Statement Issued by the Director General of Telecommunications

November 2000

Contents

Chapter 1

Annex 1

Annex 2

Chapter 1

On 3 July 2000, the Director General published for consultation a draft Direction to BT Cellnet under Condition 66 of its mobile Public Telecommunications Operator’s licence. The draft Direction, if confirmed, would have required BT Cellnet to demonstrate to the Director General each quarter, commencing with the quarter ending 31 December 2000, that it was not engaging in any unfair cross-subsidy of its ‘tied’ service providers, as measured by the formula set out in the annex to the Draft Direction and known as the ‘Oftel formula’.

For the reasons set out more fully in Annex I (‘Background’), the Director General has concluded that in the light of audited returns from BT Cellnet showing (separately) performance for pre-pay and post-pay services for the quarters January-March and April–June 2000, BT Cellnet’s operations comply with the Oftel formula and do not, in consequence, give rise to an unfair cross-subsidy contrary to Condition 66. Accordingly the Director General has decided not to confirm the Draft Direction.

Also annexed to this Statement is a summary of the responses received to the consultation and Oftel’s comments on them (Annex II).

ANNEX I

BACKGROUND

Introduction

1.1 Condition 66 of Cellnet’s licence provides that, where the Director General considers that the Licensee is unfairly cross-subsidising, among other things, any activity carried on by another member of the Licensee’s Group, the Licensee shall take such steps as the Director General may direct for the purpose of remedying the situation.

1.2 In its 1999 statement on the mobile telephony market Oftel indicated that there was prima facie evidence that BT Cellnet and Vodafone had conducted a margin squeeze by cross-subsidising their tied service providers (TSPs) with profits taken at the wholesale level, limiting the ability of independent service providers (ISPs) to compete with differentiated products. Both BT Cellnet and Vodafone had failed to achieve the required minimum rate of return set by Oftel in 1997. Accordingly, Oftel had launched a formal investigation with a view to deciding whether or not to issue a Direction or to take other enforcement action. ( Review of the Mobile Market, July 1999 paras 2.5-2.8)

1.3 That statement indicated that (in accordance with Oftel’s normal procedures) the Director General would, in deciding on what action to take, take account of factors including:

• whether there had been a breach of a licence condition, and if so, whether such breach was continuing or likely to be repeated;
• whether any cross-subsidy had, or was likely to have, a material effect on competition;
• in the case of a licence breach, what steps might be required to remedy it; and
• whether it was appropriate to issue Directions under the licences of BT Cellnet and Vodafone to remedy the effects of any cross-subsidy, and if so, in what terms.

The ‘Oftel Formula’

1.4 In order to determine whether Vodafone and BT Cellnet are unfairly (ie anticompetitively) cross-subsidising their TSPs, Oftel has developed a statistical formula, based on returns of financial information from BT Cellnet and Vodafone. The Oftel formula uses a discounting technique to assess  whether a service provider is making an adequate return over the life of a subscription once the cost of acquiring that customer has been taken into consideration. The results are expressed in terms of pounds per subscriber. If, after the data has been put in to the Oftel formula, a positive value results, then the TSP is earning a profit for each subscriber. If the results are negative then the TSP is making a loss for each subscriber, implying a cross-subsidy from elsewhere within the Licensee’s Group. In the case of a supplier with market power, a negative figure would suggest that an ‘unfair’ cross-subsidy had occurred, because of the possibility that it would have had a material effect on competition.

1.5 In the course of its investigation Oftel updated the formula to take into consideration the changing characteristics of the mobile market. Changes made to the formula included the use of cost and revenue information disaggregated between pre-pay and post-pay products and services. Oftel also recalculated the ‘churn rate’ (the frequency with which an average subscriber is likely to switch suppliers) to include a number of different possible periods for the average duration of a subscription (‘subscription life’). These ranged from 22–26 months for post-paid and from 45-56 for pre-paid. Oftel also updated the cost of capital calculation used in the Oftel formula to reflect the changes in the financial markets.

The Draft Direction

1.6 It appeared to the Director General, based on information before him and supplied to him by BT Cellnet in June 2000, that for the period January -March 2000, BT Cellnet’s cross-subsidy of its tied SPs – which according to that information amounted, on average, to between £1.50 and £8.00 per subscriber – was anti-competitive and accordingly unfair, contrary to the terms of Condition 66 of BT Cellnet’s mobile PTO licence. The information indicated that BT Cellnet, as an operator with some degree of market power in the mobile market, was able to leverage its market power in the upstream (network) level of the market into the downstream (service provision) level and so distort competition. The cross-subsidy could have the effect of driving ISPs from the market by allowing BT Cellnet’s TSPs to operate at a level of loss that ISPs, who did not have the benefit of this additional source of revenue, could not sustain. The Director General therefore issued for consultation, on 3 July 2000, a Draft of a Direction which, if confirmed, would require BT Cellnet to demonstrate to the Director General’s satisfaction, within 21 days of the end of each quarter and commencing with the quarter ending December 2000, that it had complied with the Oftel formula. Further details of the formula, the assumptions on which it is based, and the alternative churn rates used, are contained in the Explanatory Memorandum and Annexes accompanying the Draft Direction.

Subsequent Developments

1.7 Since the Draft Direction was published for consultation, the Director has received additional data from BT Cellnet. This additional data comprised audited Oftel formula returns for the periods of January - March 2000 and for the subsequent quarter, April - June 2000, split between pre- and post-pay. This data –which in the case of that for January-March contained some revisions to correct certain inaccuracies in that originally submitted - shows that for both the last two quarters, BT Cellnet’s operations complied with the Oftel formula. Overall, the operations show an average return of between £3 and £10 per subscriber for the period January to March 2000, and of between -£[…] and + £[…] ( figures removed for publication) for the period April to June 2000. Both these ranges being predominantly positive, it appears that BT Cellnet’s operations do not involve an unfair cross-subsidy. However, Oftel has noted the variance in results between the quarters and will take account of this in its decisions on future monitoring.

Conclusion

1.8 Accordingly, in the light of this information it does not appear to the Director General that BT Cellnet is unfairly cross-subsidising contrary to the requirements of Condition 66. Consequently, the Director General will not confirm the Draft Direction.

Next Steps

1.9 Oftel has noted the concerns raised by the various respondents to the consultation. Oftel intends to continue, at least for the time being, to monitor the compliance with the cross-subsidy condition of BT Cellnet and Vodafone, by means of the Oftel formula, and to take appropriate action in respect of any breaches of the condition that may be identified. As also discussed below it will, however, in future assess the results of the quarterly returns provided by BT Cellnet on a 6-monthly basis.

1.10 Later next year, Oftel will also re-examine its approach to the prevention of unfair cross-subsidy in the light of the outcome of the 2000-01 Mobile Market Review, which is due to be completed in June 2001.

1.11 Among the matters to be considered by the Review which may be particularly relevant to this issue are market definition (eg, whether pre-pay and post-pay are in the same product market for the purposes of competition analysis); the impact of the regulation that supports a wholesale market; and the extent of any market power among operators. Oftel will keep its approach to monitoring in this area under review in the light of the outcome of the Mobile Market Review and the results of monitoring the formula returns in the intervening period.

ANNEX II

Oftel’s Comments on specific issues raised during the consultation

2.1 Oftel received five submissions in response to the consultation, including one from a trade association group representing the interests of ISPs. They covered a number of issues: these, and Oftel’s comments on them, are outlined below.

The Investigation

2.2 A number of respondents raised concerns at the length of time taken by the investigation.

2.3 Oftel accepts that the investigation took substantially longer than the Office’s normal target timescale for such action. The delay was due to the inherent complexity of the investigation, deficiencies in the evidence submitted from various sources, and staff turnover. Oftel anticipates that any future investigations in this area would be progressed more swiftly on the basis of lessons learnt from this investigation.

2.4 It was suggested  that operators had not been formally notified by Oftel that the basis of any decision on whether an operator is unfairly cross subsidising its TSPs would be the updated Oftel formula described in Chapter 1.

2.5 Oftel does not accept this suggestion. In the course of the investigation both operators were informed that Oftel was intending to make certain changes to the formula, in particular the pre-pay/post-pay split, and was requesting additional data from BT Cellnet and from Vodafone in order to analyse performance against such a revised formula.

The Formula

2.6 One of the respondents was of the view that it is not appropriate to disaggregate Oftel formula returns for post-pay and pre-pay.

2.7 It is Oftel’s current view that pre-pay and post-pay do not represent separate markets, but merely separate operations within the market for mobile telephony as a whole. On the other hand, in terms of data gathering and analysis it is advisable to treat the two sectors as separate. This is due to the fact that the two sectors are very likely to show different characteristics with respect to the parameters used to calculate the formula. These differences must be taken into account when calculating the net present value of revenues, since failure to do so would compromise the reliability of the results. Thus Oftel came to the conclusion that it is best to study pre-pay and post-pay separately during the investigation, while using a consolidated result for the purpose of the Oftel formula test. From this perspective, the analysis of each sector is useful insofar as it allows Oftel to reach its decisions on the market as a whole.

2.8 Consolidation of the two partial results for pre-pay and post-pay must take into account the different weight of the two sectors in terms of the economic value of new subscribers. This cannot be achieved by merely adding up the different variables used for the calculation of the post-pay and pre-pay formulae, as one of the respondents was proposing. There are substantial differences in airtime profits and acquisition costs between the sectors, which are not reflected in the size of subscriber populations. Consequently, consolidating by merely adding up variables may result in a mathematical anomaly - in that in some circumstances the Oftel formula result for the whole market may lie outside the range of the results for the two sectors. In other words, aggregating the two sectors by adding up the simple figures from each sector may produce the result that although neither of the two sectors would have passed the test on its own, the consolidated variables would give rise to a positive value. This would be misleading.

2.9 Oftel considers that a better way to consolidate the results is by a weighted average, where the weights used are the number of subscribers acquired during the quarter: this number provides a better measure of the relative growth of the two sectors. We expect firms to base their decisions about cross-subsidising their TSPs upon the relative growth of the specific sector, on the hypothesis that companies will be more interested in subsidising the highest-growth sector, so enabling them to gain a greater market share of the highest-growth sector and greater expected profits. By using gross acquisition figures as weights in the weighted average, a greater economic value is attributed to the higher-growing sector. The consolidated result will therefore reflect the relative economic value of each sector.

2.10 Some respondents expressed concern that Oftel only used one quarter’s data upon which to decide whether to take any enforcement action. Several respondents pointed out that the choice of using quarterly data might lead to misleading results, in particular because of seasonality effects.

2.11 It is true that there could be variations in the volume of business that depend on seasonality. However, Oftel’s formula works at the individual subscriber level, and calculates returns in terms of profit or loss per subscriber. This measure is less likely to be seasonal than revenues or costs. Oftel believes that it would be inadvisable to average figures over a whole year, because of the delay that would be caused and, more importantly, because the level of aggregation of data would be too high. When trying to correct for seasonality effects, it is preferable to use more disaggregated rather than more aggregated data (eg, monthly, rather than yearly, returns), and then apply models that are especially designed to take account of seasonality. Oftel has taken the view that quarterly observations strike the best balance between, on the one hand, collecting information which can be analysed and if necessary acted upon reasonably speedily and, on the other hand, imposing an unreasonable burden on the companies concerned.

2.12 As regards the interpretation of the data, however, Oftel also recognises, that a single quarter’s data may not always provide an accurate indication of a company’s performance over a longer period. Whether or not a particular level of cross-subsidy should be considered unfair if it lasts no longer than three months can only be judged case-by-case. Where information for subsequent quarters is provided which shows a different result from the previous one it would be inappropriate for a Draft Direction based on the earlier results to be confirmed, since the breach would no longer be present. Accordingly, and taking into account the views expressed by respondents on this point, Oftel now intends to continue to collect (audited) data on a quarterly basis, but to review it every six months.

2.13 One respondent commented that it was not clear what values of subscription life were appropriate for assessing compliance with the formula.

2.14 Oftel assesses the mobile operators’ returns on the basis of a range of values for subscription life to assess the sensitivity of returns to changes in this variable. The estimation of future subscription life is by nature uncertain and it is important that the formula reflect this fact. Therefore Oftel takes into account possible changes in the outcome under a range of input values when making a determination of cross-subsidy. These input values currently reflect, to the best of Oftel’s knowledge, the main characteristics of the market in a meaningful way.

2.15 A number of respondents argued that the formula should take into account the capital value of the subscriber base, when considering the profitability of the TSPs.

2.16 The Oftel formula estimates revenue from subscribers on a discounted cashflow basis over the estimated life of a subscriber. The expected revenues are not based on the contract period of a new subscriber, but the expected life of the subscription. Therefore the formula includes all the expected revenue from a new subscriber. To add a capital value to this calculation would be double counting the expected revenues from a new subscriber. On the other hand, using the capital value of the subscriber base instead of the expected revenues from the new subscribers does not seem to be correct. Estimating the capital value of the subscriber base is by its very nature highly uncertain and depends heavily on the prevailing market conditions. Estimating the expected revenues from the population of new subscribers, on the other hand, needs the estimation of only two parameters, whose value can be more easily assessed by observation of the characteristics of the market.

2.17 Concern was also raised that the formula did not take into account new revenue possibilities from m-commerce including mobile data, internet and messaging.

2.18 While revenue streams from these products may in some cases arise during an average subscription period, no evidence has been presented that, for the time being at least, they have a substantive impact on the outcome as measured by the formula. Furthermore, Oftel notes that revenues and costs are both subject to substantial variation due to a very large number of factors. Since the precise nature of these factors, and the intensity and direction of future changes in them, are by their very nature impossible to predict, Oftel considers that the approach currently used provides the most reasonable way to assess the existence of cross-subsidies in the mobile telephony market. However, Oftel remains open to suggestions for revising the formula in this respect if evidence emerges of an appropriate way in which m-commerce revenues can be specifically factored into it.

2.19 One respondent argued that the failure to include discounts/bonuses to service providers with a subscriber base of over 50,000 and allowing for a small firm risk premium in the cost of capital was unfair to TSPs.

2.20 In order to ensure that discount schemes do not put TSPs in a favoured position, Oftel takes the view that it is necessary to exclude any discounts and/or bonuses received by any Service Provider with a subscriber base above the threshold of 50,000. The exclusion of discounts for larger Service Providers allows Oftel to assess whether or not it is the existence of the discount itself that allows the TSP to pass the test. Discount schemes naturally tend to favour TSPs, as the average size of ISPs is probably much smaller. Oftel has not seen any conclusive evidence that the average size of ISPs has substantially increased since 1997.

2.21 The cost of capital has been calculated on the basis of listed mobile operators, as data on components of the cost of capital are not publicly available for mobile service providers. The inclusion of a small firm risk premium allows for the fact that UK service providers will be substantially smaller than global mobile operators, for which data are available. Service provision may be a higher risk business than network operation, although the point is difficult to assess due to the limitations of the available data, even for those mobile service providers which are publicly listed. The estimation of the cost of capital on the basis of publicly available data for joint fixed and mobile service providers produces a cost of capital in line with Oftel’s estimation of the cost of capital for mobile service providers. The results of the formula are also considered with a cost of capital excluding the small firm risk premium. In no case has the inclusion of the small firm risk premium had a material impact on the results.

2.22 One respondent suggested that public accounting statements from TSPs be considered by Oftel as evidence of whether or not an unfair cross-subsidy was occurring.

2.23 Oftel considers that public accounting statements provide little worthwhile information to determine whether or not TSPs are being cross-subsidised. Public accounting statements are historical cost records of the revenues and expenditure of the TSPs in a particular accounting period. They will not necessarily provide an accurate view of whether the expected revenue from TSP operations is greater than the relevant costs, which can better be verified by looking at whether the net revenue from an average additional subscriber exceeds the acquisition cost. For example, an increase in the rate of acquisition of subscribers in any accounting period will show immediately as increased costs, whereas the increased revenues accruing from them may not appear fully until later accounting periods. Furthermore, using public accounting information will ignore the calculation of and changes in the churn rate. Oftel considers that the formula provides a more effective and fair means of assessing cross-subsidy between operators and tied service providers.

2.24 It was suggested that Oftel’s use of monthly airtime operating profit, which is based on current figures, takes no account of possible higher future revenues.

2.25 The test used by Oftel makes use of the comparison between acquisition costs and expected revenues on a per-subscriber basis. This reduces to a minimum the degree of uncertainty associated with estimation of future revenues, since the only parameter which embodies information on the future is the average life of a subscription. Oftel considers that this approach provides the best estimate of the future profitability of a subscriber from the company’s perspective. It can be expected that a firm’s decision on how to invest in acquiring new subscribers will depend mainly on the current revenue conditions and on the expected period of retention of that subscriber. Although revenues can be expected to grow over time in absolute value, and although companies are likely to allow for this when planning how much to spend in subscriber acquisition, profits per subscriber are not subject to the same degree of variability. Oftel considers that it would be impossible to identify all the factors that might affect variations in revenues and costs in the future. From this perspective, Oftel is satisfied that the current approach provides the most robust approach to assessing whether cross-subsidies exist.

2.26 It was suggested that Oftel had considerably overestimated the average subscriber life for pre-pay services at between 45-56 months, and considerably underestimated the monthly churn rate at below 2%.

2.27 Oftel has used three main sources for the estimation of the churn rates: data provided by companies in the market, including Vodafone, BT Cellnet, and three Independent Service Providers; data from an independent survey commissioned by Oftel on mobile phone usage in the UK; and the Oftel returns from Vodafone and BT Cellnet. While most of these data provided a fairly clear picture of subscription lives for the post-pay sector, they shed limited light on the length of ‘subscriptions’ in the pre-pay sector. This is entirely due to the relative novelty of the sector, and on the fact that more detailed data are not yet present. However, on the basis of all the information available at this time, Oftel is satisfied that the chosen range of 45-56 months, consistent with churn rates of 1.8% - 2.2% per month, adequately reflects the characteristics of the pre-pay sector. If more detailed data points to a different range in the future, Oftel will review the parameters.

2.28 A respondent proposed that actual rates of decay should be used as part of the Oftel formula rather than churn rates.

2.29 In principle, decay rates and churn rates are equivalent. A decay pattern is a set of rules that describes the process of a population of objects decreasing period after a period. Under some assumptions, the decay rate is a key factor of the decay pattern, as it describes the relationship between the population in any one period and the changed population in the following period. As long as the definition and the actual figures used are realistic, Oftel does not see any meaningful difference between churn rates and decay rates, since the two approaches should lead to similar results.

Other Issues

2.30 Oftel also received the following comments and suggestions:

2.31 Oftel should consider the tied channels as a totality, as they exist, including all costs and revenue streams.

2.32 Data were requested for each of the TSPs because it had been assumed this would have made it more convenient for them to gather it from their accounts, as they are separate companies. Also, since it is common business practice for different TSPs to target different types of customers, data gathered on this basis gives Oftel the opportunity to better understand the characteristics of the market. However, for the purpose of the formula Oftel aggregates all post-pay and pre-pay data in two different categories. The final result of the formula furthermore consolidates these two categories.

2.33 It was suggested that Oftel should not assume that a given cross-subsidy is unfair simply because BT Cellnet has been found to have Market Influence. The mobile market was increasingly competitive, and the existence and growing strength of Orange and One2One was sufficient to prevent any anticompetitive practices.

2.34 The existence and extent of any market power, and the relevant markets in which such power might be exercised, are matters for consideration during the 2000-2001 mobile market review, now under way. More generally, Oftel believes that regulatory intervention should be kept to a minimum, and is justified only when there are substantial reasons to think that without intervening market conditions would be distorted. While cross-subsidisation is not an anticompetitive practice per se, standard economic analysis suggests that it may result in a significant distortion of competition when it is practised by a firm with market power that cross-subsidises a tied firm in a related market.

2.35 If a firm has market power at the Network Operation level, it can leverage this market power into the downstream Service Provision level by cross-subsidising its TSPs. This may result in the profit margin of competitors being squeezed, and their being unable to compete with the TSPs, because they face the same wholesale charges but must compete with the TSPs’ retail charges. Hence they face the risk of being excluded from the market. Condition 66 of BT Cellnet’s licence reflects this concern.

2.36 One respondent was of the view that TSPs benefit from the group marketing and advertising/branding of the network operators. The respondent felt that some sort of charge for this should be reflected in the Oftel formula as at the moment it implicitly acts as a cross subsidy because of the net benefits accruing to TSPs as a result of sharing a brand.

2.37 Oftel has considered this view but does not feel that the Oftel formula should be changed to include a specific charge for corporate advertising and branding. Advertising which promotes the network brand may stimulate increased turnover by encouraging consumers to use mobile telephony services and to join or stay with the network in question. It is accordingly likely to have some benefits for service providers of both types, since all network operators engage in it to some extent and most ISPs supply several brands. The Oftel formula takes into consideration all costs and revenues relating to customer acquisition and the provision of calls. This includes, amongst other things, operating costs relating to marketing and corporate management.

2.38 A respondent suggested that there was hidden cross-subsidy in the supply of phones by the networks to their TSPs, which Oftel has not considered.

2.39 Oftel has in the past received a small number of complaints about the distribution of phone handsets to Service Providers. This is mostly a retail distribution issue, rather than a mobile telephony issue. However, Oftel would be concerned if, as a result of anti-competitive distribution practices of centralised suppliers of handsets tied to the Network Operators, competition in the mobile telephony market was distorted. From Oftel’s analysis, there is no evidence that seems to point to this. All costs related to handset subsidies are incorporated in the subscriber acquisition cost, thus the formula already takes them into account.

2.40 It was suggested that Oftel should publish the data which it has received from the licensees in relation to this investigation.

2.41 The present Statement, and the Explanatory Memorandum on the Draft Direction, include some of this data. Certain other aggregated data has been published previously, in the Oftel consultation document on the 1999 Review of the Mobile Market. The financial data provided by the licensees for the present purpose is, however, commercially sensitive, and in Oftel’s view it is not necessary for that purpose to publish that data in any further detail.

2.42 Oftel should use a Long Run Incremental Cost (LRIC) basis in order to determine cross-subsidy. If revenue exceeded the LRIC, the service would be sustainable in the long run, suggesting that it would not require to be cross-subsidised.

2.43 The figures used by Oftel are likely to approximate reasonably closely to the LRIC of service provision taken as a whole. Compared to a straight LRIC-based approach, the approach used by Oftel offers the considerable advantage that it does not need a complex model to be built, calibrated and used regularly, with the substantial costs involved in gathering very specific data from the operators.

2.44 Oftel’s approach to the issue of cross-subsidy of service providers is not objective as it looks only at the performance of TSPs. Returns should also be requested from the ISPs.

2.45 Oftel does not believe that cross-subsidies are anti-competitive per se; however they are presumed to have potential anti-competitive effects when they are used by a vertically integrated operator with market power in a relevant market. In this context, cross-subsidies can be used to leverage market power from the area where the operator has market power into a related one. Oftel thus monitors the returns of TSPs because a concern exists that such market power could be leveraged from the upstream network operation level to the downstream service provision level. Considering returns from ISPs would not be useful, in that ISPs are not active at the network operation level (by definition); thus they cannot have market power at that level.

2.46 The Efficient Component Pricing Rule (ECPR) would be a more economically efficient and objective test for measuring the true economic sustainability of stand-alone mobile service provision.

2.47 The ECPR has been devised to answer the question of how a vertically integrated monopolist owner of a gateway-type of facility should set the price for access to the facility, if efficiency is to be considered the priority. This rule is typically used in those situations where it is necessary to set a price for the use of an essential facility through which it is possible to provide services in a downstream, related market. However the ECPR by its very construction aims to ensure that any prospective producer, or entrant, of the service in question be allowed to provide the service only if it is at least as efficient as the monopolist in producing it. In other words, the main objective of the ECPR is that production of a service be not diverted to inefficient producers/entrants. The approach used by Oftel is broadly similar, being based on a ‘retail minus’ concept: we expect TSPs to be able to earn a reasonable profit, as given by the retail revenues they earn on the market less the acquisition costs they incur. The Formula ensures that only ISPs which are at least as efficient as TSPs can compete on the market, since they will be able to make the same adequate return if they are equally efficient, and a higher return if they are more efficient. Oftel’s approach seems to be of more immediate applicability, and it focuses more directly on detecting the existence of cross-subsidies, while retaining the more interesting properties of ECPR.