Access key 0 - Accessibility, Access key 2 - Jump to content, Access key 7 - Jump to navigation
Skip To Content | Skip To Navigation
 

Home > Consultations > Consultation Documents > Mobile call termination > Statement > Annex C


Annex C - LRIC plsu target charge, Statement on Wholesale Mobile Voice Call Termination consultation

LRIC+ target charge

Use of LRIC as the cost base

C.1 As stated in Annex E of the May consultation and Annex F of the December consultation, Ofcom's view is that the most appropriate and economically efficient basis for regulatory charge controls is forward-looking LRIC. The LRIC of voice termination is the additional cost an MNO incurs to provide termination. This can also be seen as the cost that the firm would avoid if it decided not to provide voice termination, taking a long-run perspective. It corresponds more closely to the charges that would prevail in an effectively competitive market than accounting-based measures of cost. It is a fundamental goal of price regulation to mimic the effects of a competitive market and this consideration underpins the use of LRIC.

C.2 LRIC is widely used as a regulatory costing technique, for example by other NRAs in Europe such as the PTS in Sweden, and by the FCC in the US. It has also been identified as the most appropriate methodology to use for setting interconnection charges by the European Commission in its 1998 Recommendation on Interconnection. For further details, see The Use of Long Run Incremental Cost (LRIC) as a Costing Methodology in Regulation, 12 February 2002, http://www.ofcom.org.uk/static/archive/oftel/publications/mobile/
ctm_2002/lric120202.pdf. Furthermore, the CC has agreed, as stated in paragraph 2.251 of the CC report, with the use of LRIC as the appropriate costing methodology for setting termination charges.

C.3 Ofcom's view remains unchanged that the only relevant costs for the purposes of setting the charge controls are those relevant to 2G voice termination. This excludes 3G costs which the MNOs can recover through their unregulated 3G charges. The CC took the same view on this issue as stated in paragraph 2.251 of the CC report.

LRIC model

C.4 The purpose of the LRIC model is to derive the costs of a reasonably efficient 2G mobile operator in the UK. In April 2002, the latest version of the model was made available which considered a voice-only network. The model and supporting documentation are available at http://www.ofcom.org.uk/static/archive/oftel/publications/mobile/
ctm_2002/april02_model.zip (-42-). Further detailed papers are also available: Source of algorithms, data, assumptions and estimates, http://www.ofcom.org.uk/static/archive/oftel/publications/mobile/
ctm_2002/analysis300102.pdf; and Manual for the Oftel LRIC model, http://www.ofcom.org.uk/static/archive/oftel/publications/mobile/
ctm_2002/slides300102.pdf. In the December consultation, some amendments were made to the model as a result of responses to the May consultation, as documented in paragraphs F.15-F.16, F.23-F.24, F.26-F.27 and F.31-F.35.

C.5 In designing the model, five key issues were considered as listed in the December consultation and summarised briefly in the May consultation:

C.6 In its review of the charges for calls to mobiles, the CC agreed with these general principles and that the April 2002 LRIC model was a suitable starting point for the assessment of the costs of terminating calls on mobile networks (paragraph 2.287 of the CC report).

Responses to the December consultation

LRIC model output

C.7 In the opening paragraphs of section 3.3.2 of its response to the December consultation, Orange reiterates its belief that the only way to derive reliable outputs from the LRIC model is to amend the input parameters and the underlying methodology. Ofcom believes that this point has been substantively addressed in paragraphs F.12-F.14 of the December consultation.

C.8 The remainder of this section considers the other issues raised in response to the December consultation regarding the output of the LRIC model and presents Ofcom's response.

Cost of capital

C.9 As stated in Annex D of the May consultation and Annex E of the December consultation, Ofcom believes that the appropriate cost of capital in the context of this market review is the cost of capital for a reasonably efficient 2G mobile operator in the UK, and in particular, the cost of capital regarding 2G termination services.

C.10 The model made available in April 2002 calculates the LRIC on the basis of a 12.5% pre-tax real cost of capital in 2003/04 and subsequent years. Ofcom has updated the components of the CAPM used to derive an estimate for the cost of capital in the light of more recent information. On this basis, Ofcom estimates the pre-tax real cost of capital to be in the range of 9.8% to 14.0% with a mid-point of 12%. This is a small decrease from the 12.25% proposed in the December consultation (due to a fall in the estimate of the risk free rate) and identical to the value proposed in the May consultation which were both based on the best available information at that time. Further details of the derivation of this range are provided in Annex B.

C.11 The resulting adjustment to the output from the April 2002 LRIC model, taking account of the amendments described in the December consultation, is shown in the table below with the LRIC+ figures for 2005/06 unchanged to within 1%.

Table 1: Adjustment to April 2002 LRIC model termination output (after December consultation amendments) for 12% cost of capital
Pence per minute (real 2000/01) 2001/02 2002/03 2003/04 2004/05 2005/06
900/1800MHz operators  
LRIC+ (12.25% CoC - Dec consultation) 5.07 4.76 4.15 3.95 3.76
LRIC+ (12% CoC) 5.09 4.78 4.12 3.92 3.73
Difference 0.02 0.02 -0.04 -0.03 -0.03
Percentage 0.4% 0.5% -0.9% -0.8% -0.7%
1800MHz operators  
LRIC+ (12.25% CoC - Dec consultation) 6.26 5.85 5.03 4.75 4.50
LRIC+ (12% CoC) 6.29 5.88 4.98 4.70 4.46
Difference 0.03 0.03 -0.05 -0.05 -0.04
Percentage 0.4% 0.5% -1.0% -1.0% -0.9%

Administrative fees for 2G spectrum

C.12 Ofcom is currently undertaking a review of the annual administration fees paid by MNOs for their 2G spectrum allocation. A report by Indepen, Aegis and Warwick Business School was made publicly available in February 2004 (see http://www.ofcom.org.uk/research/industry_market_research/m_i_index/
spectrum_research/independent_review/ and Ofcom is expecting to publish a document for consultation towards the end of the summer. Potentially, the review may result in a change to the 2G spectrum fees paid by MNOs within the time period of the charge control. Orange notes this issue in section 3.3.4 of its response.

C.13 The Indepen et al report reviews the application of Administered Incentive Pricing (AIP) to radio spectrum, as part of the programme of work to implement the Government's response to the Cave Review of Radio Spectrum Management. The report estimates the marginal opportunity cost of spectrum to be (GBP) (GBP) £1.680m for a 2x1MHz carrier in both the 900MHz and 1800MHz bands. This figure is about 2.4 and 3 times higher than the existing charges for the 900MHz and 1800MHz bands respectively. However, the report does not determine final spectrum licence charges as Ofcom will undertake that task following consultation, having also taken into consideration a range of other policy issues and objectives beyond simply that of maximising efficiency in the use of spectrum.

C.14 To derive a LRIC+ target charge for 2005/06 it is necessary to make assumptions about future spectrum pricing. One option would be to ignore Ofcom's forthcoming review, perhaps on the basis that the new charges are not yet known. But, given that Ofcom is concerned not to understate costs and that there is some indications that the administration fee may rise, Ofcom considers it preferable to set assumptions for the 2G spectrum licence charges from 1 April 2005 on the latest available information which is the marginal opportunity cost estimates provided by Indepen et al. For the avoidance of doubt, Ofcom emphasises that such an approach is not intended to, and does not, pre-judge the outcome of Ofcom's forthcoming consultation on administered incentive pricing for spectrum.

C.15 In examining this issue, Ofcom has made a correction to a minor error in the original calculation of the 2G spectrum fees presented in the April 2002 LRIC model regarding conversion of nominal charges to real 2000/01 values and also updated the 1800MHz charges for 2001/02 to 2003/04 reflecting the actual implemented charges, as this information (-43-) was not available when the model was first constructed. Together with the revised estimates for future 2G spectrum pricing, this results in a net amendment to the input cells C85:D135 of the Unit_cost_data sheet in the Netw_R2.xls model file which is shown in the table below for the years 2001/02 to 2005/06.

Table 2: Adjustment to spectrum pricing to April 2002 LRIC model inputs for 2001/02 to 2005/06
Spectrum pricing ((GBP) (GBP) £ real 00/01) Original April 02 LRIC model Revised inputs
GSM 900 GSM 1800 GSM 900 GSM 1800
Year  
2001/2002 17,436,672 24,330,240 15,294,727 16,242,188
2002/2003 17,855,152 24,914,166 14,936,256 15,861,511
2003/2004 18,283,676 25,512,106 14,586,188 15,489,757
2004/2005 18,283,676 25,512,106 14,188,899 15,067,857
2005/2006 18,283,676 25,512,106 34,422,796 44,416,511

C.16 Re-calculating the model output to reflect this revision to 2G spectrum pricing results in an increase in the model's LRIC+ of termination for 2005/06 of about 0.13ppm and 0.15ppm for combined 900/1800MHz and 1800MHz operators respectively.

Table 3: Adjustment to April 2002 LRIC model termination output (after December consultation amendments) for 12% cost of capital and revision to 2G spectrum pricing
Pence per minute
(real 2000/01)		 2001/02 2002/03 2003/04 2004/05 2005/06
900/1800MHz operators  
LRIC+ (12% CoC) 5.09 4.78 4.12 3.92 3.73
LRIC+ (12% CoC / revised 2G spectrum) 4.86 4.53 3.90 3.69 3.86
Difference -0.23 -0.25 -0.22 -0.22 0.13
Percentage -4.4% -5.2% -5.3% -5.7% 3.5%
1800MHz operators  
LRIC+ (12% CoC) 6.29 5.88 4.98 4.70 4.46
LRIC+ (12% CoC / revised 2G spectrum) 5.94 5.50 4.65 4.37 4.61
Difference -0.35 -0.38 -0.32 -0.33 0.15
Percentage -5.6% -6.5% -6.5% -7.0% 3.4%

Treatment of equipment that declines in quantity

C.17 In paragraphs B.16-B.21 of its response, whilst Vodafone agrees that the approach regarding swapped out microwave links described in paragraph F.24 of the December consultation is reasonable, Vodafone has a greater concern regarding the treatment of equipment that declines in quantity but then recovers in later years in response to demand.

C.18 Specifically, Vodafone believes that whilst the approach described in paragraphs F.26-F.27 of the December consultation to prevent equipment quantities temporarily falling in response to reduced demand appears a plausible correction, it ignores consideration of the depth and duration of the temporary "dip" in demand. Vodafone believes that an efficient operator would not maintain surplus equipment for the duration of the dip. Rather, in Vodafone's opinion it is the initial peak requirement that is transient and a 'quirk' of the model's look-ahead algorithm. Vodafone proposes that a more reasonable assumption is to remove the transient peak on the basis that an efficient operator would not purchase the equipment in 1999/2000 but instead expand its network more evenly.

C.19 Vodafone correctly notes that the peak in equipment quantity requirements is in part due to the model's look-ahead algorithm as described in paragraph F.26 of the December consultation. In reality network planning for equipment is a complex undertaking influenced by a number of factors and the model necessarily takes a simplified and more mechanistic approach. Ofcom considers that there is merit in the arguments put forward for the treatment of decline in equipment taken in the December consultation: an efficient operator may be expected to satisfy full traffic demand and, after a period of rapid growth, may not be able to accurately anticipate a significant decrease in that rate of growth, leading to a subsequent period during which equipment levels are higher than ideally necessary. However, there is also merit in Vodafone's observation about the depth and duration of the dip in network requirements. Ofcom recognises that in reality an efficient operator may be able to manage the impact of significant changes in growth of demand to some degree thus 'smoothing' the timing of equipment purchase and reducing the extent of under-utilisation of equipment for the period in question. The net impact of adopting Vodafone's approach in place of that proposed in the December consultation is an increase in the LRIC+ target charge for 2005/06 of about 0.12ppm after reconciliation with the MNO data in 2001 (see paragraph C.22 below). A case can be made for either approach, but on balance, Ofcom considers it reasonable to adopt Vodafone's amended approach to this issue in the interests of ensuring that costs are not understated.

C.20 With regards to the detailed amendments to the model, Vodafone proposes a smoothing adjustment for 3-sector cell sites and for inter-switch transmission links. However, Ofcom believes that to single out only two asset elements results in an inconsistent treatment and it is more appropriate to apply this adjustment to all asset elements which are subject to temporary decline in quantity. In particular these elements are:

C.21 The quantities of the above asset elements can be smoothed over the relevant period by overwriting the values of the network_design_full sheet of the Netw_R2.xls model file according to the table below.

Table 4: Amendments to specific modelled equipment quantities on network_design_full sheet of Netw_R2.xls - where there is no entry the model is allowed to calculate equipment quantities unmodified
Equipment asset element Row 98/99 99/00 00/01 01/02 02/03 03/04
Macrocells - Urban (900) 152 - 1660 1670 1680 1690 -
Macrocells - Urban (1800) 152 - 1700 - - - -
Macrocells - Suburban (900) 153 - 1340 1380 1395 1410 -
Macrocells - Suburban (1800) 153 - 1075 - - - -
Macrocells - Suburban (900) 420 - - - 1 - -
Macrocells - Suburban (1800) 420 - - - - - -
Macrocells - Suburban (900) 421 - - - 0 - -
Macrocells - Suburban (1800) 421 - - - - - -
BSC: MSC ports 710 710 - - - - 2625 2650
MSCs 727 - - - - 87 -
MSC: interconnect ports 740 - - 4400 4450 - -
MSC: switch ports 749 - -   7400 - -
Inter-switch transmission links 767 - 6050 6100 6150 - -

C.22 The impact of the amendments is a lower economic cost model output in 2005/06 due to reduced quantities of equipment, and a lower model GBV and operating cost in the reconciliation year. The lower output in the reconciliation year results in a greater upwards adjustment after comparison with the MNO data in 2001, from 35.6% to 38.7% for GBV and from -14.2% (-44-) to -8.5% for operating costs. This adjustment counteracts the lower economic cost model output in 2005/06 (by -0.11ppm and -0.06ppm for combined 900/1800MHz operators and 1800MHz operators respectively) so that the net impact of this change is a slight increase of about 0.12ppm in the final LRIC+ target charge for 2005/06. There is also an implication for the differential between combined 900/1800MHz operators and 1800MHz operators which is discussed in paragraph C.87 below.

Table 5: Adjustment to April 2002 LRIC model termination output (after December consultation amendments) for 12% cost of capital, revision to 2G spectrum pricing and revision of treatment of equipment that declines in quantity
Pence per minute
(real 2000/01)		 2001/02 2002/03 2003/04 2004/05 2005/06 GBV ((GBP) (GBP) £m)
(-45-)		 Opex ((GBP) (GBP) £m)
(-46-)
900/1800MHz operators  
LRIC+ (12% CoC / revised 2G spectrum) 4.86 4.53 3.90 3.69 3.86 1835 348
LRIC+ (12% CoC / revised 2G spectrum / revised declining assets) 4.73 4.41 3.79 3.58 3.76 1758 318
Difference -0.13 -0.12 -0.11 -0.11 -0.11 -76 -30
Percentage -2.7% -2.7% -2.9% -3.0% -2.7% -4.2% -8.5%
1800MHz operators  
LRIC+ (12% CoC / revised 2G spectrum) 5.94 5.50 4.65 4.37 4.61 1983 403
LRIC+ (12% CoC / revised 2G spectrum / revised declining assets) 5.87 5.43 4.59 4.31 4.55 1973 387
Difference -0.07 -0.06 -0.06 -0.06 -0.06 -10 -17
Percentage -1.1% -1.2% -1.3% -1.4% -1.3% -0.5% -4.2%

Asset lifetimes

C.23 In section 3.3.2.A of its response, Orange repeats its concern that the asset lifetimes used in the model are unrealistically long. In particular, Orange raises the following issues:

C.24 T-Mobile expresses similar concerns in paragraphs II.67-II.70 of its response. Additionally, T-Mobile requests further detail regarding the treatment of T-Mobile's asset lifetimes.

C.25 Neither Orange nor T-Mobile has submitted new evidence regarding this issue which was responded to previously in paragraphs F.40-F.47 of the December consultation. Ofcom therefore finds no compelling evidence to adopt a different position from that stated previously. The further, more detailed, concerns raised by these two operators are addressed below.

C.26 Whilst Orange is correct in noting that asset lifetimes are an input in the April 02 model, rather than derived as part of the economic depreciation calculation as in the September 01 model, it is incorrect to characterise this particular change as an improvement designed to address a 'flaw' in the original version of the model. The primary purpose of the later version of the model was to improve transparency of the calculations and simplify the analysis where appropriate (specifically to present a voice-only model rather than one which also models data services). As a secondary consideration, the opportunity was taken to improve the economic depreciation algorithms, and in particular, to implement an 'all-instances' calculation which explicitly considers the number of assets purchased in each year, rather than a 'single-instance' approach (see Mobile phones inquiry: Mobile termination - re-presented cost model, letter to CC, 8 April 2002 (-47-)). Whilst this change makes it more difficult to calculate the asset lifetimes endogenously (hence the direct inputs for asset lifetimes in the April 02 model), it does not change the appropriate asset lifetimes which seek to recognise the trade-off between increased capital costs if assets are replaced more frequently with increased operating costs associated with older assets if they are replaced less frequently. Taking the endogenously calculated lifetimes from the September 01 model as direct inputs for the April 02 model is a pragmatic approach to respecting this trade-off in an internally consistent way.

C.27 Regarding Orange's second concern, Ofcom agrees that it is relevant to consider the impact of changes in asset lifetimes after comparison with the MNOs' accounts. It was noted in paragraph F.47 of the December consultation that Orange's proposed asset lifetimes did indeed result in a slightly higher economic cost of termination in 2005/06 after reconciliation with operator data, however, this was not confirmed by the results obtained using T-Mobile's asset lifetimes. As noted in paragraph F.45 of the December consultation, implementing shorter asset lifetimes has two counteracting effects: capital cost recovery is increased and operating cost recovery is reduced. Following the amendments to the model described earlier in this annex, Ofcom has recalculated the impact of different asset lifetimes based on the information previously provided by Orange and T-Mobile. In the case of T-Mobile's information the net effect is a lower model output for 2005/06 due to a stronger operating cost effect prevailing. Ofcom has then determined the impact after comparison with operator data in 2001 recognising that amending asset lifetimes has a direct impact on the capital / operating cost proportions over time as noted by Orange in section 3.3.2.B of its response (see also paragraph C.57 below). Consistent with the findings in paragraph F.47 of the December consultation, whilst one set of lifetimes leads to a slightly higher cost of termination in 2005/06, the other set leaves the efficient charge level unchanged to within about 1%. In any case, the key observation is that the reconciliation with MNOs' GBV figures is a relatively high level process and sensitive to the particular timing of asset replacement - essentially due to a modelling artefact which simplifies the nature of asset replacement. For this reason, Ofcom believes that it would be inappropriate to place too great a weight on significant changes to the results which arise solely from reconciliation with operator data following changes in asset lifetimes, which in themselves have a negligible impact on the model output (as noted in paragraphs F.44-F.46 of the December consultation).

C.28 Orange re-articulates its concern regarding capital cost recovery by providing a simple worked example. However, this example does not reflect the relevant calculation as it illustrates a point where the underlying path of capital cost recovery is simple straight-line accounting depreciation, rather than an economic depreciation calculation. Orange is correct in observing that capital costs are determined by the initial gross book value (GBV) of the asset as well as the capital recovery profile. However, Orange is incorrect in its belief that the second aspect has not been addressed. The reconciliation with operator data seeks to compare quantities of assets at appropriate prices; hence it is reasonable to consider a comparison of GBV. However, the path of cost recovery chosen to determine regulated charges (economic depreciation) is very different from that used in the operator data (accounting depreciation) for reasons discussed in paragraphs C.88-C.93 below. It would therefore be inappropriate to attempt to reconcile capital cost recovery in the model against that presented in the MNOs' accounting information. Whilst the quantity of assets is addressed through a GBV adjustment, the increased capital cost recovery associated with shorter asset lifetimes is addressed internally within the model through the economic depreciation calculation and, as described in paragraph F.44 of the December consultation, this effect is broadly balanced by the reduced operating cost recovery associated with younger assets.

C.29 In response to T-Mobile, as described in paragraph C.26 above, it is questionable whether asset lifetimes in the model should be amended without also reflecting the likelihood of faster falling modern equivalent asset (MEA) price trends and / or faster rising operating costs with age. Nevertheless, a simple sensitivity analysis is undertaken (as in the December consultation) where the lifetimes of network elements in the LRIC model are reduced in accordance with those suggested by T-Mobile (see paragraph F.44 of the December consultation). The same approach is taken with Orange's asset lifetime information. As is transparent from examination of the April 02 model (and highlighted above), asset lifetimes in the model are direct inputs and so can simply be replaced for the purpose of undertaking sensitivity analysis regarding shorter lifetimes. T-Mobile did not provide revised asset lifetimes for all network elements in the LRIC model but only for the major network elements. Hence, in undertaking the sensitivity, asset lifetimes for the remaining categories are based on the more comprehensive set of lifetimes provided by Orange, given that these lifetimes are shorter than those originally used in the April 02 model and hence more favourable to T-Mobile.

C.30 Regarding T-Mobile's final point in paragraph II.70 of its response, the sensitivity analysis involving shorter asset lifetimes still results in an upwards capital cost adjustment and a downwards operating cost adjustment. However, the key factor is not the impact of shorter asset lifetimes on the comparison with operator data in 2001 but the impact of the higher capital cost recovery and lower operating cost on the economic cost in 2005/06 which is captured within the model through the economic depreciation calculation.

Asset price trends after 2010, 3G migration and dynamic uncertainty

C.31 Vodafone (paragraphs B.29-B.33 of its response) submits that equipment prices should continue to fall after 2010, following the observed trend in the model from 1990. The only factor that Vodafone can identify for a stabilising in 2G asset prices is the emergence of 3G technology, and the subsequent diverting of manufacturers' focus on the development of 3G equipment. Vodafone asserts that whilst migration of voice traffic from 2G to 3G has been ignored, the impact of 3G on 2G MEA prices has been recognised selectively.

C.32 Following the above point, in section 3.3.1 of its response, Orange proposes that there are significant longer term uncertainties in a dynamic market and that it is inappropriate to ignore 3G considerations in the cost modelling. Orange does not believe that modelling 2G costs exclusively results in estimates which represent a ceiling on reasonable costs. Whilst acknowledging that it is potentially impossible to reflect the dynamic uncertainties accurately, Orange believes that caution should be exercised in issues of discretion and attempts should be made to model particular outcomes to provide a range of estimates.

C.33 In particular, Orange states that the Director's approach essentially assumes that MNOs will only migrate traffic from 2G to 3G when unit costs of 3G are lower than 2G and believes that this is unreasonable due to a number of exogenous factors which influence an MNO's migration including: the competitive landscape; 2G / 3G interoperability issues; and regulatory uncertainties.

C.34 T-Mobile (paragraph I.27 of its response) also argues that insufficient consideration has been given to dynamic implications, in particular, the effect on MNOs' incentives to invest in risky projects. In response to this point, Ofcom's position is clearly stated in paragraphs L.22-L.23 of the December consultation: future projects (such as 3G) should be assessed on their own merit, rather than on the basis of receiving subsidy from 2G termination charges set in excess of costs. Ofcom believes that it has given due consideration to the 2G investment made by MNOs in accordance with section 88(2) of the Act.

C.35 As stated in paragraph F.49 of the December consultation, for the purposes of regulating 2G termination charges, future voice traffic is modelled as if it is carried on a 2G network of a reasonably efficient operator. This scenario does not intend to negate the existence of 3G technology and its impact, nor the likelihood that in reality voice traffic will at least partially migrate to 3G networks. As to the appropriate MEA price trend, whilst it is typical for rapid reductions to occur in the unit cost of production in the early stages of a product's lifecycle due to significant increase in volume production and economies of scale, it seems optimistic to assume that prices will decline indefinitely and, in the limit, tend to zero. Over the explicitly modelled period, the model already assumes that the price of TRXs fall to 12% of their initial value. Vodafone's belief that price declines should continue after 2010 would lead to unit TRX prices that are less than 3% of the initial value by the end of the modelled period. Ofcom does not believe that it is appropriate to adopt Vodafone's assumption and, as noted previously, this is consistent with the CC's viewpoint (paragraph 2.317 of the CC report).

C.36 It is therefore inappropriate for Orange to interpret the explicitly modelled period of 50 years as an indication that Ofcom assumes no migration to 3G will occur before 2040 or that it is an attempt to predict how the market will look in 2040. Rather, the modelling approach seeks to determine an appropriate basis for calculating the level at which to regulate 2G termination charges. This is achieved by considering a scenario where future voice traffic is modelled as if it is carried on a 2G network of a reasonably efficient operator.

C.37 Orange acknowledges that it is "potentially impossible" to accurately reflect dynamic uncertainties in the modelling, and instead requests that discretion should be exercised in a cautious manner to take account of these uncertainties. Ofcom believes that caution has indeed been taken to ensuring costs are not understated in a number of respects and therefore does not share Orange's concern that the resulting charge is below the floor of reasonable costs. For example, the treatment of equipment that declines in quantity (see paragraph C.19 above); the approach taken to calculating the non-network common cost mark-up (see paragraph C.116 below); the reconciliation with MNOs' accounting data in 2001 which does not make any adjustment for potential operator inefficiency at that time; and the use of economic depreciation, rather than current cost accounting, which results in a significantly higher efficient charge level (see paragraphC.92 below).

C.38 Orange believes that it would be more appropriate to model particular outcomes, including the migration of 2G services to 3G, in order to obtain a range of estimates as a basis for determining the efficient charge level. Ofcom has undertaken significant sensitivity testing in the past, however Ofcom has now focussed on a specific reasonable representation of the costs of 2G termination, based on this earlier work.

C.39 In any case, Ofcom maintains the view that the timing and rate of migration of voice traffic from 2G to 3G networks is essentially a decision to be made by operators and hence the modelling of future voice traffic as if it is carried on a 2G network is reasonable for the purposes of setting regulated 2G termination rates. Orange claims that it cannot entirely control this migration due to the competitive requirement to offer 3G services and the effectiveness of interoperability between 2G and 3G in dual coverage areas since "the handset will automatically select the 3G technology if available". Ofcom is aware that there are already handsets available which enable network selection. Furthermore, as Orange states, future 3G network equipment software releases may also address this concern. Ofcom is of the opinion that whilst there may be a transitory period in the early stages of 3G take-up during which a network operator does not have full control of the rate of migration of voice traffic, within a reasonable period the technology should enable MNOs to have sufficient control. Therefore, Ofcom believes there is insufficient evidence to suggest that Orange's concern will have a material impact.

Summary of revisions to LRIC model results

C.40 The overall impact of the amendments made to the model regarding pricing for 2G spectrum and treatment of equipment that declines in quantity is summarised in the table below.

Table 6: Summary of impact of amendments on LRIC model results
Pence per minute
(real 2000/01)		 2001/02 2002/03 2003/04 2004/05 2005/06
900/1800MHz operators  
LRIC+ (12% cost of capital) 5.09 4.78 4.12 3.92 3.73
LRIC+ (2G spectrum) 4.86 4.53 3.90 3.69 3.86
LRIC+ (2G spectrum + declining assets) 4.73 4.41 3.79 3.58 3.76
Difference -0.36 -0.38 -0.33 -0.33 0.03
Percentage -7.0% -7.8% -8.0% -8.5% 0.7%
1800MHz operators  
LRIC+ (12% cost of capital) 6.29 5.88 4.98 4.70 4.46
LRIC+ (2G spectrum) 5.94 5.50 4.65 4.37 4.61
LRIC+ (2G spectrum + declining assets) 5.87 5.43 4.59 4.31 4.55
Difference -0.42 -0.45 -0.39 -0.39 0.09
Percentage -6.6% -7.6% -7.7% -8.3% 2.1%

Comparison with MNO data

2001 GBV costs understated

C.41 In section 3.3.2.D of its response, Orange states that the average GBV derived from MNO data for 2001 used to compare with the LRIC model is understated due to simple averaging of T-Mobile's GBV which fails to take account of T-Mobile's subsequent network coverage investment. Specifically, Orange believes that the conclusion that T-Mobile's network has primarily been driven by improvements in quality rather than coverage (see paragraph F.95 of the December consultation) is incorrect.

C.42 Ofcom has no compelling reason to revise the position described in paragraph F.95 of the December consultation for the reasons discussed in paragraphs C.71-C.81 below. Ofcom acknowledges that if it accepted that T-Mobile's GBV is understated on a like-for-like basis and if it accepted Orange's proposed solution, the efficient charge in 2005/06 would increase by about 0.05ppm. This increase is small, and in any case, Ofcom does not accept Orange's point for the reasons set out in paragraphs C.71-C.81.

Data adjustment factor

C.43 Vodafone seeks clarification (paragraph B.22 of its response) that in deriving an average data adjustment figure of 4.5% (paragraph F.72 of the December consultation), whether O2's submission to the CC was used in this calculation.

C.44 Orange asserts in section 3.3.2.G of its response that the proposed data adjustment factor is based on the proportion of radio channels dedicated to GPRS which ignores the modularity of equipment. As raised in its previous response, Orange believes that the correct approach is to exclude the incremental costs of adding GPRS equipment to an existing GSM voice network. Further, Orange states that any concerns expressed in paragraph F.71 of differences in cost allocation methodologies is quite unfounded when referring to specific incremental costs.

C.45 Finally, Orange believes that when averaging across operators for the purpose of modelling inputs, a consistent approach should be taken across all operators to ensure the exclusion of data specific costs.

C.46 The purpose of the data adjustment factor and methods for estimating it are set out in paragraphs F.70-F.72 of the December consultation. In particular, two approaches are described. The first approach (adopted by the CC) is based on consideration of the capacity dedicated to non-voice services (see paragraph F.71 of the December consultation). The second approach is based on consideration of the incremental costs exclusively associated with data services (see paragraph F.70 of the December consultation). Ofcom believes that both approaches are a reasonable way of determining the percentage by which a total GBV figure (representing both voice and data) should be reduced to represent the GBV associated with a voice-only network. The data adjustment figure of 4.5% is based on averaging the most recent estimates received from all four MNOs (consistent with the approach taken more generally to the comparison with operator data), using the first approach in the case of O2 (information submitted to the CC), and the second approach in the case of the remaining three MNOs (who submitted further information in response to the May consultation).

C.47 Orange's assertion that the figure of 4.5% is "based on the proportion of radio channels dedicated to GPRS" is therefore not correct since this figure is primarily based on incremental costs associated with GPRS. Regarding Orange's second point that there are no allocation issues when referring to specific incremental costs, this might be a valid observation if a very narrow interpretation of the word "allocation" were taken. However, Ofcom did not intend such a narrow interpretation - paragraph F.71 of the December consultation expressed the broader point that in determining appropriate cost causality there are typically various issues in deciding exactly how to measure the increment and how to attribute costs accordingly. The key sentiment is that a capacity measure is likely to be more neutral in this respect.

C.48 Ofcom believes that the approach it has adopted of deriving an average voice-only GBV figure by calculating the average total (voice and data) GBV figure across all operators and then reducing this by an average data adjustment factor is a consistent approach to take across all operators. If Orange's final comment regarding consistency is intended as a proposal that the total (voice and data) GBV figure for each operator should be reduced by that operator's data adjustment factor to obtain a voice-only GBV figure for each operator which is then used to derive an average voice-only GBV figure, this may well be a reasonable approach. A case can be made for either approach, however, since Ofcom has rounded down the average data adjustment factor to 4.5% (see paragraph F.72 of the December consultation) in the interests of not understating costs, the resulting upwards capital cost adjustment (to be applied to the capital proportion of the 2005/06 economic cost) is essentially the same regardless of which approach is adopted (-48-).

Varying capital / operating cost proportions over time

C.49 In response to the proposal that a more accurate estimate of the economic cost in 2005/06 is obtained by recognising that capital / operating cost proportions vary over time (see paragraph F.80 of the December consultation), O2 (page 2), Orange (section 3.3.2.B-3.3.2.C), T-Mobile (paragraphs II.48-II.52) and Vodafone (paragraphs B.24-B.27) believe that insufficient reasoning has been given for this change. O2 states that it is not clear why the LRIC model estimates have been used rather than other empirical evidence. Orange believes that it is inconsistent to use a theoretical model to determine the proportions over time whilst applying a fixed calibration adjustment solely on outputs for 2001.

C.50 More specifically, Vodafone, T-Mobile and Orange highlight that the capital / operating cost proportions of the economic cost in 2005/06 depend on the asset lifetime assumptions, which they consider questionable.

C.51 Furthermore, Vodafone and T-Mobile question whether it is appropriate to derive capital / operating cost proportions on the basis of the split of the economic cost output from the model. Vodafone believes that this split is merely a feature of the way the economic depreciation calculation recovers cost over time and proposes that it would be more appropriate to determine the split on the basis of an accounting view of operating costs. T-Mobile appears to share similar sentiments stating that the capital and operating cost trends in the LRIC model are an artificial output of the model.

C.52 Finally, T-Mobile believes that the change in the capital / operating cost proportions is inconsistent with arguments used elsewhere that a high level comparison is more appropriate than a more detailed comparison (for example in regard to Vodafone's concern of structural bias discussed in paragraphs F.83-F.89 of the December consultation).

C.53 The purpose of the reconciliation is to establish the extent to which capital costs and operating costs are understated or overstated in the model when compared with operator data. The underlying assumption is that if capital costs are understated by x% and operating costs are overstated by y% in 2001 then this is still the case in 2005/06. These percentages are determined by comparing cost output from the model for 2001 on an historic cost accounting (HCA) basis with MNOs accounts for the same period (given that the MNOs do not state cost information on an economic depreciation basis). However, Ofcom believes that economic depreciation is the most appropriate path of cost recovery for the purpose of setting regulated charges and has calculated the efficient charge level on this basis. In this context the correct split of capital and operating costs in determining the appropriate 2005/06 target average charge is the capital / operating cost proportions of the economic cost in 2005/06.

C.54 As stated in paragraph F.80 of the December consultation, it was recognised in Annex E of the May consultation that varying the capital / operating cost proportions over time may well be appropriate. Indeed, there appears to be no disagreement that operating costs will grow in importance relative to capital costs under economic depreciation over the period in question. Given this acknowledgement, it is analytically more appropriate to recognise this trend in determining the proportion of economic cost in 2005/06 that is regarded as capital cost recovery rather than operating costs. The May consultation proposed to take a more conservative approach using the overall model forecast trend. However, given the reassessment of various detailed points following substantial responses to that consultation, and subsequent amendments of various aspects of the modelling in order to achieve more accurate results, it was consistent to also re-examine this issue in order to derive a commensurately accurate estimate.

C.55 With regards to O2's concern about using the LRIC model's trends rather than empirical evidence, the appropriate proportions of capital and operating costs in the 2005/06 target average charge are determined fundamentally by the specification of the path of cost recovery (as discussed below). In any case, it is unclear what evidence O2 intends given that 2005/06 lies in the future. Indeed, no MNO (including O2) has provided any further empirical evidence. It appears that T-Mobile believes that the approach taken in the December consultation followed the acquisition of MNO data in later years. This is not the case. Furthermore, as noted in paragraph F.94 of the December consultation, one of the reasons for choosing a date of comparison in 2001, rather than a later date, was the CC's desire to minimise the extent to which the information it obtained, such as GBV, was influenced by network costs relating to data services.

C.56 Contrary to Orange's opinion, Ofcom believes that it is appropriate to use the capital / operating cost trends in the model as this is entirely consistent with the underlying model trends which are used to determine the economic cost in 2005/06. Ofcom reiterates that the approach taken is to use the bottom-up LRIC model to provide an understanding of cost structures and relationships. As noted in paragraph F.115 of the December consultation, deriving cost trends from this bottom-up model and undertaking a single point comparison with top-down operator information constitutes a reasonable and conventional approach.

C.57 Ofcom acknowledges that the capital / operating cost proportions of the economic cost in 2005/06 do depend on the asset lifetimes assumed in the model and, as expected, shorter asset lifetimes result in a small increase in the proportion of economic cost in 2005/06 which can be attributed to capital cost recovery. However, Ofcom believes that the lifetimes used in the LRIC model produce a reasonable estimate for the cost of termination and for the reasons discussed in paragraphs C.23-C.30 above does not believe there is compelling evidence to adopt a different position when the results of the model in 2005/06 are not materially influenced by the adoption of shorter asset lifetimes.

C.58 With regards to T-Mobile's assertion that a more detailed approach is pursued regarding this issue but avoided elsewhere, T-Mobile appears to have failed to appreciate the difference in nature between this issue and the others with which it seeks a comparison. Whilst Ofcom has reservations about comparing data at a more detailed or 'granular' level, since this may result in less accurate results for the reasons set out in the December consultation (for example, paragraph F.87), the issue in this case concerns analytical consistency of model trends. As already noted, it is not disputed that operating costs increase as a proportion of total costs over time and therefore Ofcom believes that it is more accurate to recognise this trend in determining the efficient charge level.

C.59 The amendments to the LRIC model for cost of capital, pricing for 2G spectrum and treatment of equipment that declines in quantity (summarised in paragraph C.40 above) results in a small change to the capital / operating proportions from the LRIC model in 2005/06 in comparison to those stated in the December consultation (see paragraph F.81). The appropriate split becomes 28% and 72% for capital and operating costs respectively for a combined 900/1800MHz operator, and 30% and 70% for capital and operating costs respectively for an 1800MHz operator.

C.60 The overall impact on adjustments following comparison with the MNOs' data, implementing a capital cost adjustment of 38.7% and an operating cost adjustment of -8.5% (see paragraph C.22 above) together with the revised capital / operating cost proportions (but excluding any adjustment for the differential in costs between combined 900/1800MHz and 1800MHz operators), is an increase to the LRIC model output in 2005/06 of 0.19ppm for combined 900/1800MHz operators and 0.27ppm for 1800MHz operators. This compares with adjustments in the December consultation of 0.01ppm and 0.05ppm for the two operator types respectively.

Structural bias in the model and termination-specific adjustment

C.61 Vodafone (paragraphs B.1-B.3 of its response) still believes that calibration of the model at more than a single point in time is preferable and the argument that additional resources would be required to verify the data is insufficient. As in its previous submission, Vodafone goes on to state that it believes a termination-specific calibration should be undertaken with its FAC information as well as the information provided by its own version of the LRIC model.

C.62 In particular, Vodafone is concerned that there is a structural bias in the model so that different services display different degrees of discrepancy with the MNO data. Vodafone argues in paragraphs B.7-B.13 of its response that this necessitates a further termination-specific adjustment.

C.63 A response to Vodafone's arguments has already been set out in the December consultation. Specifically, as stated in paragraphs F.87 and F.115 of the December consultation, Ofcom believes that its approach to reconciliation is appropriate and fit for purpose. Vodafone appears incorrectly to believe that the main reason for not pursuing further calibration is that the activity would prove resource-intensive. Whilst it was noted that this approach would involve significant effort, as explained in paragraphs F.87 and F.113 of the December consultation, Vodafone's proposal has not been pursued primarily due to considerable doubt whether it would lead to more accurate results.

C.64 Ofcom has not received any compelling evidence from Vodafone to amend its position stated in the December consultation; however, its detailed points are addressed below.

C.65 Vodafone notes that the unamended results from the April 2002 LRIC model were described by Ofcom at that time as an adequate representation of an efficient operator, but amendments were subsequently conceded following the CC inquiry. In the light of this and its FAC estimates, Vodafone does not believe there can be confidence that the "estimate of the cost of termination in 2005/06 is reasonable" (paragraph F.112 of the December consultation). The comment regarding the appropriateness of the LRIC model results prior to the CC inquiry needs to be taken in context. In paragraph A3.14 of the Review of the Charge Control on Calls to Mobiles, 26 September 2001, it was stated that:

Reconciliation with accounting-based figures has not, however, been possible, because the MNOs have supplied neither detailed accounting information nor top-down models. Furthermore, little evidence has been supplied by the MNOs to inform the data and assumptions for the bottom-up model. The robustness of the model's results would have been improved by the provision of such information. Nevertheless, the figures in the LRIC model represent the best estimates available for the cost of mobile termination and form a reasonable basis for setting charge caps.

C.66 Hence it was appropriate to update the estimates of the cost of termination in the light of the new evidence from MNOs supplied to the CC which had not been made available previously. In contrast, the further information provided by Vodafone regarding structural bias, already considered in the December consultation, is far from conclusive.

C.67 Vodafone emphasises that its FAC allocation algorithms were identical to those used in the LRIC model and thus its FAC results should be regarded as a floor to the average MNO FAC rather than dismissed as unusable. Vodafone's argument appears to be that if the basis of allocation used in the LRIC model and Vodafone's FAC construction are identical, then any discrepancies must be due to other factors than cost allocation, such as understated cost inputs, or structural bias in the LRIC model. These other factors are addressed through the high level reconciliation exercise and specific investigation of whether there really is structural bias in the LRIC model. This suggests that there is then little further value to be gained from undertaking a reconciliation with the FAC estimates for its own sake since the reasons for variation have been addressed by Ofcom in other ways.

C.68 For the reasons stated in paragraph F.87 of the December consultation, Ofcom believes that it is more appropriate to undertake a high level model reconciliation exercise (as adopted by the CC) rather than to conduct a more granular reconciliation exercise. Nevertheless, a short analysis of the available network depreciation information is described in paragraphs F.88-F.89 of the December consultation which suggests that Vodafone's data is unrepresentative of the four MNOs as a whole. Vodafone states that it is self-contradictory to analyse network depreciation data when this is the underlying information supporting the FAC analysis which Ofcom believes to be unreliable. Ofcom disagrees with Vodafone's conclusion. The difficulties associated with obtaining reliable FAC estimates is primarily due to the differences in cost allocation to services (specifically the termination service in this case), rather than a concern, necessarily, regarding the underlying cost data itself. Secondly, Vodafone provides two reasons why its network depreciation figures may differ significantly from the other MNOs. Ofcom does not contest that these may be plausible explanations; however, providing reasons for why Vodafone is unrepresentative of the average MNO simply provides support for Ofcom's argument that it is inappropriate to use Vodafone's specific figures given that the objective is to determine whether the model has a structural bias when compared to the average MNO. In any case, as set out in paragraph F.87 of the December consultation, Ofcom does not believe that Vodafone's proposed approach is appropriate. The use of network depreciation data is intended to highlight that, on the basis of the available information, a more granular analysis does not appear justified.

C.69 Vodafone also raises the point that substituting Vodafone's actually experienced MEA unit costs into the LRIC model results in a change in high level component cost categories indicating that the existing cost mix is inappropriate. This observation advocates a more detailed reconciliation exercise, however, for the reasons articulated in the paragraph above, this is not Ofcom's chosen approach and, for the reasons already described, it is unclear that adopting Vodafone's proposal would provide a representative result for the four MNOs taken as a whole.

C.70 If Ofcom was attempting to model each operator's specific network then Vodafone's version of the LRIC model would provide a very useful comparison. However, the chosen approach has been to model a generic average operator's network (taking account of differences in spectrum allocation) and to undertake a reconciliation with average information across all four MNOs. In this context Vodafone's version of the model is of more limited value and Vodafone's specific information and data is reflected through the aggregated comparison.

Combined 900/1800MHz and 1800MHz operator cost differential

1800MHz operator cost disadvantage underestimated

C.71 In its response to the May consultation, T-Mobile argues that the costs of an 1800MHz operator are higher than those of a combined 900/1800MHz operator for providing a similar level of quality and coverage. Orange supports this view by arguing that the conclusion reached by both Ofcom and the CC that the network costs of each network type are similar was erroneous since the comparison was undertaken at a point in time when T-Mobile had significantly lower coverage and call quality. Arguments and supporting evidence from the 1800MHz operators was considered in paragraphs F.92-F.98 of the December consultation, which ultimately concluded that there was no compelling reason to make further adjustments. In paragraphs II.58-II.66 of its response, T-Mobile sets out in detail why it still believes that there is a cost disadvantage suffered by 1800MHz operators which should be recognised.

C.72 T-Mobile claims in paragraph II.60(a) of its response that the core of its argument is not addressed in the December consultation. T-Mobile's key concern is that the conclusion that no cost difference existed between networks of different types arises only because of an invalid comparison in 2001 resulting in a downward bias of the average costs of a 1800MHz network. More specifically, T-Mobile raises a number of points relating to:

C.73 In paragraphs II.63-II.65 of its response, T-Mobile comments on the definitions of network coverage, call quality and grade of service used in the December consultation stating that the reference to call quality being modelled in the LRIC model by means of a blocking probability percentage demonstrates confusion between grade of service and call quality. Specifically, T-Mobile disagrees with the definition of coverage presented in paragraph F.95 of the December consultation. Ofcom's view is that its definition of coverage is appropriate in the specific context of comparing the outputs of the LRIC model for different operator types. Ofcom acknowledges that its definition might not be appropriate for other uses, and agrees that the interaction of technical factors influencing coverage is a complex one. But this issue is not key in the context of comparing the economics of service provision using the two different network technologies.

C.74 T-Mobile disputes the suggestion in paragraph F.95 of the December consultation that the roll-out carried out by T-Mobile after the CC's comparison of data in 2001 was driven by the need to make improvements in quality rather than coverage. T-Mobile states that its investment was aimed at improving both these aspects of its network. Ofcom has considered the evidence presented by T-Mobile in support of this point, noting T-Mobile's submission that its cell count and land area coverage have both increased. In practice there are substantial difficulties inherent in attempting to quantify the extent to which the network roll-out undertaken by T-Mobile since the date of comparison has been driven by each of: (1) improving coverage to bring T-Mobile's network reach in line with the other MNOs; (2) improving quality; and (3) meeting increasing capacity requirements. This is especially so given the specific definition of coverage appropriate to this reconciliation exercise. As outlined in paragraph C.75 below, at the date of comparison, the average combined 900/1800MHz operator had broadly the same quality of service as the average 1800MHz operator. This means that only additional roll-out falling into the first of the above categories would be relevant in comparing the relative costs of generic combined 900/1800MHz and 1800MHz operators. Ofcom notes that T-Mobile has not provided a quantification of roll-out arising specifically from the category above. Ofcom has formed its view on the extent to which the costs of combined 900/1800MHz and 1800MHz operators differ based on a high level consideration of a number of measures. It has considered the extent to which the conclusions it has drawn from an analysis of these measures would be altered by modifying the data submitted by T-Mobile to reflect increased geographic coverage. Ofcom's view, in the light of the practical difficulties referred to above, is that it does not have compelling evidence to change its view that neither operator type had a significant cost advantage over the other (on an accounting basis) at the date of comparison.

C.75 With regards to differences in quality, T-Mobile supports its view by referring to Oftel's call success rate surveys which clearly indicate a difference in quality between the operators at September 2001, with T-Mobile's percentage of calls connected and successfully completed below that of both O2 and Vodafone. Ofcom agrees that ensuring comparability of data at the given point in time is a key issue in the context of its reconciliation exercise and hence a valid comparison of MNO data requires a comparable level of, amongst other things, network quality. Contrary to T-Mobile's suggestion, it is not clear to Ofcom that the average 1800MHz operator's service quality differed materially from that of the average combined 900/1800MHz operator at the date of comparison. As highlighted by T-Mobile, the Oftel call success rate survey indicates that T-Mobile had a lower service quality than the other three MNOs. However, the relevant comparison is the average network quality of the combined 900/1800MHz operators (Vodafone and O2) compared with that of the 1800MHz operators (Orange and T-Mobile). Since Orange's network quality had, historically, been the highest of the MNOs, the network quality for the average 1800MHz operator was very similar to that of the average combined 900/1800MHz operator in 2001.

C.76 T-Mobile states that if MNO network equipment information had been considered for a given level of quality of service, not only would it demonstrate the need for a greater number of cell sites required by 1800MHz operators than combined 900/1800MHz operators, but also that the number of cell sites for the average 1800MHz operator was substantially above that calculated by the LRIC model. The operating and capital cost adjustments made to the outputs of the LRIC model ensure that its cost estimates are in line with the average of the four MNOs. T-Mobile's arguments, however, relate to the relativities of the averages of the two combined 900/1800Mhz operators and the two 1800MHz operators. The networks of the MNOs have each evolved at different rates, based on the evolving capacity requirements and roll-out strategies of each MNO. Therefore, using high level information such as total network GBV, total network operating cost, and total number of sites is an imperfect means by which to establish whether or not the economics of a generic combined 900/1800MHz operator are more than, less than, or approximately as favourable as those of a generic 1800MHz operator. Ofcom has analysed data submitted to the CC by the MNOs regarding their number of cell sites, capacity in sectors, number of TRXs, total network GBV and depreciation, and total network operating costs. Each of these indicators has various merits and flaws when used to indicate which of the two mobile technologies is more costly to deploy. Ofcom has taken account of all of these measures. Based on its analysis, Ofcom's view is that the CC's conclusion, that neither the combined 900/1800MHz operators nor the 1800MHz operators had a significant cost advantage over the other at the date of comparison, is reasonable. Ofcom has also considered technical arguments made by both T-Mobile and Vodafone, each of which argues that their own network type is the more costly to deploy. Ofcom's view is that both submissions have a degree of merit, and the analysis of the MNO data is likely to represent the best way of determining the relative costs of each operator type.

C.77 T-Mobile argues that callers to mobiles experience direct benefits of increased coverage and quality and the proposals are discriminatory as they only allow the 1800MHz operators to recover a contribution based on a lower level of quality than that for combined 900/1800MHz operators. As stated in paragraph F.95 and expanded upon in paragraphs F.163-F.165 of the December consultation, it is unclear to Ofcom that callers to mobiles should be obliged to pay more for a higher quality of service chosen by call recipients. Ofcom does not believe that this position is discriminatory. As stated above, the average quality of the combined 900/1800MHz operators and average quality of the 1800MHz operators at the point of comparison appears to be similar, on the basis of the available information.

C.78 T-Mobile comments further in paragraph II.62 of its response why it believes that the calculation described in paragraph F.96 of the December consultation demonstrating the minimal impact on the LRIC model outputs to an increase in quality is flawed. It appears that T-Mobile may have misunderstood the intention of the analysis. The sensitivity was not designed to test T-Mobile's core concern relating to the average service quality and costs associated with each operator type but rather to test the impact of increasing overall modelled quality in the model after 2001. This sensitivity was aimed only at introducing a trend in blocking probability over time, i.e. departing from the assumption that it was constant in all periods.

C.79 Ofcom notes T-Mobile's clarification in paragraphs II.58-II.59 of its response regarding 3G cell site deployment. Paragraph F.94 of the December consultation sought to reiterate one of the advantages of relying on the CC's 2001 calibration point, rather than collecting new data, due to the need to focus on costs associated with 2G voice services. This text was not intended to assert that all of an operator's additional expenditure after 2001 would be driven by 3G costs. Information subsequently provided by T-Mobile does not change Ofcom's view of the relative costliness of running combined 900/1800MHz and 1800MHz networks. The operational and financial data provided to Ofcom by all the four MNOs is the key evidence that has led to this view.

C.80 In section 3.3.2.F of its response, Orange states that the adjustment to cancel out the operator-type cost differential ignores the significant higher costs of 1800MHz operators in providing rural and in-building coverage. Ofcom emphasises that it has not ignored factors arising from the difference in the nature of the 900MHz and 1800MHz spectrum in its consideration of this issue, as evidenced, for example, by the discussion in paragraph F.101 of the December consultation.

C.81 In conclusion, on the basis of the information available to Ofcom and the reasoning set out in paragraph F.98 of the December consultation, Ofcom believes that the view that neither operator type had a significant cost advantage over the other (on an accounting basis) is reasonable. Having given careful consideration to the potential differences in costs between the two types of networks, Ofcom considers that its treatment meets the tests in section 47(2)(b) of the Act.

1800MHz operator cost disadvantage overestimated

C.82 In paragraphs B.34-B.39 of its response, Vodafone continues to question the adequacy of the proposed adjustment of 0.2ppm to satisfactorily resolve the inter-operator network size differential. Vodafone states that this value only attempts to adjust for the perceived equality in network size looking forwards from the date of comparison and hence implies that the relative size of the differential generated by the model in prior years is regarded as reasonable. Vodafone states its continued belief that the network size differential in the model is exaggerated with the consequence that the economic depreciation differential is also exaggerated in years after 2001/02.

C.83 Regarding the detailed calculation of the adjustment, Vodafone states that its examination of the LRIC model indicates that the difference in HCA output in 2005/06 is 0.18ppm. However, Vodafone argues that this differential needs to be uplifted by the capital adjustment factor to reflect the fact that the figure has been generated by a model which underestimates the underlying network size. In section 3.3.2.F of its response, Orange notes that the CC derived a differential in the range of 0.13ppm to 0.2ppm and hence believes that the proposals are unfairly biased against 1800MHz operators by using the maximum within this range.

C.84 The path of cost recovery chosen for determining the efficient charge in 2005/06 is economic depreciation. Whilst Ofcom recognises that the use of economic depreciation means that the inter-operator differential in 2005/06 is partially determined by the extent of cost recovery achieved in earlier years, Ofcom does not believe there is compelling evidence to support Vodafone's belief that the differential has been overstated by the model in the period prior to 2001. A response to Vodafone's arguments regarding this issue is set out in paragraphs F.99-F.106 of the December consultation. Vodafone has not provided any further underlying reasoning for why the algorithms in the LRIC model should result in an overstatement of this differential.

C.85 However, Vodafone supplements its previous response by presenting a table showing that the GBV differential for the two types of operators predicted by the model is significant in earlier years. Ofcom regards this comparison as irrelevant. The objective is to determine the inter-operator differential, in terms of economic cost, in 2005/06. The underlying approach taken to estimating this differential compares the model with 2001 accounting data and then allows the model output to reflect any changes that may be appropriate in this differential between 2001 and 2005/06 on an economic cost basis. Since the basis for cost recovery used to determine the efficient charge level in 2005/06 is economic depreciation, the cost differential in historic years in accounting terms is not relevant.

C.86 Furthermore, Vodafone makes a number of observations which it claims indicate that the historic inter-operator differential is overstated. Specifically, Vodafone disagrees with Ofcom's position since: it is based on comparison of quantities rather than taking account of the cost of network units; the only differences between the two operator types exist at the radio layer; and Vodafone believes that structural bias in the model means that costs in the radio layer are overemphasised. In response, Ofcom does not believe that its position regarding the inter-operator differential only considers differences in quantities rather than costs. The position that, at current traffic levels of the MNOs, both operator types have a similar amount of network equipment and hence similar costs (see paragraph F.90 of the December consultation) is based on a comparison not only of inter-operator equipment levels but also inter-operator GBV. The unit costs used in model are identical for both types of operators, consistent with the view that prices for equipment are now very similar (see paragraph 2.303 of the CC report). Secondly, Ofcom agrees that the only difference between the two operator types exists at the radio layer (which is reflected in the LRIC model), however, Ofcom does not accept Vodafone's point regarding structural bias (as discussed in paragraphs C.61-C.70 above).

C.87 Regarding Vodafone's final point, following the amendments described in paragraphs C.13-C.15 and C.19-C.21 above relating to the revision of 2G spectrum pricing and treatment of assets which decline in quantity, the difference in HCA outputs in 2005/06 increases from 0.18ppm to 0.25ppm. Ofcom agrees that it would be more accurate to reflect the adjustments resulting from comparison with operator data in determining the size of the inter-operator differential since this adjustment is aimed at reflecting parity in the size of the operators' networks. However, Ofcom disagrees that it is appropriate to simply uplift the 0.25ppm figure by the capital cost adjustment. Rather, it is more appropriate to apply an adjustment which reflects the capital cost adjustment (38.7%) and operating cost adjustment (-8.5%). Since the objective is to determine the differential, in HCA terms, for 2005/06 it seems most appropriate to apply the capital and operating cost adjustments to the HCA mix of capital and operating costs in 2005/06 (33% in capital costs and 67% in operating costs according to the model output). The net impact is a 7% uplift to the 0.25ppm difference in HCA outputs in 2005/06 which results in a difference of 0.27ppm. Given that the LRIC model outputs have already been adjusted to reflect the average MNO across both operator types, this figure translates into an upward adjustment of 0.135ppm for combined 900/1800MHz operators and a downward adjustment of 0.135ppm for 1800MHz operators in 2005/06.

Appropriate basis for cost recovery

C.88 O2 (pages 2-3 of its response) notes that whilst the December consultation states the continued belief that combined 900/1800MHz operators and 1800MHz operators have similar costs on an HCA basis, it maintains the use of economic depreciation which has the effect of increasing the future termination revenues that the 1800MHz operators would enjoy, resulting in distortion of competition in the outgoing market.

C.89 In response to paragraph F.109 of the December consultation, which states that it is for the MNOs and their investors to decide upon the most appropriate basis for accounting and performance measurement, O2 believes that it is not practicable to amend its accounting methodology and hence the use of economic depreciation is discriminatory and Ofcom should use accounting depreciation instead.

C.90 Furthermore, O2 states that economic depreciation means that the costs for 1800MHz operators should have been lower than the costs for combined 900/1800MHz operators in the past. However, the proposals do not take into account the fact that 1800MHz operators have applied higher charges in the past.

C.91 In response to O2's first point, as set out in paragraphs F.108-F.110 of the December consultation, economic depreciation does give rise to a differential in the efficient charge level for the two types of operators. This differential reflects the difference in economic costs which are higher for 1800MHz operators. Ofcom maintains the opinion that economic depreciation provides the best view of the appropriate path of cost recovery over time for the purpose of setting regulated charges, being the path of costs that would prevail in a competitive market.

C.92 Regarding the appropriate basis for accounting and performance measurement, contrary to O2's belief, the December consultation (paragraph F.109) did not suggest that MNOs should amend their accounting methodology but merely sought to note that accounting methodology for statutory or commercial purposes is a separate matter which, in this context, is of less relevance than the appropriate basis of cost recovery for regulatory purposes. Ofcom notes that if a current cost accounting approach was taken to setting regulated charges, rather than economic depreciation, the efficient charge level in 2005/06 would be lower by almost 1ppm for combined 900/180MHz operators and about 1.5ppm for 1800MHz operators.

C.93 O2 is incorrect to suggest that economic depreciation ought to have resulted in the cost of termination for 1800MHz operators in the past being lower than for combined 900/1800MHz operators. Indeed, examination of the output of the LRIC model shows that the economic unit cost for 1800MHz operators is higher than for 900/1800MHz operators at all points in time (reflecting lower average utilisation).

Summary of net adjustments following comparison with MNO data

C.94 The derivation of the net adjustments to be made for the combined 900/1800MHz operators and 1800MHz operators in the years up to 2005/06 is shown in the table below.

 

Table 7: Net adjustments following comparison with MNO data
Pence per minute
(real 2000/01)		 2001/02 2002/03 2003/04 2004/05 2005/06
900/1800MHz operators  
LRIC+ (revised) 4.73 4.41 3.79 3.58 3.76
Capital cost adjustment 0.95 0.82 0.58 0.50 0.42
Operating cost adjustment -0.19 -0.19 -0.19 -0.19 -0.23
900MHz / 1800MHz adjustment 0.14 0.14 0.14 0.14 0.14
Net adjustment 0.89 0.76 0.52 0.44 0.33
Resulting LRIC+ 5.62 5.17 4.31 4.02 4.09
1800MHz operators  
LRIC+ (revised) 5.87 5.43 4.59 4.31 4.55
Capital cost adjustment 1.26 1.09 0.76 0.65 0.54
Operating cost adjustment -0.22 -0.22 -0.22 -0.22 -0.27
900MHz / 1800MHz adjustment -0.14 -0.14 -0.14 -0.14 -0.14
Net adjustment 0.91 0.74 0.40 0.29 0.14
Resulting LRIC+ 6.78 6.17 4.99 4.60 4.69

C.95 The net adjustments of +0.33ppm and +0.14ppm (in real 2000/01 terms) for combined 900/1800MHz operators and 1800MHz operators respectively are approximately 0.2ppm higher than those proposed in the December consultation (net adjustments of +0.11ppm and -0.05ppm for combined 900/1800MHz operators and 1800MHz operators respectively). These increases reflect the amendments set out above: revisions to the cost of capital; 2G spectrum pricing; the treatment of equipment that declines in quantity; and the size of the inter-operator differential.

Network common costs

Magnitude of network common costs

C.96 Vodafone raises the issue of the appropriate definition of fixed and common costs in paragraphs 1.48-1.52 of its response. Whilst Vodafone states that it will not repeat its previous arguments on this matter, it notes that the consultancy Analysys has taken a different view in its work for the Swedish national regulatory authority.

C.97 Ofcom's position regarding minimum coverage network common costs has been clearly set out in a number of documents, most recently in paragraphs F.128-F.142 of the December consultation and the references therein. For that reason, Ofcom's reasoning is not repeated here.

C.98 With regards to Vodafone's further observation, whilst Ofcom employed the consultants Analysys as its advisors on LRIC modelling and therefore gave consideration to Analysys' views, Ofcom itself was responsible for leading this process and developing this work. In this context, Ofcom does not consider the approach taken by Analysys in its work elsewhere to be relevant in itself, in the absence of new substantive arguments.

Allocation of network common costs to handsets

C.99 Vodafone submits that the presence of a handset cost in the LRIC model has no function other than for constructing an annual total which absorbs a portion of network common costs (paragraph B.23 of its response). It is unclear to Vodafone that this approach is appropriate given that network common costs are related to traffic not handsets.

C.100 As recognised in paragraph 7.18-7.19 of the Review of the Charge Control on Calls to Mobiles, published on 26 September 2001, some network costs depend on the number of customers that a network has rather than the volume of traffic. For example, the cost of handsets is clearly driven by the number of customers rather than the volume of traffic. Such costs are an integral aspect of the supply of mobile services. This is reflected in the LRIC model which considers two increments: subscribers and traffic (-49-). As explained in paragraph F.153 of the December consultation, Ofcom considers subscription to be a service in its own right with a demand, a cost (driven by number of customers) and a price. Giving due consideration to economically efficient prices, it is reasonable that the recovery of common costs should reflect that the cost of supplying mobile services is driven by both traffic volumes and subscriber numbers.

Non-network common costs

Treatment of customer acquisition, retention and service (CARS) costs

C.101 In paragraphs I.16-I.19 and II.41-45 of its response, T-Mobile reiterates its argument that the majority of CARS costs concern the acquisition and retention of customers and since the purpose of this expenditure is to earn revenue, including termination revenue, these costs represent a common cost. T-Mobile argues that common costs should be recovered in a way which maximises overall consumer welfare but the failure to recognise CARS costs as common costs results in retail prices which are substantially higher than is optimal. Since customers are sensitive to up-front charges, these costs will need to be recovered from outgoing calls resulting in an inefficient, unbalanced pricing structure.

C.102 T-Mobile also argues that the higher termination charges are above incremental cost, the greater the incentive to MNOs of attracting and retaining customers hence the greater the investment in CARS. Given that the proposed level of the charge for termination is above LRIC, T-Mobile claims that an element of CARS is incremental to termination.

C.103 T-Mobile has not raised any new substantive arguments to support its belief that CARS costs represent a common cost across all services. Ofcom remains of the view (shared by the CC (-50-)) that CARS costs are not causally related to termination since they do not vary with incoming traffic and therefore should not be included in either the LRIC, or the mark-up for common costs. Ofcom's reasoning is set out in detail in paragraphs F.147-F.157 of the December consultation. As stated in paragraph F.154 of the December consultation, Ofcom believes that a contribution to the recovery of CARS is relevant to the termination charge only through the network externality surcharge.

C.104 Ofcom considers CARS costs to be incremental to the provision of retail services, specifically subscription and mobile-originated calls, hence these costs should be recovered from the (unregulated) price of retail mobile services which MNOs are free to structure as they wish (see paragraphs F.149 and F.153 of the December consultation). Since Ofcom does not agree that CARS costs are common, it does not accept T-Mobile's view that Ofcom's efficient charge level for termination will lead to retail prices which are higher than is optimal or that the resulting pricing structure would be inefficient (since it is inappropriate to recover the vast majority of these costs from termination services).

C.105 Ofcom agrees with T-Mobile that the higher termination charges are raised above incremental cost, the greater the value of attracting / retaining subscribers and the greater the willingness to invest in CARS. However, this does not mean that there is an element of these costs which are incremental to termination. T-Mobile's view would appear to be based on a possible misunderstanding of what is meant by incremental in the context of the principle of LRIC. A cost should not be included in the LRIC of a service if it is not incurred or caused in the long run by that service (-51-). T-Mobile's argument appears to be based on a demand-driven view rather than a regard for the cost of supplying a service. T-Mobile's apparent confusion is further evidenced by its claim (footnote 57 of its response) that "the CC accepted that an element of acquisition and retention costs was incremental to termination" in paragraph 2.327 of the CC report. This is not the case. The CC accepted that "there is probably some correlation between the intensity of call making and that of call receiving" and hence a correlation between investment in CARS and the volume of incoming calls. However, this observation does not lead to the conclusion that part of CARS costs is incremental to termination. As stated in paragraph F.149 of the December consultation, CARS costs are not incurred by the supply of termination services but are caused by the desire to acquire and retain retail customers and so are incremental to retail activities.

Inclusion of all MNO cost accounting data

C.106 Vodafone (paragraphs B.42-B.44 of its response) seeks confirmation that Orange's 'further network overhead costs' discussed in paragraph F.173 of the December consultation have been included in the calculation of the average MNO network operating cost figure of (GBP) (GBP) £338m and that Vodafone's network operating cost total, as reported to the CC, is also appropriately included in this calculation.

C.107 As indicated in paragraph F.173 of the December consultation (-52-), whilst the CC removed Orange's 'further network overhead costs' to derive a non-network administration cost figure, these 'further network overhead costs' have been reallocated to Orange's network operating cost figure, which is then used in the calculation of the average of (GBP) (GBP) £338m, rather than failing to be recovered. Furthermore, Ofcom confirms that Vodafone's network operating cost figure, as reported to the CC, is indeed included in the derivation of the average of (GBP) (GBP) £338m.

Allocation of non-network administration costs

C.108 As stated in paragraph B.53 of its response, Vodafone continues to believe that the appropriate cost-causation approach to allocation of non-network administration costs should consider retail costs composed of sales and marketing and customer care costs but excluding the costs associated with handsets, discounts and incentives.

C.109 However, following the methodology to allocate administration costs between network and retail costs set out in the December consultation (paragraph F.181), Vodafone (paragraphs B.45-B.52 of its response) believes that all cost elements included in the retail activities category have been taken account of, but the costs included in the network category are not similarly complete. Orange makes a similar point in Section 3.3.2.E of its response. On this basis, Vodafone and Orange believe that there are two network costs missing:

C.110 Orange makes the further point that there is a trend for retailers to source handsets directly and so it is unclear that operators will continue to provide handsets in the future. Orange also states that if Ofcom shares the CC's objective that the frequency of handset replacement is reduced then it is only consistent for the purpose of cost modelling to assume that lower levels of handset subsidy prevail in 2005/06.

C.111 T-Mobile argues in paragraph II.46 of its response that these non-network common costs should be recovered so as to maximise efficiency and overall consumer benefits. Instead, T-Mobile states that the proposed approach in the December consultation is arbitrary and not even based on EPMU which would result in percentage mark-ups on costs which are the same rather than mark-ups on different voice minutes which are equal.

C.112 In response to Vodafone's previous point regarding the correct treatment of administrative costs from a cost-causation principle, Ofcom's view is that common costs should be recovered from all the activities that these costs help to support. This position is stated more fully in paragraph F.186 of the December consultation.

C.113 Regarding Vodafone and Orange's concern that network costs have been understated because depreciation alone fails to capture the full cost of acquiring the network, Ofcom agrees that a better measure of this cost is the sum of network depreciation and the cost of capital associated with the network assets. Ofcom has therefore amended its assessment of the relevant average MNO network costs in 2001 for the purpose of allocating non-network administration costs. As before, Ofcom has taken the average network cost (excluding cost of capital) of (GBP) (GBP) £607m (-53-) but added a further (GBP) (GBP) £248m to allow for cost of capital, derived from the average MNO network net book value of (GBP) (GBP) £1,639m (-54-) and a nominal cost of capital of 15.14% (-55-) which corresponds to the real rate of 12%. This gives rise to a total average MNO network cost of (GBP) (GBP) £855m. Comparing this network cost to the total average retail (CARS) costs in 2001 of (GBP) (GBP) £1,276m (-56-) results in 40% ((GBP) (GBP) £855m / ((GBP) (GBP) £855m + (GBP) (GBP) £1,276m)) of the (GBP) (GBP) £159m (-57-) administrative overheads being allocated to network activities using an equal proportionate mark-up (EPMU). Dividing the resulting administrative cost attributable to network by the average total minutes for the year (comprising incoming, outgoing and on-net voice minutes) of 15.5 billion (-58-) results in a non-network common cost mark-up to the LRIC of voice call termination of 0.41ppm (in real 2000/01 terms). This compares with a non-network common cost mark-up of 0.33ppm proposed in the December consultation.

C.114 In contrast, Ofcom disagrees with Vodafone and Orange that a further amendment should be made for interconnect costs. The (GBP) (GBP) £410m referred to by Vodafone, and presented by in Table 7.2 of the CC report, refers to interconnect payments associated with outgoing calls. As noted in Table 7.2 of the CC report, these costs are not relevant to incoming calls and therefore not relevant to the mark-up on termination. These costs are not incurred in relation to a MNO's own network but are costs incurred in the provision of retail (outgoing) calls. Hence there is an argument for including these interconnect costs in the retail (non-network) cost category but not in the network cost category. In the interests of ensuring that the non-network common cost mark-up is not understated, Ofcom has chosen not to add interconnect costs to the retail cost category but to maintain its existing approach.

C.115 In response to Orange's further points, Ofcom notes Orange's view that MNOs' provision of handsets may reduce in the longer term if retailers increasingly source handsets directly from suppliers, however, Ofcom believes that the impact of this is likely to be minimal during the period of the charge control which ends in March 2006. With regards to reflecting lower handset subsidies in 2005/06, the level of handset subsidies is irrelevant to the calculation of the non-network common cost mark-up which refers to gross handset costs.

C.116 T-Mobile's belief that these common costs should be recovered so as to maximise efficiency and overall consumer benefit appears to imply the use of Ramsey mark-ups. As stated in paragraph 6.8 of Chapter 6 and Annex K of the December consultation, Ofcom believes that it is more appropriate to recover these costs on the basis of the relative proportions of network costs to the total of network and retail costs. This results in equal proportionate mark-ups on network services and retail services. However, T-Mobile is correct in observing that the mark-up approach used is a variant of an EPMU approach. There is a two-stage approach. First, a part of the non-network common cost is allocated to network traffic services using an EPMU on retail and network traffic costs. Then, the allocation of non-network common costs to network services ((GBP) (GBP) £64m) is recovered through a fixed mark-up per minute which is the same for all traffic services (outgoing, incoming and on-net voice minutes), rather than through a mark-up on the cost of each of these traffic services. Ofcom has calculated the non-network common cost mark-up on termination resulting from a mark-up on the cost of each service rather than a fixed mark-up per minute, using the economic cost recovered for the three traffic services in 2005/06. Whilst T-Mobile is correct in its belief that this leads to a higher per minute mark-up on termination than on outgoing calls, on-net call minutes have a significantly higher cost than termination (because they include both ends of the call) and therefore attract an even higher per minute mark-up under this approach. The net effect would be a non-network common cost mark-up on termination of 0.31ppm which is lower than the 0.41ppm using the existing approach, after correction for the network cost of capital. Whilst both methods are a reasonable approach to the recovery of non-network administrative overheads, Ofcom sees insufficient reason to merit changing its approach and believes the existing method has the additional benefit of ensuring that the non-network common cost mark-up is not understated.

Conclusion

C.117 In conclusion, the table below sets out the amended LRIC+ figures (at 12.25% cost of capital) from the December consultation and the further adjustments discussed in this annex to derive the LRIC+ figures that include the EPMU for network and non-network common costs. The efficient charge level is then determined by the sum of the adjusted LRIC+ figures below and the surcharge for the network externality (discussed in Annex D).

C.118 The resulting LRIC+ figures (excluding the network externality surcharge) in 2005/06 of 4.50ppm and 5.10ppm (in real 2000/01 terms) for combined 900/1800MHz and 1800MHz operators respectively, are approximately 0.30ppm higher than those proposed in the December consultation.

Table 8: Revised LRIC+ (excluding network externality) figures

 

Table 8: Revised LRIC+ (excluding network externality) figures
Pence per minute
(real 2000/01)		 2001/02 2002/03 2003/04 2004/05 2005/06
900/1800MHz operators  
Model LRIC+ (December consultation) 5.07 4.76 4.15 3.95 3.76
Cost of capital adjustment - 12% 0.02 0.02 -0.04 -0.03 -0.03
Revision for 2G spectrum pricing -0.23 -0.25 -0.22 -0.22 0.13
Revision for declining equipment quantities -0.13 -0.12 -0.11