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

Introduction

D.1 In the May and December consultations, Ofcom proposed that it would be appropriate to allow MNOs to add an additional mark-up on cost when setting charges for mobile termination services. This mark-up (or surcharge) was designed to ensure MNOs account for the external benefits that callers to and from mobile telephones receive from the addition of new subscribers to the network, and the maintenance of existing subscribers on the network.

D.2 In both of these consultations, caution was expressed regarding the estimation of a surcharge, noting that the conceptual and practical estimation obstacles were formidable. A judgement was made on the basis of a range of estimates produced by different models of behaviour in wholesale and retail mobile markets. Each of these estimates provided a relevant, although incomplete, perspective on the efficient surcharge.

D.3 Ofcom maintains a similar approach to the calculation of the appropriate surcharge in this Statement. The following sections contain a brief outline of the approach in the December consultation, responses to the December consultation and Ofcom's comments on these, and conclusions.

Summary of previous approach

D.4 As noted in paragraph D.1, the purpose of the network externality surcharge is to correct for potential economic inefficiencies that may be created if the subscription charge levied by MNOs only reflected the costs of supply. Consumer welfare could be potentially improved if some consumers who would not otherwise join a mobile network had their subscription subsidised, because these consumers' joining decisions increase the welfare of existing subscribers. To the extent that any such subsidies need to be funded by MNOs, Ofcom believes it would be appropriate for wholesale mobile termination charges to include a contribution towards the recovery of these subsidies.

D.5 Paragraphs G.4 to G.9 of the December consultation provide further justification for a network externality surcharge.

D.6 As explained in paragraph D.2, estimation of an appropriate surcharge is very complex. Six estimates were provided in the December consultation, each reflecting different but individually relevant considerations. These estimates, and the relevant considerations, are presented below for reference.

 

Table 1: Summary of estimates used in the December consultation

Source	Description	Optimal surcharge
Rohlfs targeting model*	Incorporates ability of MNOs to distinguish marginal and infra-marginal subscribers through price discrimination	0.07ppm
Rohlfs principal-agent model*	Incorporates MNOs sub-optimal use of higher mark-ups on termination	0.07ppm
Previous surcharge (MMC)	See MMC report, appendix 5.2 for further details.	0.50ppm
CC report	See CC report, appendix 8.1 for further details.	0.45ppm
Rohlfs no targeting model*	A linear pricing model (no price discrimination) with some internalisation of externalities by MNOs assumed.	0.49ppm
Rohlfs model - reduced internalisation*	Reduces assumptions about amount of externality internalised by MNOs (increasing the usage cross-elasticities, j2 = 0.5, j4 = 0.5, n = 0.25).	0.67ppm

* The Rohlfs models were updated to take account of revised LRIC inputs.

D.7 The Rohlfs models referred to in the table are described in more detail in Dr Rohlfs' paper A Model of Prices and Costs of Mobile Network Operators (May 22, 2002). The Rohlfs models provide an estimate of the optimal mark-up to recover common costs as well as an adjustment for externalities. Given that Oftel preferred to use the EPMU approach for the recovery of common costs, adjustments to the estimates to isolate the effect of externalities on the optimal set of prices were required. The method by which this done is discussed further at paragraph D.36.

D.8 On the basis of the available evidence, 0.4ppm was considered to be a reasonable externality surcharge.

Responses to the December consultation

BT / UKCTA

D.9 BT did not believe an externality surcharge was justified. BT suggested that an untargeted scheme, which provides for a subsidy to both infra and inter-marginal consumers, is unlikely to deliver efficiency benefits. Fixed line users will subsidise all mobile users, most of whom would have a handset even without a subsidy. Further:

• there is a surcharge on fixed users to subsidise mobile users, but no equivalent surcharge to fund fixed users; and
• there is no scheme to attract marginal mobile users in place.

D.10 Hence, the effect is that the subsidy effectively promotes mobile services over fixed services. The UK Competitive Telecommunications Association (UKCTA) provided similar comments.

D.11 Ofcom recognises that targeting of marginal consumers is an important issue with respect to the calculation of the appropriate externality surcharge. It is indeed the case that where there is no specific scheme in place to subsidise marginal consumers, subsidies could well be offered to both infra-marginal and marginal subscribers. It is also true that if substantial targeting of subsidies to marginal consumers occurs, the appropriate subsidy is likely to be significantly lower. However, the lack of ability to target does not imply that no subsidy is appropriate - the welfare costs of raising the surcharge must be balanced against the potential benefits that a surcharge will deliver. This is illustrated in the Rohlfs "no targeting" model, in which it is assumed all subscribers are offered the same (subsidised) subscription price.

D.12 On the issue of consistency of treatment between the fixed and mobile sectors, Ofcom believes that Ofcom's comments in the December consultation address this point (see Annex G.18). Similar objectives exist with respect to marginal subscribers on both types of networks. However, for a combination of historical and efficiency reasons (closely related to its USO obligations), BT has financed schemes aimed at marginal subscribers to fixed networks out of profits from supplying call services. Ofcom's forthcoming review of the Universal Service Obligation (scheduled for summer 2004) will likely examine this issue further.

D.13 The UKCTA suggested that the practical evidence cast doubt on the need for a mark-up to cover network externalities. In particular, it referred to Professor Martin Cave's statement that up until 2000-01 the MNOs aimed to maximise subscriber numbers, but after this:

"... they re-focused their growth policy, reduced subsidies and some of them saw the number of subscribers fall. They changed their growth policy from maximisation of subscriber growth to optimisation and maximisation of profits. This suggests that we might have had a higher than optimal level of mobile penetration at that time."

(UKCTA, page 5)

Ofcom's response

D.14 Ofcom does not consider this point undermines the rationale for a network externality surcharge. The surcharge effectively promotes behaviour (subsidisation of marginal subscribers) intended to promote overall consumer welfare. This may involve providing subsidies to consumers to either join a network, or to maintain their network subscription. Consequently, even if it could be shown that at a point in time penetration was already at or above the efficient level, it would not follow that for future periods no subsidy was justified. It may well be efficient for subsidies to be provided to maintain existing subscribers on the network.

D.15 UKCTA also cites from the Cave report to question an adjustment for the network externality, but not for the "call externality". The call externality relates to the originating party, who pays for the call, failing to take into account the benefits of the call to the receiving party (who may value the call, but pays nothing). Hence, internalising the call externality might result in optimal call termination charges being below cost.

"Mobile networks, as well as other telecoms networks, are characterised by network and call externalities. If network externalities prevail, access to the network should be subsidised in an efficient manner to internalise those effects. Handset subsidies in mobile networks partially fulfil this function. Fixed network users may contribute to that subsidy by paying termination charges above costs. On the other hand, termination charges below cost help to internalise call externalities."

(UKCTA, page 6)

Ofcom's response

D.16 Call externalities - while they almost certainly do exist - probably do not justify any adjustment to call prices. As noted in Oftel's Review of the Charge Control on Calls to Mobiles (2001), and in the CC report, these are likely to be effectively internalised by callers, as a high percentage of calls are from known parties and there are likely to be implicit or explicit agreements to split the origination of calls.

Vodafone

D.17 Vodafone rejects Oftel's economic reasoning in relation to the Rohlfs-Griffin ^(-59-) (R G) factor, and states that it is 'purely and simply an empirical matter'. Vodafone also rejects Oftel's comments about the exclusion of unobserved taste effects, noting that the Frontier estimates contain time trends which were specifically designed to pick up such effects.

D.18 Vodafone also re-iterates its view that off-net minutes should be excluded from the denominator of the externality calculation (in relation to the CC estimate). Vodafone suggests Oftel's position - that the externality surcharge should not be solely levied on fixed-to-mobile calls - would be correct if all prices were to be adjusted by Oftel to their optimum levels, but as the industry generates no net revenue from termination of off-net calls, it cannot be assumed that MNOs can generate funds for targeting marginal subscribers from off-net mobile-to-mobile calls.

Ofcom's response

D.19 Ofcom rejects Vodafone's interpretation that a priori economic reasoning should be ignored when deriving the R-G factor. As noted in both the May and December consultations (see Annex G), empirically-estimated R-G factors of above two are simply implausible and strongly suggestive of estimation bias. The CC also agreed with this approach (see paragraph 2.372).

D.20 Ofcom has partially addressed Vodafone's point about off-net minutes in the December consultation (see paragraphs G.75-G.76), noting that, in principle, it would be more efficient to recover the surcharge across all mobile termination services (whether used for fixed-to-mobile or off-net mobile calls). Ofcom does not agree with Vodafone's further suggestion that higher termination charges for off-net calls will be 'revenue neutral' and will not provide more funds to subsidise marginal subscribers. Vodafone's interpretation would only be correct if retail charges for off-net calls were invariant to the termination charge. However, Ofcom believes that higher termination charges are highly likely to feed into higher retail prices (because these charges form part of the marginal cost of a call, from the perspective of the originating operator). This implies that higher termination charges are likely to generate funds to subsidise marginal subscribers, and Ofcom therefore rejects Vodafone's argument that the CC calculation was incorrect because it excluded mobile-to-mobile call minutes.

Orange

D.21 Orange does not make any additional comments to those in response to the May Consultation, only noting that in its view the Director had chosen a surcharge at the extreme lower end of a reasonable range of estimates.

T-Mobile

D.22 T-Mobile states that five key assumptions underlying Oftel's analysis raise concern:

1. the level of the R-G factor assumed;
   
   
2. the ability of MNOs to target funds raised by a surcharge at marginal customers;
   
   
3. whether the surcharge can be recovered from all mobile services;
   
   
4. the assumed nature of retail competition; and
   
   
5. whether the level of surcharge can be determined independently of the level of fixed and common costs.

Level of R-G Factor assumed

D.23 On point (a), T-Mobile claims that Oftel's assumption of 1.5 for the R-G factor implies that half of the external benefits generated by increasing mobile subscription are internalised, which is inconsistent with the CC's evidence.

D.24 It was the view of Oftel (see annex G.26-G.29, December Consultation) and the CC (see paragraphs 2.372) that reasonable bounds for the R-G were between 1 and 2, with R-G factors of over two being implausible. Similarly, both Oftel (see paragraph G.23, December consultation) and the CC (see paragraphs 2.350 and 2.374) concluded that a reasonable upper limit for the R-G factor in practice - that is, taking into account the likely gross externality factor and likely internalisation by consumers, was 1.5-1.7. This upper limit accounts for both the likelihood that the 'gross' externality factor is below 2, as well as allowing for some internalisation by consumers.

Ability to target funds on marginal customers

D.25 T-Mobile made a number of further sub-points on targeting. In particular, that:

• Oftel has regard to models which Rohlfs has rejected, and doesn't have regard to other results from Rohlfs which are not inconsistent with a surcharge of 2ppm (nor has Oftel sought to update the reasonable range in the Rohlfs model taking into account new cost information). (T-Mobile, part II, paragraph 75)
• Oftel does not provide any support for the assumption in its modelling that all of the termination surcharge raised will be spent on non-subscribing marginal customers. (T-Mobile, part II, paragraph 76)
• As a surcharge increases profitability of all customers, competition will force MNOs to use funds on all subscribers (i.e. including infra-marginal subscribers). MNOs will only bring on marginal customers if they are self-financing. (T-Mobile, part II, paragraph 77)
• Because subsidies are a transfer, Oftel's suggestion that limited targeting means that the subsidy should be lower is incorrect. There is no reason it should be small - it should be at the level which creates the optimal number of subscribers on mobile networks. T-Mobile, part II, paragraph 78)

D.26 The issue of targeting was extensively addressed in Annex G of the December consultation (which also contains references to earlier Oftel work and the CC report). While the majority of T-Mobile's points were already addressed in that consultation, some additional comments and clarifications are now added.

D.27 It is Ofcom's view that T-Mobile's statements regarding the use of the Rohlfs model are incorrect. Rohlfs rejects estimates of 2ppm that were derived using his model as not being the most reasonable ^(-60-) . Rohlfs also does not reject the results of his targeting model, merely noting that MNOs may not have the ability to price discriminate to the extent modelled in the 2 or 3 part pricing plans ^(-61-) . The Rohlfs model has also been updated to correct for revisions to the LRIC model, and is updated further for this Statement. The input values and parameters used in the December consultation were the same as those used in the cost-benefit analysis; see Annex L of the December consultation. In this Statement, these inputs have been further updated.

D.28 The estimates used in the December consultation to inform the judgement as to the appropriate surcharge made a number of different assumptions with respect to targeting. For example, the estimate from the "Rohlfs no targeting" model assumes a surcharge is set on all fixed-to-mobile calls and this is passed through as a lower average subscription charge for all subscribers. The Rohlfs targeting model estimates the surcharge under the assumption that subsidies can be targeted to separate marginal and infra-marginal subscribers. Both of these estimates form part of a range, reflecting that while it is unreasonable to assume no incentive or ability to target, it is also probably unreasonable to assume that all subsidies would be directed to marginal subscribers.

D.29 As indicated in the December consultation, if it was the case that the incentive to target was overstated, and the assumption that all or most of the surcharge was spent on infra-marginal consumers was adopted, it would not lead to the conclusion that the surcharge should be increased. It may be that any subsidies that accrue to infra-marginal users are transfers - that is, of themselves they have no net welfare consequence - but there are clear adverse welfare consequences from the higher termination charges which must be raised to finance the (ever-larger) subsidies. Given these welfare losses, substantial wastage of the subsidy would therefore suggest that the subsidy should be reduced ^(-62-) .

Surcharge recovered from all services

D.30 On point (c), T-Mobile re-iterates the points made in its response to the May consultation, namely that the surcharge can not be levied on competitive services, as suggested in modelling.

"Competition does not permit MNOs suddenly to set their subscription and outgoing call charges higher so that they can generate a pool of funds to bring unprofitable customers onto their networks."

(T-Mobile, part II, paragraph 79)

D.31 Again, this is an issue that Ofcom believes was addressed in the December consultation (paragraphs G.46-G.51). In determining the set of efficient Ramsey prices including externality effects, there is a trade-off between the benefits from effectively correcting for the externality (by subsidising the price of subscription) and raising the price of other services to fund the subsidy. This means that the 'second best' price for subscription will clearly be above the 'first best' price, in which the subsidy is assumed to be funded outside of the model. When considering questions of funding, it is more efficient to raise the price of all other services supplied by MNOs rather than just mobile termination.

D.32 Ofcom believes that T-Mobile has misinterpreted the modelling of the recovery of the surcharge across all services. In the "no targeting" model, once the first best price for subscription is determined, it is necessary for all prices to be marked up for common cost and subsidy recovery so that MNOs do not lose money. This result - that all prices will be higher than otherwise - is driven by efficient cost recovery and is independent of the level of competition between the MNOs. To see this, suppose that the optimal subsidies to marginal subscribers were provided (the first best). If the mark-up on call termination is then fixed by regulation, all MNOs will need to raise prices for these other services (including subscription) so as to ensure cost recovery. The results of the model are hence consistent with a competitive market in which Ramsey-type cost recovery principles are used.

Assumed nature of retail competition

D.33 On point (d), T-Mobile argues that the use of the estimate based on a 'principal-agent' model of regulation was inappropriate, as the model results run counter to market facts - with a market outcome of lower subscription and high call prices, rather than the other way around as predicted by the model.

D.34 The principal-agent model is designed to capture relevant features of MNO behaviour - in particular, their maximisation of the surplus accruing to their own subscribers, and not of subscribers (including fixed subscribers) in general. But this involves simplification of complex issues, for example, there is no distinction between the subscription or usage prices paid by different users. This model results in a lower optimal mark-up on mobile termination, as the model assumes MNOs will use higher mark-ups to lower usage prices to infra-marginal subscribers, rather than lowering subscription prices. Given the smaller subscription subsidy, the external benefits derived from a surcharge are lower and it is therefore optimal to have lower mark-ups. This is further addressed in the May consultation, page 216. While T-Mobile claims the outcomes of this model are not consistent with current market outcomes, T-Mobile suggests earlier in its submission (part II, paragraph 77) that it will use the additional revenue earned from a mobile termination mark-up to compete for infra-marginal customers. Further, while the average prices paid may appear different, the model takes no account of the price discrimination and non-linear pricing which is commonly practiced by the MNOs in the retail market (which leads to many different subscription and call charge combinations). Ofcom considers that the modelling of retail competition in the 'principal-agent' model is reasonable, although it accepts that this is a complex matter and other models of retail competition would be possible.

Whether the surcharge can be determined independently of common costs

D.35 On point (e), T-Mobile argues that Oftel underestimated the optimal surcharge by removing fixed costs from the Rohlfs model:

" In fact, Oftel will underestimate the appropriate surcharge if it chooses now to assume that there are no fixed and common costs when estimating the surcharge (i.e. by modelling the appropriate mark-ups separately). In particular, the existence of fixed and common costs implies a higher overall level of mobile prices (than if there were no such costs) and therefore a lower number of subscribers as the base from which to determine the appropriate externality surcharge. The lower initial subscriber base will imply a higher optimal externality surcharge than estimated by Oftel when it assumes there are no fixed or common costs in its externality modelling."

D.36 Before further discussion of this issue, it may be helpful to recap the approach taken in the May and December consultations, taking the Rohlfs model with no targeting of subsidies, i.e. with all subscribers paying the same price (the "no targeting" model) as the example. With the base case assumptions, this model yielded an optimal mark-up over LRIC for fixed-to-mobile retail calls of 0.77ppm. However, this mark-up included an allowance for the recovery of common costs as well as an adjustment for externalities. Ofcom has decided to use an EPMU approach to common costs (for reasons set out from paragraph 6.8 onwards). It has used the Rohlfs models to inform a reasonable figure for the externality surcharge, not the common cost mark-up. To 'strip out' the common cost recovery element, an amount equal to the EPMU (0.28ppm, leaving an optimal surcharge of 0.49ppm) was removed.

D.37 It was noted in the December consultation that the method used in the May consultation was imprecise, and a new method was developed. At the same time, adjustments were made to the cost inputs - in particular, common costs increased due to the inclusion of non-network common costs. These common cost adjustments led to a significant increase in the optimal mark-up (to around 1.5ppm), although as this was primarily driven by common cost adjustment, the effect on the mark-up for externalities was thought to be minimal. The new methodology effectively abstracted from the issue of mark-ups to recover common costs - that is, common costs were set to zero - and the optimal mark-up was then recalculated. The 'upper bound' mark-up was found to be 0.49ppm - the same as that previously used. The other reported mark-ups (in the table on page 1) were also recalculated using this approach.

D.38 In response to T-Mobile's comments, Ofcom has further examined its methodology and concluded that the level of common costs does have an influence on the optimal externality surcharge - although this influence is not straightforward, and depending on the circumstances, could change the surcharge in either direction. In the specific circumstances of the calculations in the December consultation, in which common costs were removed, Ofcom finds that it is likely that the optimal surcharge has been understated for two of the models (specifically in the "no targeting" model and the "less internalisation" models).

D.39 Ofcom's purpose in using the Rohlfs models is to inform the size of the externality surcharge, given that the EPMU approach is to be used for common cost recovery. Consequently, Ofcom has revised its approach to calculate the optimal mark-up for fixed-to-mobile calls by using as the cost inputs LRIC plus EPMU (with common costs set to zero). This approach ensures that all relevant costs to be recovered are included.

D.40 This approach leads to the following revisions to the estimates using the Rohlfs model. Again, these estimates have been re-calculated to take into account the latest available information on LRIC and common costs.

 

Table 2: Revised estimates

Source	December surcharge	Revised surcharge
Rohlfs targeting model	0.07ppm	0.06ppm
Rohlfs principal-agent model	0.07ppm	0.02ppm
Rohlfs no targeting model	0.49ppm	0.66ppm
Rohlfs model - reduced internalisation	0.67ppm	0.90ppm

D.41 It can be seen this leads to revisions of the December consultation estimates, in the first two models by small reductions, and in the latter two models by increases of around 0.2ppm.

D.42 This means that the previous figure of 0.4ppm appears below the midpoint of the estimates, whereas it had previously been above the midpoint. Taking into account the revised figures in Table 2, it is considered that an upwards revision is appropriate. Based on the revised figures, and the other figures in Table 1, Ofcom concludes that a surcharge of 0.5ppm is reasonable.

Conclusion

D.43 Ofcom considers that, broadly speaking, the estimates used in the previous consultations remain relevant to the decision about an appropriate externality surcharge. However, given the upwards revisions to two of these estimates, Ofcom believes it would be appropriate to allow an additional 0.1ppm for the externality surcharge. This takes the appropriate surcharge to 0.5ppm.

---

Footnotes:

59:- The R-G factor (or gross externality factor) is the ratio of social benefit to private benefit. In this context, it provides a measure of the externalities associated with the addition of subscribers to a network. See Annex G of the December consultation for more detail.

60:- See Oftel, Ramsey prices and Network Externalities: Dr Rohlfs' Analysis, 23 May 2002, footnote 6.

61:- See Rohlfs, Response to the Competition Commission - Estimates of targeted subsidies, 19 June 2002, p. 5.

62:- T-Mobile's analysis of publicly-provided education or health, which is considered to be desirable even if it not targeted to those who could otherwise not afford to pay, is clearly lacking in the same respect. The Government in this instance would clearly need to consider whether the deadweight losses from higher tax revenue (which relate to the total size of the subsidy) are outweighed by benefits which accrue to less-well-off citizens.

next section

Back to top