costs (including any costs that arise because of legal obligations). The model documentation should provide evidence that the operating costs are based on efficient operating practices. While TDC s incurred operating costs might not be significantly inefficient given the vintage of the actual network, it is important to be aware that this cost level gives very little insight into the efficient cost levels for operating a modern core-network and a new copper access network. Part C. Specific guidelines for the revised hybrid model Criterion RH 1 The starting point when building the revised hybrid model is the level of demand in Denmark for all the services using the access and the core network of an SMP operator with a Universal Service Obligation along with an allowance for growth. When building the model, the natural starting point should be that the model in all included periods should meet the demand for services using the access and core network for the SMP operator. Whether this SMP operator has a universal service obligation or not is irrelevant. As argued elsewhere, one of the many shortcomings of the current model is that it has been unable to adjust to changes in demand over time in the access network. For instance, a 23 per cent decrease in the demand for copper lines has not had any consequence for the estimated requirements for trenching and ducting in the model. In particular this means that it is implicitly assumed that drop-wire costs are relevant to include whether or not these access lines are active. Or to put in other words, the demand side are assumed only to have very minor influence on the scale of the access network. This means that the model on the cost side currently tries to mimic a full scale access network for a universal service provider while at the same time prescribing that only the active lines should contribute to the cost recovery. Modelling a full scale access network independent of current demand would clearly only make sense if all modelled lines also contributed to the cost recovery. If it then should turn out that TDC had a universal service obligation deficit, this calculation could be used in the ensuing recovery procedure.

Criterion RH 3 The revised hybrid model should utilise a technically feasible network based on packet switched, IP, technology. The quality level, functionality and other characteristics of the modelled network should however correspond to the voice interconnection products that the SMP operator s network is designed for. Criterion RH 6 Although the model will be based on an all-ip network, it will still need to interconnect with TDM networks. Thus the costs of interconnecting with such networks (media gateways) should be included. The most cost effective solution today is to use IP interconnect between operators at very few regional points instead of continuing the current practice with many traditional local interconnection points. Given the deregulation of regional and inter-regional interconnect it is of course important that local interconnect is costed in the model. However, to give the correct economic incentives for TDC to enter alternative and more efficient commercial interconnection agreements at higher network levels, it is important that local interconnect is priced at the minimum cost of providing interconnect instead of including an inefficiency premium to TDC. Including media gateways that apparently has the sole purpose of enabling a PSTN/IP conversion at the local interconnect level, is an example of such an inefficiency premium that should be avoided. Criterion RH 9 The model should estimate equipment quantities for the access network using detailed maps and other information for a sample of areas. In the absence of such information alternative approaches, such as data from the SMP operator, may be used. The meaning of this criterion is not clear to the LRAIC Forum. In particular, what would be the purpose of asking TDC for data if TDC in the first place has told NITA that detailed maps and other relevant information does not exist for a reasonable sample of areas? If no hard facts exist to base the modelling on, it is not seen as a very attractive alternative to simply ask TDC for input. The estimation of equipment quantities (trenching, duct, cables etc.) based on different samples is in the current model set-up the single most important step in making a bottom-up calculation of the access costs. If this estimation is compromised by lack of data and/or other types of transparency, it is therefore the entire process which is being seriously compromised. NITA is therefore asked to further elaborate on this criterion. Criterion RH 10 The models should show the anticipated growth per annum for each service for one, three and five years, with the modelled networks dimensioned appropriately. The models should be flexible enough to allow for changes in margins for growth. Not only should the model show the anticipated growth for each service; the model should also be capable of adapting to these changes through explicit cost volume relationships. For instance, changes in the demand for copper access should have an immediate impact on trenching and ducting cost, and changes in the demand for fibre access should have an immediate impact on the proportion of shared costs allocated to fibre access.