Vodafone, O2, Orange

and
T-Mobile
Reports on references under section 13 of the Telecommunications Act 1984
on the charges made by Vodafone, O2, Orange and T-Mobile
for terminating calls from fixed and mobile networks
Volume 1: Summary and Conclusions

Vodafone, O2, Orange

and T-Mobile
Reports on references under section 13 of the
Telecommunications Act 1984 on the charges made by
Vodafone, O2, Orange and T-Mobile for terminating calls
from fixed and mobile networks

Volume 1: Summary and Conclusions

Presented to the Director General of Telecommunications

December 2002

 Competition Commission 2003

Web site: www.Competition-Commission.org.uk

Members of the Competition Commission as at 31 December 2002

Dr D J Morris (Chairman)
Professor P A Geroski1 (Deputy Chairman)
Mrs D P B Kingsmill CBE (Deputy Chairman)
Professor J Baillie
Mr R D D Bertram
Mrs S E Brown
Mr C Clarke
Dr J Collings
Dr D Coyle
Mr C Darke
Mr L Elks
Dr G G Flower
Mr N Garthwaite1
Mr C Goodall
Professor C Graham
Professor A Gregory
Mrs D Guy1
Mr G H Hadley
Professor A Hamlin
Miss J C Hanratty
Professor J E Haskel1
Mr P F Hazell
Mr C E Henderson CB
Mr R Holroyd
Professor B Lyons
Professor P Klemperer
Dame Barbara Mills QC
Professor P Moizer
Dr E M Monck
Mr R J Munson1
Professor D Parker
Mr A J Pryor CB
Mr R A Rawlinson
Mr T S Richmond MBE
Mr J B K Rickford
Mr E J Seddon
Dame Helena Shovelton DBE
Mr C R Smallwood
Mr J D S Stark
Professor A Steele
Mr P Stoddart
Mr R Turgoose
Professor C Waddams
Mr S Walzer
Mr M R Webster
Professor S Wilks
Mr A M Young
Mr R Foster (Chief Executive and Secretary)

These members formed the Group which was responsible for this report under the chairmanship of Professor
P A Geroski. Mr N Garthwaite was selected by the Chairman of the CC under section 13(10A) of the Telecommunications Act 1984.

iii

Note by the Office of Telecommunications

Certain material has been excluded from this version of the report following a Direction made by the
Secretary of State for Trade and Industry to the Director General of Telecommunications acting in
accordance with section 14(6) of the Telecommunications Act 1984 as affected by the requirement of the
EC Directive on a common framework for general authorisations and individual licences in the field of
telecommunications services in respect of the obligation of professional secrecy and certain business
secrets.

The omissions are indicated by a note in the text or, where space does not permit,
by the symbol .

iv

Volume 1 contents

Page

Part ISummary and Conclusions

Chapter

Summary.....................................................................................................................3

Conclusions.................................................................................................................9

Part I

Summary and Conclusions

Summary

1.1. On 7 January 2002 the Director General of Telecommunications (the DGT) made two
references to us in exercise of his powers under section 13 of the Telecommunications Act
1984 (the Act). One reference concerned the charges made by Vodafone Limited (Vodafone)
and BT Cellnet Limited1 (BT Cellnet), and the other reference the charges made by Orange
Personal Communications Services Limited (Orange) and Mercury Personal Communications
Limited (One2One2), to operators of fixed or mobile public telecommunications systems for the
termination of calls to handsets connected to their respective mobile networks. Under each
reference, we are required to investigate and report on whether these termination charges
would, in the absence of a charge control mechanism on them, be set at levels which operate, or
may be expected to operate, against the public interest, and if so, whether the effects adverse to
the public interest could be remedied or prevented by modifications to the licences of one or
more of the four mobile network operators (MNOs) concerned. The full terms of reference are
set out in Appendix 1.1.
1.2. Following a review of the current control of the call termination charges of O2 (UK)
Limited (O2) and Vodafone, the DGT announced in a press release dated 12 December 2001
that, in his view, mobile termination charges were substantially in excess of costs; and that he
had proposed charge caps of RPI12 each year for four years until 31 March 2006 on the call
termination rates, not only of O2 and Vodafone, whose average charges had been at the maximum level permitted under the cap for the whole period of their existing charge control, but
also of Orange and T-Mobile (UK) Limited (T-Mobile), whose unregulated charges had consistently been above those of Vodafone and O2 over the same period. The four MNOs had
objected to the proposed charge controls, however, and the DGT therefore referred the matter
to us. Both references are in similar terms and we therefore decided to investigate the two
references in parallel, with the same Group of CC members.
1.3. Our investigation concerned the call termination charges which O2, Vodafone, Orange
and T-Mobile levy on each other for terminating calls on their respective networks (off-net
calls) and on fixed network operators (FNOs), of which the largest is BT, for terminating
fixed-to-mobile calls on their networks. Call termination charges are incurred by the MNOs and
FNOs as costs at the wholesale level, and are taken into account in the retail prices which they
set for their own customers. FNOs call termination charges are separately regulated and are
much lower than the termination charges levied by the MNOs. Calls to mobiles from FNOs
accounted for a much larger proportion (70 per cent) of termination minutes than off-net calls
(about 30 per cent) in 2001/02 (mobile-to-mobile, or on-net calls, do not attract termination
charges), and provided nearly all the MNOs net revenue from call termination charges in that
year. This is because the MNOs pay out in termination charges to other MNOs (for off-net
1
In 2001, British Telecommunications plc (BT) demerged its domestic and international wireless subsidiary, BT Wireless, into a
stand-alone company, mmO2. mmO2 included BT Cellnet, BTs UK mobile operator. In May 2002, mmO2 rebranded all its
operations under the O2 banner and BT Cellnet became O2 UK Limited. We refer to this company as O2 throughout this report,
except where the context otherwise requires.
2
In September 2001, One2Ones parent company, Deutsche Telekom AG, announced that One2One would be renamed T-Mobile
UK and the transition took place on 18 April 2002. We refer to this company as T-Mobile throughout this report, except where the
context otherwise requires.

calls) broadly what they earn in termination revenue from other MNOs, and so the net revenue
they receive from other MNOs or the net amount they pay to them is small compared with the
call termination revenue they receive from fixed-to-mobile calls.
1.4. We found that each MNO has a monopoly of call termination on its own network. This
is because there are currently no practical technological means of terminating a call other than
on the network of the MNO to which the called party subscribes and none that seems likely to
become commercially viable in the near future. There are also no ready substitutes for calling a
mobile phone at the retail level, such as calling a fixed line instead.
1.5. We considered whether competitive pressures on the MNOs at the retail level constrained call termination charges in any way, or forced the MNOs to pass on excess termination
revenues to retail customers through lower prices. We concluded that competitive pressures at
the retail level did not constrain termination charges. There is vigorous competition among the
MNOs to attract and sign up subscribers to their networks, for example through the payment of
incentives and discounts to retailers, and handset subsidies to customers, but this is funded by
excess returns from termination charges. We found that this structure of incentives put in place
by the MNOs distorts the volume and direction of traffic on the network, leading to a distorted
pattern of usage by consumers. We found less evidence of effective competition in call origination: thus the MNOs appear to display at least some power to set price structures that suit
them for on-net and off-net calls.
1.6. In the light of arguments put to us by the MNOs, the DGT and others, we concluded
that termination charges should in principle be cost-reflective and that the most appropriate
method for determining the costs of termination was long-run incremental costs (LRIC). Some
costs, however, are fixed and common to outgoing and incoming calls, and we allocated those
costs on the basis that, because call termination charges are ultimately borne by the caller, the
only costs that should be allowed should be those costs that the caller himself causes (which we
term the cost-causation principle). We concluded that the call termination charges of the
MNOs were well in excess of a fair charge, this being based on the LRIC of call termination
(including fixed and common network costs) and a mark-up for relevant non-network costs. We
also concluded that there should be a small mark-up for the network externality, a justified
addition to the fair charge because the caller benefits from having a large, accessible pool of
people to call and be called by, and should make a contribution to the recruitment and retention
of marginal subscribers. The MNOs argued that the costs of call termination should be calculated on a demand-led (Ramsey) basis, but we rejected such an approach, as it would breach
the cost-causation principle. Further, we concluded that the MNOs would, in practice, set
neither the correct structure nor the level of retail prices in accordance with Ramsey principles
if we were to set termination charges at Ramsey levels; hence, implementing Ramsey pricing
would require us to set retail prices, and we do not believe it would be appropriate for us to
attempt to do so; and, in any case, there are formidable problems associated with computing
correct Ramsey prices.
1.7. It was put to us by the MNOs that, because most people now have a mobile phone,
what consumers lose in paying high termination charges they gain on cheap handsets and
competitively priced on-net calls. We rejected those arguments. Some callers to mobiles from
fixed-line telephones or from payphones do not themselves own a mobile phone, and so subsidize mobile customers through high call termination charges, with almost no reciprocal benefit.
Moreover, to the extent that callers with both fixed and mobile phones use their fixed lines to
call mobiles more than they use their mobile phone, they suffer a detriment due to the high
termination charges of the MNOs. The high prices of fixed-to-mobile, and low prices of on-net,
calls also tend to skew usage from the lower-cost (that is, fixed) technology to the higher-cost
(that is, mobile) technology.
1.8. We conclude that the termination charges of O2, Vodafone, Orange and T-Mobile operate against the public interest, with the adverse effects that termination charges in 2002/03 are
4

30 to 40 per cent in excess of our estimation of the fair charge; and that, in the absence of a
charge control on them, the termination charges of O2, Vodafone, Orange and T-Mobile may be
expected to operate against the public interest, with the adverse effect that termination charges
may be expected to be up to double the level of the fair charge by 2005/06, if such charges
were retained at their current levels in real terms. As a result, the following detriments occur
and may be expected over at least the next three years to occur:
(a) the costs incurred by the FNOs and the MNOs through paying mobile call termination
charges that are in excess of the fair charge are wholly or mainly passed through by
them into their customer tariffs, with the result that consumers pay too much for fixedto-mobile and off-net calls;
(b) the high price of fixed-to-mobile calls discourages such calls and, as a result, too few
such calls are made, thereby distorting patterns of telephone use;
(c) consumers who make more fixed-to-mobile or off-net calls than on-net calls unfairly
subsidize those who mainly receive calls on their mobile phones or who mainly make
on-net calls, or who make little use of their mobile phones;
(d) the excess charges for termination have the further effect that they serve to encourage or
facilitate significant distortions in competition because MNOs are not obliged to charge
and subscribers are not obliged to pay the economic cost of handsets. This leads to the
undervaluation of mobile phone handsets by the MNOs customers combined with a
greater turnover (churn) than would take place if customers paid charges which
reflected the proper valuation of such handsets. This leads to yet further distortions in
greater expenditure in mobile customer acquisition than would have taken place if termination charges reflected costs more closely and if handset costs reflected the costs of
handsets more closely; and
(e) the higher prices of calls from fixed to mobile phones and the lower price of on-net
mobile calls encourage greater use of the higher-cost (mobile) technology at the expense
of the lower-cost (fixed) alternative.
1.9. We did not find that there were, or that there might be expected to be, any offsetting
public interest benefits arising from the termination charges set by the four MNOs being in
excess of the fair charge, which could justify their continuation. We considered a number of
possible remedies to address the adverse effects, and concluded that a charge control by way of
a price cap, based on the relevant LRIC of the call termination service of an MNO with a 20 per
cent market share, plus the appropriate mark-ups as set out in paragraph 1.6, was the only
remedy likely to address them effectively. The price cap can be expressed as an RPIX formula
applied to the weighted average termination charge of the MNOs and would take the form of a
15 per cent reduction in their termination charges over the period from 1 April to 25 July 2003
and then a progressive reduction to the fair charge by 31 March 2006. This proposal, which is
proportionate to the detriments and non-discriminatory, would remedy the adverse effects we
have identified by removing their root cause.
1.10. In our view, this approach is both right and fair. It addresses the adverse effects
through a methodology that is equitable, because it attributes costs on the basis of who causes
them. If termination charges are regulated on this basis, it should mean that:
(a) consumers do not pay too much for fixed-to-mobile or off-net calls;
(b) consumers who make more fixed-to-mobile or off-net calls than on-net calls, or who
make more off-net calls than they receive, will not unfairly subsidize other consumers;
5

(c) cost-reflective call charges should minimize distortion in the volumes and patterns of
calling;
(d) there should be no displacement from less resource-intensive to more resource-intensive
technology; and
(e) there will be less incentive for the MNOs to subsidize handset acquisition, which should
reduce the rate of replacement of handsets.
1.11. The DGTs proposed licence modification sought to regulate termination charges
within the period to 31 March 2006. However, the falling away of the current framework of
telecommunications regulation as the result of the coming into effect of new telecommunications Directives on 25 July 2003 means that any licence modification we recommend will
have a very short life. In our view, a period of three years is more suitable for regulatory
assessment than the shorter period required by the change of regime. Consequently, in carrying
out our inquiry, we used the longer period for the purposes of our analysis of the mobile market
and the likely developments in that market. Moreover, the DGT asked us to state our views on
the level at which termination charges should be set for periods beyond July 2003. In our view,
the analysis we have carried out enables us to take a view of the levels at which charges should
be set for the period to March 2006 and, notwithstanding that such views can have little more
than persuasive effect, we have decided to state our views on the level at which we think call
termination charges should be set to 31 March 2006.
1.12. Accordingly, we recommend that:
(a) For each of O2, Vodafone, Orange and T-Mobile, there should be two price caps, set at
the same level, one regulating termination charges for fixed-to-mobile calls and the
other, termination charges for off-net calls. In this way, the MNOs cannot load charges
disproportionately on to one or other call type.
(b) O2, Vodafone, Orange and T-Mobile should each be required to reduce the level of its
average termination charge by 15 per cent in real terms before 25 July 2003.
(c) O2 and Vodafone should be subject to further reductions in their average termination
charges of RPI15 in each of the periods 25 July 2003 to 31 March 2004; 1 April 2004
to 31 March 2005; and 1 April 2005 to 31 March 2006.
(d) Orange and T-Mobile should be subject to further reductions in their average termination charges of RPI14 in each of the periods 25 July 2003 to 31 March 2004; 1 April
2004 to 31 March 2005 and 1 April 2005 to 31 March 2006.
1.13. The precise effect of our recommendations on the MNOs and consumers is dependent
on the extent to which the MNOs seek to recover the reduction in revenue from capped termination charges by price changes in the retail sector. As the result of our recommendations,
however:
(a) We expect welfare gains of between around 325 million and around 700 million over
the period of our recommended charge control.
(b) We do not expect average retail prices to increase. The MNOs need not increase their
retail prices to restore revenue lost through the capping of termination charges, as their
business plans project a continued downward trend in retail prices and they could recoup
these revenues by reducing the rate of retail price reductions for a period.
(c) As we anticipate no increase in retail prices, we expect no significant loss of mobile
subscribers. Even if handset subsidies are reduced, people already owning mobile
6

phones are unlikely to leave the network unless their handset is lost, stolen or broken.
The MNOs have the option to offer marginal subscribers cheaper packages to induce
them to stay on the network once that happens.
(d) No threat is posed to the financial viability of the MNOs. The MNOs have had notice of
the possibility of charge controls on termination charges since at least September 2001
and their business plans all assume some reduction in termination charges over the next
three years.

Conclusions

Contents
Page
Introduction............................................................................................................................................... 12
Background............................................................................................................................................... 13
Events leading up to the references....................................................................................................... 13
The four MNOs..................................................................................................................................... 14
O2 ...................................................................................................................................................... 14
Vodafone........................................................................................................................................... 15
Orange............................................................................................................................................... 15
T-Mobile ........................................................................................................................................... 15
Summary of the MNOs UK performance........................................................................................ 16
Developments since the MMCs 1998 report ....................................................................................... 17
Features of termination charges ............................................................................................................ 21
General features ................................................................................................................................ 21
Sources of termination revenue for the MNOs ................................................................................. 22
The public interest .................................................................................................................................... 26
Framework ............................................................................................................................................ 26
Jurisdictional and other legal issues...................................................................................................... 26
Summary of views ................................................................................................................................ 27
The DGT ........................................................................................................................................... 28
The MNOs ........................................................................................................................................ 29
O2 .................................................................................................................................................. 29
Vodafone....................................................................................................................................... 30
Orange........................................................................................................................................... 30
T-Mobile ....................................................................................................................................... 31
Third parties...................................................................................................................................... 32
Market definition .................................................................................................................................. 32
Supply-side substitution.................................................................................................................... 33
Demand-side substitution.................................................................................................................. 35
Competitive pressures on termination charges.................................................................................. 35
Closed user groups........................................................................................................................ 37
On-net and off-net calls................................................................................................................. 39
Customer awareness and behaviour .................................................................................................. 41
Sensitivity to the price of incoming calls ...................................................................................... 41
Awareness on part of customers of calling a particular mobile network ...................................... 42
Knowledge of relative and actual prices ....................................................................................... 43
Customer behaviour when calling mobile phones ........................................................................ 44
Conclusion on market definition....................................................................................................... 45
Other market definition considerations ......................................................................................... 45
Competitive pressures at the retail level ............................................................................................... 46
Profitability ....................................................................................................................................... 47
Market shares .................................................................................................................................... 48
Entry ................................................................................................................................................. 49
Access and call origination issues..................................................................................................... 49
Access ............................................................................................................................................... 50
Sales and marketing and customer care ........................................................................................ 50
Customer acquisition and retention............................................................................................... 50
Switching or churn........................................................................................................................ 52
9

Conclusion on access.................................................................................................................. 54
Call origination ................................................................................................................................. 54
Margins ......................................................................................................................................... 54
Transparency of tariffs .................................................................................................................. 55
Conclusion on call origination ...................................................................................................... 56
Conclusion on competitive pressures at the retail level .................................................................... 57
The basis of the MNOs pricing............................................................................................................ 57
Costs of call termination ....................................................................................................................... 58
Introduction....................................................................................................................................... 58
Cost of capital ................................................................................................................................... 58
The risk-free rate........................................................................................................................... 59
Equity risk premium ..................................................................................................................... 60
Equity beta .................................................................................................................................... 61
Debt premium ............................................................................................................................... 61
Gearing and taxation ..................................................................................................................... 62
Alternative to CAPM .................................................................................................................... 62
Overall findings on the cost of capital .......................................................................................... 62
The network cost of call termination ................................................................................................ 63
Oftels LRIC model .......................................................................................................................... 64
The time period ............................................................................................................................. 66
Common network costs................................................................................................................. 66
The increment ............................................................................................................................... 67
The level of efficiency .................................................................................................................. 68
Economic depreciation.................................................................................................................. 70
Concerns over accuracy of the Oftel LRIC model............................................................................ 70
Comparison of outputs of the Oftel LRIC model with MNOs data ................................................. 72
Quantity and value of equipment .................................................................................................. 73
Operating cost ............................................................................................................................... 73
Combined 900/1800 MHz vs 1800 MHz costs ............................................................................. 73
Comparison with FAC estimates for 2001.................................................................................... 75
Cost trends 2002 to 2006 .............................................................................................................. 76
Non-network costs ............................................................................................................................ 77
Externalities ...................................................................................................................................... 81
Introduction................................................................................................................................... 81
The concept of a network externality............................................................................................ 82
Measurement of the network externality....................................................................................... 82
Internalization ............................................................................................................................... 83
Whether the R-G factor changes with penetration levels.............................................................. 84
Funding of the subsidy.................................................................................................................. 84
The effectiveness of a surcharge on call termination in increasing mobile penetration................ 85
Consequences of a reduction in the subsidy.................................................................................. 86
Estimate of the externality surcharge............................................................................................ 87
Conclusion on the externality surcharge ....................................................................................... 89
Costs of call terminationconclusion .............................................................................................. 90
Arguments for and against high termination charges............................................................................ 90
Distributional argument .................................................................................................................... 91
Consumer welfare ............................................................................................................................. 93
Reduction or removal of the subsidy would be detrimental to mobile sector and benefit FNOs ...... 94
Financial position of the MNOs........................................................................................................ 96
Roll-out of 3G ................................................................................................................................... 96
Distortion of consumer choice .......................................................................................................... 97
Conclusion on arguments for and against high termination charges................................................. 98
Termination charges in the absence of a price control.......................................................................... 98
Conclusions on the public interest ...................................................................................................... 102
Remedies................................................................................................................................................. 104
The DGTs proposals for charge controls........................................................................................... 104
The MNOs views on a charge control ........................................................................................... 105
Other possible remedies ...................................................................................................................... 109
Bilateral agreements........................................................................................................................ 109
Yardstick regulation........................................................................................................................ 111
Technological solutions .................................................................................................................. 113

10

Receiving party pays....................................................................................................................... 113

Pass-through and price transparency measures ............................................................................ 114
Non-discrimination ......................................................................................................................... 115
Further alternative remedies suggested by MNOs .......................................................................... 116
Conclusion on other possible remedies........................................................................................... 117
Remedying the adverse effects ........................................................................................................... 117
Our proposals for a charge control.................................................................................................. 117
Alternative approaches to calculating a regulated price control ..................................................... 118
Implementation of charge control ....................................................................................................... 122
Finding on whether the adverse effects identified can be remedied or prevented by a
modification of the MNOs licences .............................................................................................. 122
One or more charge caps................................................................................................................. 125
Other matters relevant to the implementation of a charge control .................................................. 126
Ported numbers ........................................................................................................................... 126
Target average charge ................................................................................................................. 126
2G/3G calls ................................................................................................................................. 127
Impact of price cap.............................................................................................................................. 127
Impact on welfare ........................................................................................................................... 128
Conclusion on welfare ................................................................................................................ 129
Impact on MNOs............................................................................................................................. 129
Impact on consumers ...................................................................................................................... 130
Price impacts for mobile customers ............................................................................................ 130
Volume impacts of regulation..................................................................................................... 131
Conclusion on impact on consumers........................................................................................... 131
Overall conclusion .......................................................................................................................... 131
Proportionality and effectiveness........................................................................................................ 131
Formal recommendations on licence modifications............................................................................ 133

11

Introduction
2.1. On 7 January 2002 the DGT made two references to us in exercise of his powers under
section 13 of the Act. One reference concerned the charges made by O21 and Vodafone, and the
other reference concerned the charges made by Orange and T-Mobile,2 to operators of fixed or
mobile public telecommunications systems for the delivery (or termination) of calls to handsets
connected to their respective mobile networks. Under each reference, we are required to investigate and report on whether these termination charges would, in the absence of a charge control mechanism on them, be set at levels which operated, or might be expected to operate,
against the public interest, and if so whether the effects adverse to the public interest could be
remedied or prevented by modifications to the licences of one or more of the four MNOs. The
terms of each reference are set out in full in Appendix 1.1.
2.2. A termination charge is a wholesale charge made by one telecommunications operator
to another for connecting calls to a phone on its network. Our investigation concerns charges
made by an MNO to other network operators, whether fixed or mobile, for terminating calls
made to its network. An FNO incurs a wholesale termination charge whenever one of its subscribers is connected to a call recipient who subscribes to any of the mobile networks (a fixedto-mobile call). Similarly, an MNO incurs a wholesale termination charge whenever one of its
subscribers is connected to a call recipient who subscribes to a different mobile network (a socalled off-net call) or to a fixed-line network. An MNO does not charge itself a termination
charge for handling calls made between two people both of whom subscribe to its network (a
so-called on-net call). As we are concerned with the level of call termination charges made by
the four MNOs, it is fixed-to-mobile and off-net calls that are therefore the focus of our inquiry.
2.3. Under the Act, the DGT is able to modify provisions of a public telecommunications
operators licence by agreement with that operator or, failing this, following a reference to the
CC that results in an adverse public interest finding and a conclusion that specified licence
modifications would remedy the adverse effects identified. In a press release issued on
12 December 2001, Oftel announced that, following its recent review3 of the charge control on
calls to mobiles, it had concluded that mobile termination charges were substantially in excess
of cost and that there was little incentive for MNOs to reduce them; that it had accordingly proposed charge caps on the termination rates of O2, Orange, T-Mobile and Vodafone of RPI12
per cent each year for four years until 31 March 2006; and that, as the MNOs had objected to
the proposed charge controls, it intended to refer the matter to the CC early in the new year.
The reference was subsequently made and we were given six months from 7 January 2002 to
make our report, with the possibility of a further six-month extension. In May 2002, we sought
and obtained from the DGT a six-month extension of our inquiry until 6 January 2003.
2.4. As both references are couched in similar terms and raise the same issue, namely the
level of termination charges made by the MNOs in the absence of a charge control on them, we
have carried out a single investigation by one group of CC members, and the body of our report
addresses the two references together. However, certain of the appendices form a part only of
the O2, Vodafone, Orange, or T-Mobile report, as the case may be.
2.5. Before addressing the public interest questions set out in each of our terms of reference,
we provide relevant background to our inquiry. We first summarize the events leading up to the
references (see paragraphs 2.6 to 2.11), describe briefly each of the four MNOs involved (see
paragraphs 2.12 to 2.21) and provide a summary of their UK performance (see paragraphs 2.22
and 2.23). We then look at a number of developments in the mobile market over the past four
or five years which are of relevance to our inquiry (see paragraphs 2.24 to 2.42). We end the
1

See footnote to paragraph 1.1.

See footnote to paragraph 1.1.
Review of the charge control on calls to mobiles: A Statement issued by the Director General of Telecommunications on
competition in mobile voice call termination and consultation on proposals for a charge control, Oftel, 26 September 2001.
2
3

12

introductory background section by identifying the distinguishing features of mobile termination charges, the main sources of termination revenue for the MNOs and trends in mobile termination charges since 1998 (see paragraphs 2.43 to 2.60).

Background
Events leading up to the references
2.6. In July 2000, following a period of consultation, Oftel began a review of the extent of
competition in mobile termination, to establish whether competitive pressures were likely to be
sufficient to restrain termination charges to a competitive level after 2002 and, if not, what further controls would be appropriate, and for which MNOs. The outcome of this review was a
statement published on 26 September 2001 (see the footnote to paragraph 2.3). This statement
included Oftels conclusions on the degree of competition in mobile call termination and its
final proposals for future controls of termination charges. It formed part of each of the terms of
reference made to us on 7 January and is set out in full in Appendix 1.1.
2.7. At the time of Oftels review, two of the MNOs, O2 and Vodafone, were already subject to licence conditions designed to control the level of the charges they made to FNOs for
terminating fixed-to-mobile calls. The controls imposed on O2 and Vodafone had implemented
recommendations by the Monopolies and Mergers Commission (MMC) following an investigation1 in 1998 into the termination charges which those two operators levied on FNOs (but not
those levied on other MNOs). The MMC had concluded that the level of their termination
charges was too high in relation to their costs and that this operated against the public interest.
Between August 1998 and March 1999, O2 and Vodafone had had the same weighted average
termination charge of 14.8 pence per minute (ppm). Based on the MMCs recommendations,
Oftel reduced the weighted average termination charges which O2 and Vodafone were permitted to levy on FNOs to a ceiling of 11.7 ppm for 1999/2000, and required the ceiling to be further reduced by RPI9 per cent for each of the subsequent two years. This meant a reduction in
the ceiling for the weighted average charge to 10.86 ppm and 10.2 ppm in 2000/2001 and 2001/
2002 respectively. These controls were due to expire in March 2002 (but see paragraph 2.9).
2.8. In 1998, Orange and T-Mobile were less well established in the mobile market than O2
and Vodafone, and Oftel did not include them in its reference to the MMC in that year. Since
then, both companies have, in Oftels view, become established operators with sizeable market
shares. (Oftel estimated that Vodafone and O2 each had a share of UK mobile subscriber numbers in June 2001 of 25 per cent, that T-Mobile had 22 per cent, and Orange 28 per cent.) These
considerations, together with its conclusion following the review that each MNO had market
power in respect of mobile termination to its own network, led Oftel to conclude that there was
now no justification for setting price caps for O2 and Vodafone alone and not for Orange and TMobile, so far as mobile termination charges were concerned. Accordingly, the proposed
licence modifications for all four MNOs were referred to us in January 2002.
2.9. Before undertaking its review of call termination charges, Oftel had recognized the
possibility that, should it decide that a further period of controls was necessary, there might be
insufficient time to implement them before March 2002 when the current controls on O2 and
Vodafone expired. In September 2001, Oftel therefore modified the licences of O2 and
Vodafone so as to include rollover licence conditions. These conditions extend the current
controls on O2 and Vodafone at their existing level (that is, RPI9) for a further year (that is,
from 1 April 2002 to 31 March 2003). These controls have accordingly been in effect throughout our investigation.
1
Cellnet and Vodafone: Reports on references under section 13 of the Telecommunications Act 1984 on the charges made by
Cellnet and Vodafone for terminating calls from fixed-line networks. MMC 1998.

13

2.10. The termination charges of Orange and T-Mobile have remained formally
unregulated, although in 1999 BT asked Oftel to make a determination as to what they should
be. Following discussions with Orange and T-Mobile, the DGT indicated his preliminary view
that the charges of those companies should be set at the public interest benchmark level (that
is, 12.98 ppm for 1997/98) established by the MMC in 1998 for O2 and Vodafone, adjusted as
necessary for cost variations arising from differences in their respective networks. However,
before the DGT made the determination, BT reached agreement with Orange and T-Mobile
over the level of their termination charges, taking the DGTs view into account. T-Mobile told
us that it had agreed its weighted average termination charge with BT. Orange said that its
annual price review with BT had focused on Oranges time of day charges. The current termination charges of the four MNOs accordingly reflect the exercise of formal regulation in the
case of O2 and Vodafone and informal regulatory pressure in the case of Orange and T-Mobile.
2.11. We have already seen (see paragraph 2.7) that the reference made by the DGT to the
MMC in 1998 concerned the level of termination charges for calls to mobiles from fixed lines
only. That we are now required to investigate termination charges made not only by each of the
four MNOs to FNOs but also by one MNO to another reflects Oftels view that the mobile
sector as a whole (that is, at both the wholesale and the retail level) is not yet effectively
competitive. Oftel set out its position in a separate review of competition in the (retail) mobile
sector,1 which it undertook in parallel with its review of the level of termination charges made
by the MNOs. Oftel found that, at the retail level, off-net calls were on average priced at a
much higher level than on-net calls. Oftel said (in paragraph 2.65 of its competition review)
that the structure of on-net and off-net mobile-to-mobile pricing points to the presence of
strong interdependencies between operators in an oligopolistic sector when setting retail prices
and call termination prices on mobile networks, resulting in an inefficient pricing structure for
mobile-to-mobile calls. Overall, Oftel thought that competition in the retail sector was only
prospectively competitive. It was against this background that we were asked to consider termination charges made by the MNOs in relation not only to calls made from fixed networks to
mobile networks, but also from one mobile network to another.

The four MNOs

O2
2.12. O2 is the UK subsidiary of mmO2 plc (mmO2), which was established in November
2001 following a demerger of the domestic and international wireless businesses of BT. The
group supplies voice and data mobile telecommunications services in the UK, Germany,
Netherlands, Ireland and the Isle of Man. mmO2 also operates a fixed network in the Isle of
Man. Its market capitalization was 4.4 billion at the end of November 2002. Until rebranded
as O2 in May 2002, mmO2s UK mobile subsidiary was called BT Cellnet, and before that, until
May 1999, Cellnet. Cellnet, together with Vodafone, was one of the first companies to be
granted a licence to operate mobile telecommunications networks in the UK, and launched its
network in January 1985.
2.13. The mmO2 group had a total of 17.5 million customers in March 2002; of these,
11.1 million were customers of O2 in the UK. O2s importance to the mmO2 group as a whole is
shown by the fact that in the year to March 2002, O2 contributed 64 per cent (2.8 billion) of
the groups total turnover (4.3 billion), and more than 100 per cent (0.7 billion) of its earnings before interest, tax, depreciation and amortization (EBITDA, a common measure of financial performance).
1
Effective Competition Review: mobile, A Statement issued by the Director General of Telecommunications, Oftel, 26 September
2001.

14

Vodafone
2.14. Vodafone, which operates the UK Vodafone mobile network, is a wholly-owned subsidiary of Vodafone Group Plc (Vodafone Group), a UK listed company. Vodafone Group,
which is the largest MNO in the world, with around 100 million customers, had a market
capitalization of around 83 billion at the end of November 2002, having peaked at some
230 billion around March 2000, when the telecommunications industry was very highly rated
by investors worldwide. The group operates or has interests in a large number of countries
including the UK, Germany, France, the Netherlands, Ireland, Italy, Japan and the USA. It had
13.2 million UK customers in March 2002.
2.15. Vodafone is an important contributor to the Vodafone Groups business. Its
3.6 billion turnover in the year to March 2002 represented 12 per cent of the groups proportionate turnover in that year, while its EBITDA, at 1.3 billion, represented 13 per cent of
the 10 billion EBITDA for the group as a whole.

Orange
2.16. The Orange business began in the UK in 1991 and launched its mobile network in
1994. Floated on the London Stock Exchange in 1996, the company was bought by
Mannesmann, a German industrial conglomerate, in 1999. In 2000, Mannesmann was acquired
by Vodafone Group, which was required under EC merger regulation to divest itself of Orange.
Vodafone Group sold Orange to France Telecom in September 2000. Subsequently, France
Telecom rebranded almost all its mobile activities as Orange and organized them under
Orange SA, a French subsidiary. In February 2001, Orange SA was floated on the Paris and
London stock exchanges, with France Telecom retaining a majority interest currently around
86 per cent. In the UK, Orange plc is the holding company, whose principal operating subsidiaries are Orange, which operates the mobile phone network, and a retail subsidiary operating its network of shops.
2.17. At the end of 2001, Orange SA had 40 million customers worldwide, of whom some
12 million were in the UK. Although Orange was the last mobile operator to join the UK market when it began operations in April 1994, it overtook T-Mobile and O2 during 2001 in terms
of subscriber numbers, which are now similar to those of Vodafone at around 12 million.
2.18. Oranges turnover in the year to December 2001 was 3.4 billion, which represented
36 per cent of the turnover of its parent company, Orange SA, and its EBITDA was
0.8 billion, which represented 40 per cent of that of its parent. Orange SAs market capitalization was around 24 billion at the end of November 2002.

T-Mobile
2.19. T-Mobile was founded in 1993 as One2One Partnership. It was subsequently sold in
1999 by its parent companies, Media One International Incorporated and Cable & Wireless plc,
and later became part of T-Mobile International AG (TMO), owned by Deutsche Telekom AG
(Deutsche Telekom), the established fixed-line operator in Germany. T-Mobile accounts for
25 per cent of TMOs turnover. TMO in turn accounts for about 25 per cent of Deutsche
Telekom. Deutsche Telekom was privatized in 1996, although its largest shareholder (43 per
cent) is still the German Federal Government. Deutsche Telekom has large long- and shortterm debts which it has signalled its intention of reducing to 50 billion euros by 2003. [ Details
omitted. See note on page iv.
]
2.20. In May 2002, Deutsche Telekom rebranded its mobile subsidiaries and One2One
became T-Mobile UK. T-Mobiles operations in the UK are small in relation to the groups
activity as a whole, its turnover for the year to December 2001, at 2.1 billion, being only 7 per
15

cent of Deutsche Telekoms 30 billion total business in terms of assets and turnover. It had
EBITDA in that year of 0.3 billion as against its ultimate parents 11.2 billion.
2.21. By contrast with Orange, which had from the beginning sought to build up its network and customer base across the UK as a whole, T-Mobiles early strategy was to concentrate on developing its customer base within an area bounded by the M25 motorway, only then
gradually extending its coverage across the UK. By the end of 2001, T-Mobile had around
10 million customers in the UK, including those using T-Mobiles network through Virgin.

Summary of the MNOs UK performance

2.22. In Table 2.1, we set out the recent performance of each of the four MNOs UK businesses in respect of a number of key performance indicators. (The data in this table is for the
year to March 2002 in the case of Vodafone and O2 and for the year ending December 2001 in
the case of Orange and T-Mobile.)
TABLE 2.1 Comparison of the UK business activities for the four MNOs*

Vodafone

T-Mobile Vodafone

Share of total (%)

O2
Orange

O2

Orange

3,596
1,252

2,750
670

3,397
771

2,062
345

30
41

23
22

29
25

17
11

751

202

344

261

73

19

33

25

Outgoing call minutes (bn)

including on-net calls

13.7

8.5

12.9

8.7

31

19

30

20

Incoming call minutes (bn)

excluding on-net calls

6.8

4.9

6.3

4.7

30

22

28

21

Customers (m)
Contract
Prepay

13.2
5.0
8.2

11.1
3.5
7.5

12.4
3.8
8.6

10.4
1.8
8.6

28
35
25

24
25
23

26
27
26

22
13
26

Customer type
Contract (%)
Prepay (%)

38
62

32
68

31
69

18
82

ARPU per year

per month

276
23

231
19

246
21

202
17

Average customer numbers

(m)

12.8

10.8

11.1

9.4

1,070

787

1,161

925

531

454

568

500

Turnover (m)
EBITDA (m)
Profit/loss before interest
and tax (m)

Average outgoing call

minutes per year per
customer#
Average incoming call
minutes per year per
customer#

T-Mobile

Source: CC based on information from the MNOs.

*Vodafone and O2 data is for the year to March 2002, whereas Orange and T-Mobile data is for the year ended
December 2001. The figures above are based on the statutory accounts of main UK entity for each of the MNOs.
Figures for customers and minutes for T-Mobile in this table include figures for Virgin and Wholesale Partners.
Virgin Mobile Telecoms Ltd (Virgin Mobile) had 1.4 million customers at 31 December 2001.
Vodafone figures given are for Vodafone Ltd, which in 2001/02 comprised the vast majority of the UK activity of the
Vodafone Group.
Earnings before interest, tax, depreciation and amortization.
The figures given for each MNOs share of total incoming minutes do not mean that we see all incoming minutes as
constituting a single market.
ARPU is the annual average revenue per user calculated as revenue from sales to customers (which is not total
revenue which includes for example equipment sales) divided by the average number of customers.
#Average outgoing and incoming minutes per customer are based on the average number of customers.

16

2.23. Table 2.1 shows that Vodafone has the highest average revenue per user (ARPU, a
common measure used in the industry to denote the average revenue derived from each customer); and that Vodafone and Orange have the largest volume of outgoing call minutes (the
measure of call volume used in the industry); O2 and T-Mobile are shown to have the lowest.
About 60 per cent of Vodafones customers own their mobile phone on a prepay or pay-asyou-go basis (where the mobile user pays for call minutes in advance through the purchase of
a voucher of a set amount) and about 40 per cent on a post-pay or contractual basis (where the
customer enters into contractual arrangements with the MNO to pay periodic, usually monthly,
subscriptions); the corresponding prepay and post-pay percentages both for Orange and O2 are
approximately 70 and 30 per cent. T-Mobile, which in 2001 was the only unprofitable UK
operator as measured by profit before interest and tax, has the smallest customer base of the
four MNOs. Its mix of 80 per cent prepay and 20 per cent contract customers yields the lowest
ARPU of the four MNOs. Post-pay or contract customers (for example, businesses) are on
average more valuable to the MNOs (that is, they generate higher ARPU) than prepay customers. T-Mobile told us that it did not have a large share of business customers, another factor
accounting for its relatively low ARPU. The four MNOs and their businesses are described
more fully in Chapter 5.

Developments since the MMCs 1998 report

2.24. There has been a very marked growth in the number of mobile users over the past
five years, as shown by Figure 2.1. Mobile subscriber numbers grew from just over 7 million in
March 1997 to just under 45 million in December 2001. The phase of most rapid growth in the
market started in 1998 and was attributable in part to subscribers who were for the first time
able to buy a mobile phone under prepay arrangements (see paragraph 2.23). Prepay subscriber
numbers rose from half a million in March 1998 to 31 million by the end of 2001; by contrast,
at the end of 2001 there were approaching 15 million post-pay subscribers.
FIGURE 2.1

Growth in the number of mobile subscribers and fixed lines,

1997 to 2001*
50
40
Millions

Postpay
30

Prepay

20

All

10

Fixed lines

0
1997M

1998M

1999M

2000M

2001M

2001D

Source: Oftel.
*M is year-ending March and D is year-ending December.

Note: Number of mobile subscribers for 2001D is not comparable with numbers for earlier periods
due to O2 and Vodafone revising the basis on which they calculate the number of their subscribers.

2.25. Over the same period there has been a corresponding rise in the number of call
minutes from mobile phones, from 14 billion in 1998/99 to over 46 billion in 2001/02. However, notwithstanding the more marked growth of prepay subscriber numbers, it was post-pay
customers who accounted for most of these additional minutes. Their 33 billion call minutes in
2001/02 (or an average of around 190 minutes a month) compared with 14 billion minutes (or
an average of about 35 minutes a month) for prepay subscribers.
17

2.26. There are a number of reasons for this discrepancy between prepay and post-pay
usage. Post-pay customers typically include those expecting to make a large number of calls,
such as businesses. It is in such customers interests to make periodic subscriptions and obtain
the resultant cheaper call charges. By contrast, prepay users are more likely to be individual
rather than corporate customers and to include people who do not expect to use their mobile
phone often enough to make a subscription worthwhile. Some of these users may also be less
affluent than post-pay customers (especially as no credit check is required when a mobile
phone is purchased on a prepay basis). Prepay call charges being typically more expensive than
contract minutes, the relatively high price may discourage long phone calls by prepay customers.
2.27. Outgoing minutes from both fixed and mobile phones rose markedly between 1996/
97 and 2001/02 (see Table 6.7). Total call minutes from mobile phones grew by an annual average rate of 47 per cent, while total call minutes from fixed to mobile phones grew by an annual
average rate of 37 per cent. It appears that there has been no marked slowing of the growth in
the number of call minutes from fixed lines either to other fixed lines or to mobiles in the face
of the strong growth in the numbers of call minutes from mobile phones, although the number
of minutes from fixed lines did fall slightly between 2000/01 and 2001/02.
2.28. Market research carried out for Oftel over recent years shows that the percentage of
adults (in this context, persons over 15 years of age) who have a mobile phone has risen steeply
since the middle of 2000, when 49 per cent of adults in the UK had both a fixed and a mobile
phone. Oftel told us that, on the same sample basis, penetration had risen to 76 per cent by
August 2002, although on Oftels revised sample basis, the percentage had fallen to 73 per cent
in May and August 2002 (see Table 6.1). The period of rapid growth in the current mobile
technology appears to be over and it seems likely that there will be at best only modest further
growth in customer numbers.
2.29. With such a high level of mobile penetration, the MNOs can no longer rely on adding
new subscribers to their networks to increase their revenues to the extent they did previously.
They have accordingly directed their marketing efforts over the past year or so primarily to the
retention of existing retail customers and the encouragement of as much mobile phone usage as
possible by those customers. A recent development has been the offer in their tariff packages of
inclusive minutes for calling any network at any time of day (instead of merely calling on-net).
This development is a continuation of the move to widen the coverage of inclusive minutes, for
example, to include peak-time calls. Very few tariff packages now restrict inclusive minutes to
off-peak usage. The MNOs told us that, over the past couple of years, they had sought to
increase revenue by reducing or removing entirely the incentives they had previously offered to
acquire prepay customers (for example, the subsidy on handsets). As we have seen, prepay customers on average use their mobile phones less than contract customers and individually
generate lower ARPU for the MNOs. The MNOs told us that, in line with these developments,
they now primarily monitor their performance in terms of ARPU.
2.30. In recent years the short messaging service (SMS) (also known as text messaging),
which allows mobile phone users to send text messages of up to 160 characters, has provided
the MNOs with an additional, and growing, revenue stream. In 1999/2000, the first year in
which Oftel collected data on the use of text messages, there were around 2 billion such
messages; by 2001/02, this number had risen to over 13 billion.
2.31. The MNOs to a greater or lesser extent use indirect retail sales channels as means of
increasing the numbers of customers on their networks. Service providers carry on business
under their own name and act as principals, rather than as agents for the MNOs. Thus a customer has a direct contractual relationship with a service provider, who bills the customer for
his or her mobile service and handles all customer service issues. O2 and Vodafone were originally prohibited from selling mobile airtime to the public, because the Government wanted to
18

stimulate competition and growth at retail level, given the duopoly at the network level. This
prohibition did not apply to the new entrants, Orange and T-Mobile, who have always for the
most part sold direct to customers. Oftel told us that the number of service providers had
declined since March 1998, when they accounted for 26 per cent of subscribers. In September
2001, service providers accounted for 3.5 million subscribers or 8 per cent of all subscribers.
2.32. A more recent development in the UK mobile market has been the emergence of the
mobile virtual network operator (MVNO). MVNOs can take a variety of forms, the most complete involving the ownership and operation by the MVNO of a whole network, excluding only
the radio spectrum and base stations. However, in the UK the more typical form is that in which
the MVNO operates a mobile service by purchasing wholesale bulk call minutes from an MNO
and selling them on at a margin by setting their own retail tariffs. The MNO usually provides
the customer support and the billing. Most MVNOs rebrand the MNOs service, and their subscribers may not know which MNO acts as the host network. Some ten MVNOs are currently
operating in the UK, the largest at present being Virgin Mobile, which is a 50:50 joint venture
between Virgin and T-Mobile, whose network it uses. Virgin had 1.4 million subscribers by the
end of 2001. Other MVNOs include Fresh, operated by the mobile retailer Carphone
Warehouse Limited (Carphone Warehouse) on T-Mobiles network, and Energis Mobile, which
provides mobile services to companies and operates on Oranges network. MVNOs currently
operating or due to be launched in the near future in the UK are shown in Table 3.4.
2.33. At present, MVNOs cannot set termination charges for incoming calls at rates different from those of their host MNO. To do this, they would need to be assigned their own number
range by Oftel. The charges made to those calling the MVNOs customers could then be billed
at a different rate from the charges made to those calling the customers of the operator on
whose network the MVNOs calls were being carried.
2.34. Publication of the MMCs 1998 report closely preceded the introduction of mobile
number portability (MNP) in the UK. Historically, subscribers had to change their telephone
number when transferring their custom from one network to another. The DGT considered this
to be a major obstacle to competition both in fixed and mobile telephony. Oftel therefore introduced new requirements for MNOs on 1 January 1999, which required them to permit subscribers to retain their telephone number if they changed networks. The onus was put on the
original network (that is, the network with which the mobile users number was originally
registered) to act as an intermediary and route the call on to the recipient network (that is, the
network with which the mobile user was subsequently registered). Number porting is not yet
widely used; Oftels August 2002 survey of residential mobile phone customers found that only
18 per cent of those who had ever changed networks or service providers had ported their
number.
2.35. Transferring a mobile phone number from one MNO to another has implications for
the call termination charges of both MNOs involved. If a subscriber transfers their custom from
network A to network B, and ports their phone number, network B receives its standard termination charge only for calls originated on network A. For calls from all other networks, network
B receives the termination charge for calls to network A, minus a small transit fee which is
retained by network A.
2.36. A number of major technological developments occurring over the past five years
have affected the operation of the mobile networks. The four MNOs with which we are concerned in this inquiry have, since the early 1990s, rolled out their mobile networks using digital
technology known as Global System for Mobile communications (GSM). Whilst using the
same technology, O2 and Vodafone have from the outset used frequencies around 900 MHz,
although they were subsequently allocated additional spectrum around 1800 MHz. Orange and
T-Mobile, who acquired their licences later, when no more spectrum was available at 900 MHz,
19

use frequencies around 1800 MHz. However, most mobile handsets today are capable of operating on both 900 MHz and 1800 MHz and are known as dual-band handsets. The GSM technology is usually referred to as second generation or 2G technology, first generation being the
analogue service which preceded the current systems.1
2.37. More recently, GSM networks have been modified to enable them to transfer packetswitched data and thus deliver Internet-type services in a much more effective and efficient
way than using circuit-switched connections, as offered by the original GSM specifications.
This packet-switched technology is known as General Packet Radio Service (GPRS) or 2.5G,
the latter in acknowledgement of its functionality which is intermediate between the second and
third generation technologies. As 2.5G users are always connected to the network whilst in
radio coverage, information can be sent to or from their handsets continuously. All four MNOs
in the UK now offer GPRS services.
2.38. The rapid uptake of mobile services and the perceived demand for faster data services have led to the development of a third generation (3G) of mobile service, the European
version of which is known as Universal Mobile Telecommunications System or UMTS. This
service has not yet been launched commercially in the UK. The main advantage of UMTS is
expected to lie in its capacity to offer much faster data transmission speeds than GSM or GPRS,
allowing data facilities such as multimedia services and access to the Internet. UMTS also has
the potential to offer other advantages such as wider international roaming, enabling callers to
contact mobile phones in countries where, despite GSMs widespread presence, other mobile
communication technologies are used which are incompatible with GSM handsets (for
example, in Japan and the Americas). When 3G networks begin to be rolled out across the UK,
coverage will probably centre initially around major conurbations. When outside these areas of
coverage, calls will automatically fall back on to existing GSM coverage, using GSM and
GPRS to provide their voice and data connectivity respectively. Indeed, if subscribers are on
the move during a call and move in and out of 3G coverage, the call will be passed back and
forth between the 3G and GSM networks in order to maintain continuity. As 3G coverage
spreads, it will be possible to carry the majority of calls solely on the 3G network. In the UK,
handsets will need to be capable of operating on either technology, and call recipients are
unlikely to know the extent to which their voice calls have been routed over 2G or 3G technology.
2.39. UK licences for UMTS were awarded by Government following an auction during
2000. The Information Memorandum, issued on 1 November 1999, drew the attention of prospective bidders to the regulatory environment for MNOs and, in particular, to the cap imposed
on 2G termination charges following the MMCs 1998 investigation. Of the five licences
issued, one was awarded to each of the incumbent MNOs, while a new MNO entrant,
Hutchison 3G (UK) Ltd (Hutchison 3G), secured the fifth. Very significant sums were paid by
each company for its 3G licence, amounting in total to 22.4 billion. 3G spectrum was made
available to the licensees from 1 January 2002. Hutchison 3G told us that it hoped to launch an
initial service by the end of 2002. The other MNOs said they would not be launching their services until sometime in 2003, although mmO2s subsidiary, Manx Telecom, has launched 3G
services in the Isle of Man.
2.40. In his September 2001 statement following his review of the charge control on calls
to mobiles, the DGT said that he did not intend to bring 3G calls within the ambit of the price
control he was proposing. He gave as his reasons, first, that those services were not yet available; second, that it was uncertain what services would develop; and third, that Oftel did not
seek to regulate new services or technologies in advance of their launch.
1
Public mobile telephony was first introduced to the UK in 1985 with the introduction of analogue networks operated by
Vodafone (then Racal-Vodafone) and O2 (then Cellnet). The analogue networks have now been switched off.

20

2.41. Finally in this section, we note that, during the course of our inquiry, the DGT
reviewed the control on BTs retention, which was due to expire in July 2002. In 1998, the
MMC had carried out an investigation1 of the charges levied by BT for calls made by its subscribers to mobile phones on the O2 and Vodafone networks. The DGT had referred these
charges to the MMC because he considered that the price of fixed-to-mobile calls was too high;
in his view, the retail costs that BT was allocating to fixed-to-mobile calls were excessive. As
the result of its investigation, the MMC recommended that an RPIX price control be imposed
on that part of the retail charge that BT was allowed to keep when calls originating on BTs
network were terminated on the mobile networks of O2 and Vodafone (termed BTs retention). The cap provided that the average net retention should be no higher than 3.4 ppm for the
year 1 April 1999 to 31 March 2000 and that this ceiling should be reduced in each of the
following two years by an amount equal to RPI7 per cent. Moreover, BT was required to
ensure that retentions for calls to Orange and T-Mobile were at a similar level. The target
retention for calls to MNOs from BT for 2000/01 was 3.19 ppm, and for 2001/02, 3.08 ppm.
2.42. The outcome of the DGTs review was the decision that BTs retention on calls to
mobiles should continue to be regulated. However, he determined that, instead of attracting its
own charge control, the retention should be controlled as part of BTs general retail basket of
prices. Since 1 August 2002, BTs retail prices have therefore been subject to a so-called RPI
RPI control (which has the effect of freezing prices in nominal terms) for four years until
31 July 2006. Because of BTs retention, Oftel expects that any reduction in termination
charges by the MNOs will be fully passed through one way or the other into retail prices to the
FNOs customers, although such pass-through would not necessarily be into charges for fixedto-mobile calls specifically.

Features of termination charges

General features
2.43. We end this section and set the scene for our discussion of the public interest issues
by identifying, in turn, the salient features of termination charges levied by the MNOs, the
principal sources of termination revenue for the MNOs and the trend in the levels of these
charges since 1999.
2.44. As we have already noted, mobile termination charges are charges made at the
wholesale level; that is, they are imposed by an MNO and incurred by another MNO or an
FNO. MNOs and FNOs take these charges into account in the retail prices which they set for
their own customers, for example through their call tariffs, monthly subscriptions for post-pay
customers or the price of mobile handsets.
2.45. As MNOs both receive termination charges from other operators when calls are made
to customers on their network (incoming calls), and pay termination charges to other operators
when their customers make calls from their own to other networks (outgoing calls), it is clear
that such charges represent both a cost and a source of revenue for network operators. It is a
feature of termination charges that, if a network operator could unilaterally raise its own
charges, this would not only increase its own revenue but also raise the costs of other network
operators, at least in the short-term. If those other operators were to pass on the increased costs
to their customers in the shape of increased retail prices, they would make it more expensive for
their customers to use their phones to call the network which had increased its termination
1
British Telecommunications plc: a report on a reference under section 13 of the Telecommunications Act 1984 on the charges
made by British Telecommunications plc for calls made from its subscribers to phones connected to the networks of Cellnet and
Vodafone. Oftel, December 1998.

21

charges. At the same time, however, the customers of the operator that had raised its termination charges would not (at any rate by virtue of that action alone) suffer any increase in their
retail prices. Hence, if one MNO unilaterally raised its termination charges, this would, all
other things being equal, improve its competitive position vis--vis its MNO rivals.

2.46. The normal consequence of a companys raising its prices is that its customers consider leaving in search of a better deal elsewhere.1 However, in mobile telephony in the UK, it
is the person who initiates a call to a mobile phone who pays for that call. Oftel has described
this feature as the principle of calling party pays (CPP). (Oftels views on the effects of CPP
are set out more fully in paragraph 2.71.) It is a consequence of the CPP principle that, when an
MNO raises its termination charges, its own customers are not affected. To this extent (that is,
assuming no reciprocal action from other MNOs), there would appear to be an incentive for
MNOs to raise their termination charges and, by the same token, little incentive to lower them,
since lowering them would both reduce their own revenues and lower the costs of their
competitors.

2.47. An MNO facing an increase in termination charges by another MNO faces, therefore,
an increase in its costs. It has, in theory, the following alternative responses. First, it could pass
on the increase to its customers via its retail prices. Second, it might decide to absorb the
increase, thereby suffering a potential fall in profits. Third, it could raise its own termination
charges. Finally, it could refuse to terminate on its network calls made by the customers of the
MNO that had raised its termination charges or, for that matter, refuse to send calls to that
MNO. In practice, both regulatory obligations to connect and commercial necessity would preclude these final options, while the third is only possible if the MNO which is seeking to retaliate for the increase in termination charges is either not regulated or, if it is, has room to raise its
termination charges by the required amount without breaching that regulation. Otherwise, only
the first two options would remain open. Given the characteristics of termination charges that
we have been discussing, there appears to be no commercial incentive for MNOs to lower their
termination charges, either as a first mover tactic or in response to another MNOs doing so.

2.48. There may be an additional consideration for MNOs in deciding whether to raise termination charges. Surveys suggest that many consumers do not know the particular mobile
network they are calling (see paragraph 2.136). This is compounded by MNP; once a number
has been ported, it is no longer possible to identify the network to which it belongs. If an individual MNO raised its termination charges by, say, 10 per cent, consumers who did not know
which network they were calling would be likely to notice only the effect on average retail
charges, which would rise by a smaller percentage amount than the increase in termination
charges. Thus, MNOs may be more inclined to raise their charges (since they individually have
a small effect on the average) than they would if most consumers knew the terminating network
they were calling and understood clearly the full rise in price that they faced.

Sources of termination revenue for the MNOs

2.49. MNOs derive termination charges mainly from two sources: incoming off-net calls
(where the charge is levied on another MNO) and incoming calls from fixed lines (fixed-tomobile calls, where the charge is levied on an FNO). The impact of the level of termination
charges on the MNOs finances is very different as between the two sources, for the following
reasons:
1
This is what might be expected to happen if an MNO raised its termination charges under a system where the call recipient pays
for the callsuch a system exists, for example, in the USA and is termed receiving party pays (RPP).

22

(a) fixed-to-mobile calls account for a much larger proportion of incoming (or termination)

minutes than off-net callsabout 70 per cent compared with about 30 per cent (see
Table 6.8). Thus, a change in termination charges levied on FNOs for fixed-to-mobile
calls has a far greater impact on the MNOs finances than an equivalent change in termination charges levied on other MNOs for off-net calls. This would be the case even if
there were no reciprocity of mobile-to-mobile charges among MNOs; however, the
existence of such reciprocity (see (b) below) together with the effect of the balance of
trade among the MNOs (see (c) below) magnifies the already large difference in
impact;
(b) if termination charges levied by MNOs on FNOs for fixed-to-mobile calls are changed,

there will be no resultant change in termination charges levied by FNOs on MNOs for
mobile-to-fixed calls. This is because FNOs termination charges are separately
regulated (see paragraphs 2.41 and 2.42). By contrast, a change in the mobile-to-mobile
termination charges set by all four MNOs will result in any individual MNOs costsie
the termination charges it paysbeing changed in the same direction (up or down) as its
revenueie the termination charges it levies. This applies whether the change results
from price-regulation (which may be expected to affect all MNOs in a similar way), by
means of bilateral agreements as described in paragraphs 2.473 to 2.479, or by tit-fortat price changes in a completely unregulated environment. As a result, the effect of an
across the board change in mobile-to-mobile termination charges on an MNOs financial
position (and thus potentially on the retail prices it charges to its customers) is only a
fraction of the gross amount of the change. The size of the fraction depends on the
balance of trade, as discussed in sub-paragraph (c) below; and
(c) each MNO sends calls to other MNOs as well as receiving calls from them. We refer to

the comparison of such calls sent and received as the balance of trade (see Table 6.14).
The volumes of calls sent and received are not equal; that is, there is in fact an
imbalance of trade. [
Details omitted. See note on page iv.
]
2.50. For the MNOs taken together (that is, for the mobile industry as a whole), the fraction of a change in termination charges that impacts on the MNOs profits is by definition zero
whenever trade within the MNO sector is exactly balanced. Thus, changes in mobile-to-mobile
termination charges do not affect the aggregate finances of the industry as a whole, and for this
reason may prima facie be expected to have no effect on MNOs retail prices. There is a possible effect on the balance between on-net and off-net call charges (discussed in paragraphs 2.122 to 2.131), although the causal link is not easy to establish and quantify, and we do
not rely on its existence in forming our conclusions. Also, there is an effect on the finances of
individual MNOs because of the imbalance of trade within the group of four.
2.51. If the trade in paragraph 2.49(c) were exactly balanced (and provided the reciprocity
in paragraph 2.49(b) were maintained), then the finances of the individual MNOs would be
unaffected whatever the level of mobile-to-mobile termination charges, except perhaps at
extremes far beyond any level discussed in this report. Accordingly, increases or decreases in
mobile-to-mobile termination charges would necessitate no changes to MNOs retail prices so
as to preserve their financial position. This would be the case whatever the level (either absolute or comparative) of each MNOs mobile-to-mobile business. Even at the level of imbalance
shown in Table 2.2, which shows termination revenue by contributing operator in 2001, the
financial effect would be small.
23

TABLE 2.2 Termination revenue by contributing operator, 2001

million
Vodafone

O2

Orange T-Mobile

Total

Revenue from termination chargesfrom other MNOs

Costs of termination chargesto other MNOs
Net revenue
Revenue from termination chargesfrom FNOs
Paid to FNOs
A Net revenue

Figures omitted.
See note on page iv.

B Overall net revenue from termination charges

Contribution from FNOs (%)
Source: CC calculations on data provided by MNOs.

*Figures for Orange were calculated by the CC on the basis that the proportions of outgoing calls from
the different sources was similar to those of the other three MNOs.
A has been given as a percentage of B.

2.52. [

Details omitted. See note on page iv.

] From this it can be concluded that:

(a) all the incumbent MNOs (and/or their customers) would benefit from higher termination

charges, with no obvious limit to the increase, the benefit being at the cost of the FNOs
(and/or their customers); and
(b)

[
Details omitted. See note on page iv.
]

2.53. In their evidence the MNOs described in detail the effect that a reduction in termination charges would have either on their own finances or on the retail prices they charge their
customers. Insofar as these claims are valid, the above analysis shows that the effect would
derive almost entirely from changes in termination charges to FNOs for fixed-to-mobile calls,
and scarcely at all from changes in termination charges for mobile-to-mobile calls.
2.54. As a caveat to the above analysis, it is not clear that Hutchison 3G, as a new market
entrant providing a 3G service only, will exhibit a call ratio similar to that in paragraph 2.49 (a)
or a balance of trade with the incumbent MNOs similar to paragraph 2.49(c), or indeed that the
reciprocity of termination charges with the incumbent MNOs would necessarily in all circumstances be as described in paragraph 2.49(b).
2.55. The MNOs told us that they did not keep records of the balance of trade referred to
above, that determining accurate figures would be impeded by the fact that many mobile-tomobile calls were routed via BT, and that, because (so far as the MNOs were aware) there was
no systematic reason for the pattern shown in Table 2.2, such a pattern could not be assumed to
continue. We considered whether these comments would cause us to change our conclusion in
paragraph 2.52. However, the first two comments are not relevant to the conclusion, and the
24

third comment reinforces it. If the imbalance of trade is not systematic, then it is likely to continue to be relatively small (albeit perhaps fluctuating); if it fluctuates then, after averaging
pluses and minuses over time, the effect could well be even smaller.
2.56. At first sight, the comments in paragraph 2.51 might appear to indicate that priceregulation of mobile-to-mobile termination charges is inappropriate, even if regulation of fixedto-mobile termination charges is justified. However, as well as the possible distortive effect of
excessive mobile-to-mobile termination charges on the balance between on-net and off-net call
charges, the effect of the imbalance of trade on individual MNOs while relatively small is not
negligible, [
Details omitted. See note on page iv.
]
2.57. Table 2.2 shows that FNOs accounted for virtually all of the MNOs net revenue
from termination charges in 2001. Oftel has calculated that, as the result of these termination
costs being passed through to customers at the retail level, mobile termination charges make up
about two-thirds of the retail price that FNOs charge their customers for calling a mobile phone
and about 40 per cent of the retail price of calling from one mobile network to another.
2.58. Table 2.3 sets out the average termination charges of the four MNOs (in ppm) for the
current price control period. O2 and Vodafone have set their average termination charges at the
maximum level permitted under the charge control in each of the four years 1999/2000 to
2002/03. It follows that the charges shown for O2 and Vodafone in each of those years effectively show the level of the price cap in those years. (O2s and Vodafones actual charges
differed slightly from those in the table, due to mis-estimations during each control period;
these were then corrected in the subsequent year (details are shown in Table 6.12).)
TABLE 2.3 MNOs average termination charges, 1999/00 to 2002/03 (ppm)

Vodafone*
O2*
T-Mobile

Orange

1999/00
11.7
11.7
(  )

2000/01
10.86
10.86
(

2001/02
10.2
10.2


2002/03
9.4
9.4
)

1999
2000
2001
2002
(
Figures omitted. See note on page iv. )

Source: Oftel and CC calculations on data provided by MNOs.

*Charge under the price cap. Actual charges differed slightly due to mis-estimations but these are
corrected in the subsequent year (see Table 6.12).
Average inbound revenue.
Applicable to standard termination rates only.
February to March 2000.
]
[
Details omitted. See note on page iv.
For the period January to June.

2.59. The MNOs set different call termination charges by time of day and day of the week.
Under their current charge control, O2 and Vodafone are permitted to set each of these charges
at whatever level they choose, provided the resulting weighted average charge (the weights
being based on the previous years call minutes) meets the price control. The MNOs termination charges by day of week and time of day, from April 1999 to September 2002, are shown
in Table 6.13. It will be seen that, for all four MNOs, daytime charges are the highest, evening
charges generally rather less and weekend rates the lowest.
2.60. We note that, during our inquiry, O2 and Vodafone chose to reduce their termination
charges (in conformity with the charge controlsee paragraph 2.9) on their evening and weekend rates, as shown in Table 6.13. Orange and T-Mobile have broadly reduced their rates for all
time periods.
25

The public interest

Framework
2.61. As summarized in paragraph 2.1, the first of the two questions put to us by the DGT
is whether the termination charges of O2 and Vodafone, and Orange and T-Mobile, in the
absence of a charge control mechanism on them, would be set at levels which operated, or
might be expected to operate, against the public interest. If we found that in either case the level
of their termination charges would operate, or might be expected to operate, at such levels, we
are required by section 14(1)(b) of the Act to specify in our report the effects adverse to the
public interest. The second question put to us by the DGT is whether any adverse effects so
specified could be remedied or prevented by modifications of the respective licences of O2,
Vodafone, Orange or T-Mobile. If we found that they could, we are required by section
14(1)(c) to specify the modifications by which those effects could be remedied or prevented.
2.62. In carrying out our investigations on these references, we are required to have regard
to the provisions of the Act and in particular to section 3(1), which imposes primary duties on
us to secure that there are provided throughout the UK such telecommunications services as
satisfy all reasonable demands for them and that any person by whom any such services fall to
be provided is able to finance the provision of those services. The Act imposes a number of
other duties on us, including a duty to promote the interests of consumers, purchasers and other
users. The UK and EC legal and regulatory framework relevant to the current references is
described in Chapter 4.
2.63. During our inquiry, we received a substantial volume of written evidence from interested parties, the greater part of it from the five main partiesthat is, Oftel and the four MNOs.
We encouraged the main parties to send each other copies of the principal submissions they had
made to us, edited as necessary to exclude confidential material, and Oftel published many of
its submissions on its web site. We made site visits to the four MNOs and consulted them and
Oftel in writing throughout the inquiry on a range of issues, which were also discussed with
them at hearings and in a succession of staff meetings. In all, four hearings with Oftel and with
each of the four MNOs were held during the investigation. We also held a number of hearings
with third parties (as indicated in Appendix 1.1). On 1 April 2002 we published a letter which
had been sent on 31 March to the main parties setting out a number of public interest issues
(see Appendix 2.1). On 22 July we published a Remedies Statement which set out our thinking
at that time on the issues and on hypothetical remedies (see Appendix 2.2). We commissioned a
market research company, BMRB International Limited (BMRB) to conduct two surveys of
mobile users on our behalf and took the advice of technical consultants and Leading Counsel.
We also sought the views of the European Commission on the draft Market Recommendation
relating to mobile markets.

Jurisdictional and other legal issues

2.64. All parties put to us a number of points relating to the Competition Commissions
(CC) jurisdiction to determine and impose controls on the level of their termination charges, in
the light of existing and new EC legislation, and our duties under the Act. We also received
legal submissions from the DGT.
2.65. On jurisdiction, T-Mobile argued, virtually from the outset of the inquiry, that the
DGTs process in the current inquiry was an unlawful and unauthorized assertion of jurisdiction
and accordingly that we could not act on the references. We took the advice of Leading
Counsel who concluded that the CC was properly seized of the references and acting on this
advice we informed T-Mobile of our decision and proceeded with the reference without interruption.
26

2.66. The regulation of the telecommunications sector is in a state of evolution. A description of the most important aspects of this evolution is found in Chapter 4. However, central to
these developments is that the two key Directives most closely related to the regulation of interconnection, namely the Licensing and Interconnection Directives, have been prospectively
repealed by the Framework Directive.1 The Licensing and Interconnection Directives will cease
to have effect as of 25 July 2003. The transition from one scheme of EC regulation to another,
taken together with the provisions of the Act, have created a complex framework of regulation
within which the current references have to be considered. Part of the complexity of the scheme
of regulation is that it is common ground between the parties that the licences which the DGT
seeks to modify by these references will be abolished by the new Directives before 25 July
2003.
2.67. Arising from this transition, T-Mobile added two further submissions in support of its
view that we should terminate the inquiry. First, it said that Oftel, as a national regulatory
authority, had no power to impose price controls under the current Directives, which would
expire on 25 July 2003 when the new regime took effect, unless they complied with the new
EC regime due to be implemented in July 2003. Second, even if the DGT could impose price
controls, the CC could state no modification to the current licences for the period beyond
25 July 2003. The correct solution, according to T-Mobile, was to terminate the inquiry and
await the adoption of the European Commission recommendation on market definition before
embarking upon a proper market review which would comply with the new regime.
2.68. We arranged for T-Mobiles views on the legal issues to be disclosed to the other
main parties in the inquiry. Orange also questioned whether Oftel would be able to maintain,
after 24 July 2003, any regulatory obligations that might be imposed as a result of our inquiry.
The DGT provided his views.
2.69. We considered very carefully the views of the parties in this connection. The situation is believed to be unique in that the CC can report and state proposed modifications well
before the expiry of the existing regime, and the DGT can exhaust the statutory consultation
process before the expiry of the regime, yet the licence so modified will cease to take effect on
25 July 2003. We took the advice of Leading Counsel who, over the period from September
2002 to the end of the inquiry assessed the legal views of the parties and offered us advice on
the CCs powers. In summary, his advice, which was consistent with the advice offered by the
CCs own Legal Advisers, was that the CC was properly seized of the references and arguments to the contrary were devoid of merit; that the CC was obliged to identify the adverse
effects and, if possible, to quantify them; that it should satisfy itself that such adverse effects
were remediable by a licence condition, and he believed that they were so remediable; and that,
in stating appropriate licence modifications, the CC was acting in conformity with its existing
obligations under the Licensing and Interconnection Directives. Accordingly, we were satisfied
that we could proceed to make our findings and to propose the appropriate licence
modification.

Summary of views
2.70. We now turn to our substantive analysis of the level of termination charges. We
begin by summarizing the general arguments put to us by those parties who submitted evidence
to our inquiry, that is, the main parties and third parties, including the FNOs. The views of the
DGT are summarized in Chapter 10, those of the four MNOs in Chapters 11 to 14 and those of
third parties in Chapter 15.
1

2002/21/EC.

27

The DGT
2.71. As we have already noted, the DGT told us that central to his analysis of the level of
competition in call termination was the principle of CPP; that is, that the person who initiated a
call paid for that call. The DGT said that, for the majority of mobile services, the use and supply of the service (for example, the choice of network or tariff package) was initiated and paid
for by the mobile owner. Call termination was different because this service was initiated by,
and paid for by, the caller to the mobile phone (who might be a fixed-line caller or a caller on a
different mobile network), not the mobile owner. The consequence was, he said, that high termination charges were unlikely to be subject to normal competitive market disciplines and, in
particular, that a network charging high termination charges was unlikely to suffer an erosion in
its subscriber base for that reason. The DGT told us that he rejected the MNOs argument that
the MNOs were selling a bundle of services, including incoming calls, to their customers. This
was because the inbound calls were actually a wholesale service which were then sold on to a
retail customer (by definition, not a customer of the MNO making the termination charge)
making a call to a mobile.
2.72. In the DGTs view, the overall effect of the CPP principle was that, whereas MNOs
had an incentive to keep the price of those services required and paid for directly by their own
customers (that is, access and call origination) at a level to attract and retain them, they had less
incentive to keep their termination charges low. He said that this was because callers could not
take their business elsewhere if dissatisfied; a caller had to use a particular network to reach a
particular number. Moreover, the receiving party also had little incentive to act to reduce
charges which he did not pay. Each MNO had a monopoly over the termination of calls on its
network and this meant that it could set termination charges without significant competitive
pressure.
2.73. The DGT said that he had considered whether factors such as consumers switching
networks, closed user groups (that is, groups of users, such as businesses or families, who care
about the charges paid by other members of the group), call-back, text messaging, the substitution of other kinds of calls, and countervailing power from purchasers of mobile termination
would have the potential to put competitive pressure on termination charges. He told us that he
had found no clear evidence that any of these had forced the MNOs termination charges to a
competitive level, or were likely to do so in the near future. Indeed, the evidence showed that
O2 and Vodafone were charging 10.2 ppm against costs of 5.6 ppm and Orange and T-Mobile
were charging [  ] in relation to costs of 6.5 ppm. The costs were LRICs, plus mark-ups,
and including the cost of capitalwe discuss the issue of how costs should be measured in the
mobile sector later (see paragraphs 2.217 to 2.387). The DGT pointed out that the MNOs were
setting above-cost termination charges even under conditions of actual (in the case of O2 and
Vodafone) and indirect (Orange and T-Mobile) regulation. Indeed, those of O2 and Vodafone
were being set at the maximum level allowed under the price caps, and those of Orange and
T-Mobile above that level, and this had been the case over the four-year period of the price
control. Orange and T-Mobile had consistently set their average charges above the level of the
price cap, the DGT said, even though they were free to levy lower termination charges if they
had wished to do so.
2.74. This was relevant to the question posed in our terms of reference, the DGT said, of
what would happen were there no regulation. In the DGTs view, the MNOs would set termination charges at a profit-maximizing level in the absence of regulation; this would be above the
current regulated rate and could be as high as 20 ppm. Termination charges that were
significantly in excess of costs were not economically efficient, nor did they represent a fair
distribution of costs and benefits as between those who called mobiles and those who owned
them. Furthermore, as the broad mobile sector was not yet effectively competitive, it could not
be assumed that higher profits on incoming calls were competed away and reflected in lower
28

retail prices for outgoing mobile calls, he said. The DGT therefore considered that caps on
charges for the termination of calls to the four MNOs networks were appropriate and justified
in order to protect consumers.
2.75. As we note earlier, in its review of competition in the retail mobile sector (see paragraph 2.11) Oftel said that, for services bought by mobile subscribers, the market was only
prospectively competitive. There were a number of positive indicators that competition was
developing and that competitive pressures would increasingly deliver a good deal for consumers. But in Oftels view there remained some problem areas, notably a lack of consumer
awareness of different prices and tariffs, relatively high prices for some types of call (international roaming and off-net calls) and barriers to consumers switching networks. Oftel also
drew attention to the high level of profitability of one of the MNOs (Vodafone) (see paragraphs
2.157 to 2.162), and the fact that O2 and Vodafone had been able persistently to maintain higher
prices than Orange and T-Mobile (see Table 6.33).

The MNOs
2.76. The four MNOs submitted that it was inappropriate to view call termination as a
separate market, as it was just one of the bundle of interconnected services purchased by customers; that there was a single market for the provision of all mobile services in the UK; that
that market was competitive; and that none of the MNOs had the ability to earn excessive profits from call termination because the competitive pressures they all faced in respect of the
totality of the services they offered competed away any such profits. Other main points made
by each of the MNOs are summarized in the following paragraphs and set out more fully in
Chapters 11 to 14.

O2
2.77. O2 said that it was not useful to dwell on the issue of market definition; a far more
meaningful way of assessing whether price regulation was required in the public interest was
on the basis of an evaluation of the competitive constraints faced by mobile operators. MNOs
competed for customers by offering a bundle of complementary and interdependent services,
including subscriptions and calls, both outgoing and incoming, and SMS. Termination charges
were just one factor in the framing of a competitive offer. Individual service elements, including termination charges, were constrained as the result of demand responses to particular tariff
pricing strategies. In designing the prices for individual elements of the bundle of services it
offered, O2 said that it had regard to the need to balance incoming and outgoing prices, the
demand for subscription and the demand for incoming and outgoing calls. In doing this, it was
seeking an efficient way in which to recover fixed and common costs across a range of services, for which the demand varied. As the majority of costs in providing mobile services were
fixed and common, the challenge for the MNOs was to design tariff packages that would
encourage use of their networks while ensuring, so far as possible, that overall revenues were
sufficient to cover costs.
2.78. Competition for UK mobile subscribers was, O2 told us, vigorous and effective and
had ensured that prices for mobile services in the UK were, in aggregate, among the lowest in
Europe. Any margins above a cost base on incoming calls were competed away on other services so that the overall bundle of services was reasonably priced and delivered only normal
profits. Competition in the broad mobile sector was sufficient already to constrain the level at
which operators could set their termination charges. Consumer satisfaction with mobile phone
services was high, O2 said. The proposed cost-based price regulation proposed by the DGT
would put at risk the consumer benefits already realized and place those to come (especially
29

those resulting from the development of 3G services) in jeopardy. O2 said that the result would
be a fall in penetration levels and this would damage fixed and mobile customers alike, as their
ability to reach others by mobile phone would be diminished.

Vodafone
2.79. Vodafone said that the mobile market in the UK was effectively competitive. This
was evident in a variety of market outcomes, most notably market shares, switching at a level
comparable with other industries, price reductions exceeding cost efficiency gains and massive
investment in existing and new services. Customer satisfaction, as measured by Oftel, was over
96 per cent. Over the past four years, the industry had brought affordable mobile telephony to
30 million new customers.
2.80. Vodafone said that the relevant market was the provision of all mobile services in the
UK. This reflected the fact that, if an MNO were to raise termination charges to levels which
generated revenues in excess of any relevant measure of cost, it would be forced, by competition from other MNOs offering services to subscribers, to dissipate those additional revenues
via lower subscription and outbound call charges. Vodafone acknowledged, however, that an
effectively competitive retail market was not enough, in itself, to ensure that call termination
charges on any given network would be set at efficient levels (that is, levels which would
achieve optimal consumption of each service provided via mobile networks). The mobile market was unusual in that competition to win customers provided greater incentives to reduce
charges to mobile customers than to reduce termination charges. In the absence of any form of
regulation, MNOs might well be expected, it said, to set termination charges above the efficient
level; intervention of some kind (but not necessarily in the form of a charge control) might
therefore be necessary to prevent this from happening. Vodafone argued that, even if call termination charges were set at an inefficiently high level, that did not mean that MNOs would
make excessive profits, because any excess would be fed back into competition to win subscribers. It meant only that the pricing of each separate element of the MNOs services might
lead to overall allocative inefficiency.
2.81. In Vodafones view, the capped level of termination charges under current regulation, at an average 9.3 ppm, was too low. It said that the DGTs proposal that call termination
charges should fall to an average of 6 ppm by 2005/06 meant that prices could fall well below
costs, which would operate against the public interest, leading to distortions in the mobile market. This would bring about inefficiencies and depress incentives to invest in the sector. If the
CC was, however, satisfied that some form of intervention was necessary, such intervention
should be the minimum needed to bring termination charges to efficient levels; and, moreover,
the CC also needed to be satisfied that such intervention would, on balance, lead to a better
attainment of the public interest objectives laid down in the Act than continuation of the status
quo.

Orange
2.82. Orange invited us to consider whether it was appropriate to contemplate the imposition of price controls on an industry that showed clear evidence of being intensely
competitive and no signs (even on Oftels own analysis) of generating excessive profits.
Orange submitted that regulation could not be justified in respect of operators which had never
made an operating profit (T-Mobile) or which had only just done so for the first time (Orange).
It said that even Vodafone, assessed by Oftel as generating excessive profits, had seen the level
of those profits eroded by the forces of competition.
2.83. In Oranges view, competition in the mobile market was about the acquisition and
retention of customers, and encouraging the greatest possible (efficient) usage of the network.
30

Orange looked at customer lifetime value and did not separate out individual service elements
in order to evaluate its success. That competition in the sector was intense was demonstrated by
falling prices, rapid service innovation, tariffs that offered customers increasing value for
money, low barriers to consumer switching between networks, high customer satisfaction and
high consumer penetration.
2.84. Orange accepted that, taking call termination in isolation, substitutes were limited,
given the extra value afforded to the caller by the ability to contact someone personally
wherever they were. There was no reason to think that the MNOs would not want to attract
more and longer calls to their networks and to attract them at times of the day when their networks were not being fully utilized. It was of course the retail prices charged by the originating
MNO which influenced customer behaviour, not the MNOs termination charges. Orange told
us that there had been recent moves by MNOs towards more intense competition for off-net
calls. By bringing off-net calls within a bundle of inclusive minutes offered to mobile subscribers, the price of such calls had been reduced. In due course, Orange said, this should have
the effect of increasing the competitive pressure on MNOs to reduce their call termination
charges.
2.85. So far as fixed-to-mobile calls were concerned, Orange said that, if a greater
competitive focus were to be generated on calls to mobiles from the fixed market, there needed
to be a link of some kind between the retail prices charged by FNOs for originating calls and
MNOs termination charges. This could be encouraged, in Oranges view, through initiatives
such as resale partnerships with FNOs where, for example, an MNO would offer a discounted
call termination charge to an FNO in return for the FNOs reselling the MNOs services to the
FNOs customer base. One way of achieving this was through interconnection incentives. An
MNOs interconnection rates could be set in such a way that the FNO had an incentive to
reduce its retail rates. Another possibility was a reverse friends and family concept. Here, an
Orange customer might nominate five or ten fixed-line numbers. Calls from these fixed lines to
the Orange customers mobile phone would attract a lower termination charge from Orange and
a correspondingly lower retail charge from the FNO. However, Orange thought that a difficulty
with exclusive arrangements of these sorts was that they might infringe non-discrimination
regulations, and BTs systems and ways of working were in Oranges view very inflexible.
Moreover, regulation of BT made it very hard, if not impossible, Orange said, for BT to sign
agreements with other operators offering any form of exclusivity.

T-Mobile
2.86. T-Mobile said that it was plainly not dominant within any economically relevant
product or geographic market. However the market was defined, T-Mobile maintained that it
had not acted against the public interest in the past and that its existing and prospective levels
of call termination charges were not against the public interest. In T-Mobiles view, it should be
able to rely upon the principle of UK and EC law that as a non-dominant firm it could set its
prices as it saw fit in a competitive product market. There was no case for price control through
amendment of its licence, it told us.
2.87. T-Mobile told us that the UK mobile market was currently vigorously competitive
and that competition would intensify in the period ahead. It believed that, to maintain this
dynamism and competitiveness, the MNOs needed the commercial freedom to set the prices for
individual components of the bundle of services which they offered in such a way as to recover
the significant fixed and common costs of providing termination, and should not be subject to
intrusive regulation. In T-Mobiles view, Oftel had wrongly inferred from the CPP principle
that MNOs could set termination charges independently of their competitors. Given the existing
intensity of competition between the MNOs, T-Mobile saw no realistic prospect of any of them
raising the prices of any of their services, including call termination.
31

2.88. The DGTs proposed price control would, in T-Mobiles view, have a significant
impact on the development of competition in the sector and the levels of investment and innovation in the industry, to the longer-term detriment of consumers. It cited a number of such
effects. Thus, falling termination rates were likely to lead to increased subscription or outgoing
call charges and this would affect the most price-sensitive customers, some of whom would no
longer be able to afford a mobile phone. The charge controls could be expected to lead to fewer
calls being made, because they would force a greater proportion of the fixed and common costs
of the network to be recovered from services that were likely to have a higher price-elasticity of
demand than termination. If the MNOs were unable fully to raise subscription and outgoing call
prices to offset the reduction in termination revenues, this would reduce the total revenues of
the networks; given that three out of four of the MNOs were not even earning their cost of
capital, the shareholders of these networks would be forced unreasonably to incur further
losses. The ability of MNOs to compete with FNOs would be weakened, it said, at a time when
BT continued to dominate fixed telephony. The charge controls would reduce the incentive and
ability of firms to invest in existing 2G networks, and in 2.5G and 3G technology.

Third parties
2.89. We also received evidence from a number of FNOs, including BT, bodies representing consumers (which included the statutory telecommunications advisory committees) and
other interested parties. Most of these parties agreed with the DGTs analysis of the problem
and his proposed remedy, although a number of the FNOs were in favour of an immediate
reduction of termination charges to cost, rather than a stepped reduction over four years, as
proposed by Oftel (and referred to by Oftel as the glide path).

Market definition
2.90. In order to address the public interest issues raised in our terms of reference, it is
necessary in the first place to determine what, if any, competitive pressures operate to keep
termination charges at levels consistent with the public interest, in the absence of any charge
control mechanism on them. We have already seen from the brief summary of views of the
MNOs that in their view call termination on each MNO network cannot be regarded as a
separate market. Vodafone said that there was anyway no requirement under the Act for the CC
to decide on a formal market definition. Nevertheless, we consider it helpful, in assisting us to
identify competitive constraints in the mobile sector, to consider whether separate markets exist
for different mobile services and whether any of the MNOs has market power in any such markets. Under the new EC regime, this sort of analysis will be required.
2.91. We begin by identifying three distinct activities or services involved in the take-up
and use of mobile phones: seen from the point of view of the mobile customer, the first is
gaining access to a mobile network and selecting a call tariff, which we term access; the
second is the making of a call, an activity which we term call origination; and the third
enables the mobile customer to receive calls (call termination). The first two services, access
and call origination, are instigated and paid for by the mobile phone customer and occur at the
retail level. The third, call termination, is effected and charged for at the wholesale level by the
network operator of the called party, but is instigated and ultimately paid for by the calling
party, not by the mobile phone customer; it is, however, the mobile phone customer who makes
the choice of network on which the call is to be terminated and hence determines the level of
the termination charge. We discuss access and call origination later (see paragraphs 2.174 and
2.198 respectively) in the context of competition in the retail market. We begin with the subject
of our investigation, termination charges.
2.92. The hypothetical monopolist test, widely used by competition authorities, seeks to
establish whether a hypothetical sole supplier of a product would be able to increase its profits
32

by raising prices by a small amount (usually taken as around 5 to 10 per cent)that is, a small
but significant and non-transitory increase in price (SSNIP)above the competitive level. In
order to determine whether a 5 to 10 per cent increase in termination charges would be sustainable, we consider first the possibilities of supply- and demand-side substitution of mobile voice
call termination by other services at the wholesale and retail levels. In particular, we discuss
whether there would be any alternatives to terminating a call to a mobile phone on the network
of the called party, were an MNO to increase its termination charges by 5 to 10 per cent.
2.93. We pursued the SSNIP test with the parties at hearings with them and carried out a
survey of consumers (see Appendix 6.2) in which we explored the possible supply- and
demand-side substitutes for call termination on the network of the called party.

Supply-side substitution
2.94. We looked first at supply-side substitution. Supply-side substitution occurs when, in
response to a rise in price, other firms switch into the supply of the product whose price has
risen. We considered whether there were any means by which a call to a mobile phone could be
terminated on a network other than the network of the MNO to which the called party was a
subscriber for the purpose of making outbound calls. In practice, this would involve the mobile
owners being able to use a single mobile handset both for making outbound calls on network
A and for receiving inbound calls on network B, which may be another mobile network or
another means of receiving calls. For this to happen, the called party would have to be able to
switch his handset between connections to different networks. Oftel told us that this was
theoretically possible, either through manual or automatic intervention. However, we understand that it is not currently possible for a handset to be logged on to two networks simultaneously.
2.95. In the case of manual intervention, one solution would be for the mobile customer to
obtain a mobile phone with an internal dual subscriber identity module (SIM1) card holder
that would allow the subscriber to take advantage of the network with lower termination prices.
Oftel told us that, if this dual SIM card device was to act as a competitive constraint, mobile
customers would have to use the network with cheaper call termination as the default network, only switching to the other network to make cheaper outbound calls. Oftel told us that it
doubted whether called parties would have the incentive to undertake this procedure or, if they
did, whether they would switch their SIM card to benefit those calling them; it was more likely,
Oftel thought, that those switching their SIM card would do so to take advantage of differentials in outgoing call charges, for example, when roaming overseas. Our own survey evidence
supports Oftels view (see paragraphs 2.133 to 2.135). Moreover, those being called would also
need to be aware of current termination charges on different networks (a matter we examine
latersee paragraphs 2.137 to 2.141).
2.96. So far as automatic intervention to enable the mobile customer to make inbound and
outbound calls on different networks is concerned, Oftel told us that some mechanism would
need to be found to instruct the called partys mobile phone to switch networks automatically.
Such a mechanism did not currently exist and in Oftels view would be unlikely to develop in
the foreseeable future, due to significant technological and coordination difficulties, as well as
the lack of incentives on the part of the called party to make use of a facility to reduce the costs
of incoming calls. Cooperation by the MNOs in allowing access to their SIM cards might be
required, as might the installation of special software on the mobile phone of the called party.
1
A SIM card, which belongs to an individual MNO, is an essential component of a GSM handset and contains details of the
network on which the phone is to operate. The SIM card can also be used to store telephone numbers and text messages, so that
placing the SIM in a different GSM handset would transfer any stored numbers and text messages to the new handset.

33

2.97. We sought the views of the MNOs, Oftel and other parties involved in our inquiry on
whether it was likely that technological developments over the next three or four years would
make termination of a call other than on the network of the called party a commercial possibility.
2.98. Various suggestions were made in this context (see Appendix 3.1) including the
possibilities for either manual or automatic use of multiple SIM cards, just discussed. O2 and TMobile, for example, thought that in the foreseeable future two new technologies in particular
would be important for consumers wishing to communicate with each other. These were
Voice-over-Internet Protocol (VoIP) and instant messaging (IM). O2 said that VoIP would
allow one person to contact another via their email or Internet Protocol address rather than via
their mobile phone number (that is, via packet-switched rather than circuit-switched technology), and this would allow a voice call to take place via the Internet. Charging for these calls
would be dictated by the mechanisms that currently existed for charging for Internet access;
thus, parties would pay for access and there would be no such principle as CPP or the caller
having to bear a termination charge. In O2s view, it was this charging system that made VoIP
such a promising technological and commercial option. Oftel, however, thought that voice over
a packet-switched bearer was unlikely to act as an effective constraint on voice termination
charges because the same MNO would set the termination charge and quality of service for
both types of call, so that the MNO would be able to control whether prices and quality were
set at a level which would make voice over a packet-switched bearer an effective substitute for
voice over the circuit-switched network.
2.99. O2 said that the other important new technology, IM, allowed for the exchange of
voice clips in virtually real time and would be rather like a voice chat line. It was envisaged
that IM would be offered over GPRS (2.5G) networks as well as 3G, O2 told us. T-Mobile said
that other possibilities included wireless local area networks (WLANs; see paragraphs 3.108
and 3.109), which used short-range radio technologies. (Similar possibilities exist for the termination of calls on Bluetooth and other short-range wireless technologies.) However, Oftel
told us that, apart from the significant technical issues to be overcome if any of these technologies were to become a viable option for mobile users, the same lack of competitive pressure
on the inbound call would still apply because of the CPP principle. Vodafone told us that it was
unlikely that over the next five years, technological or other developments would enable voice
calls to an MNOs customers to be terminated otherwise than on that MNOs network. These
various possible technological developments are described more fully in Chapter 3.
2.100. We have no reason to doubt that important technological developments are in prospect, some of them with the potential to blur the distinction between a voice call and a data
message; and some may not involve the levying of a termination charge in its current form.
However, we are required to investigate the public interest effects of an absence of regulation
of termination charges on calls to mobiles from the expiry of the current charge controls in
March 2003. The developments mentioned above, including VoIP and IM, have as yet made
little impact on voice communication in the UK, although one MNO, O2, told us that these
radical new technologies might be expected in the relatively near future. We note that various
technological developments were quoted as likely or possible in the MMCs 1998 report, but
that none of them has turned out to be a constraint on termination charges. It is true that, since
the last report, there have been significant steps forward in many of the technologies that
underlie telecommunications networks. Nonetheless, we believe that there can be an appreciable time lag between the appearance of new technologies and their implementation in new
commercial products, and that the speed with which they are then taken up by consumers is
difficult to predict. Given the comparatively short period for which we are empowered to make
recommendations, we therefore believe that it would be wrong to base our public interest findings on expectations about the pace of introduction, and impact on existing mobile phone use,
of any such new technologies.
34

Demand-side substitution
2.101. On the demand side, Oftel and all the MNOs agreed that substitution at the wholesale level was not currently feasible, since an operator wishing to offer calls to customers
belonging to a particular mobile network must purchase termination from that network in order
to do so. At the retail level, however, several demand-side substitutes were put to us by the
MNOs which, they said, imposed constraints on call termination charges. We note that, if
prices at the retail level are to have a constraining effect on termination charges, then any
changes in the level of termination charges have to be passed through to prices at the retail
level. If the price of terminating calls to a mobile phone were raised, the MNOs said, the calling
party (who paid the charge through his retail price) could switch to calling the person on that
persons fixed line instead, or perhaps to using a text message (SMS). Another possibility
would be to request the called party to call back, if the caller was using a fixed line, as a
mobile-to-fixed call is cheaper than a fixed-to-mobile call. Or the caller could keep his call
short. Call-back and shorter calls are not, however, in our view genuine alternatives to calling a
mobile; rather, they are strategies to reduce the amount paid for calls and create added costs and
burdens (in terms of time and trouble) for both caller and call recipient.
2.102. Calling a fixed-line number instead of a mobile number is clearly a possibility, if
the called party happens to be where the fixed line is located when the call is made and the
caller knows this (or does not mind delaying making the call or the call ending in an answering
service). But fixed and mobile telephony clearly have fundamentally different characteristics
and in our view a call to a fixed line will rarely be a wholly satisfactory substitute for locating
someone on a mobile phone. While a mobile phone is associated with an individual, not a
place, fixed-line telephones are associated with a place rather than an individual. A call to a
fixed line will therefore be an inadequate substitute whenever the called party is away from the
fixed-line telephone or is on the move, which is precisely when the mobile phone is most
valuable.
2.103. Text messaging could be a partial substitute for a call to a mobile. But because it
enables parties to exchange only relatively short messages, and these messages can be delayed,
this form of communication is nothing like a complete substitute for a telephone conversation
in most cases. Oftels research showed that text messaging was regarded, especially by the
young, as an activity separate from voice calling, and that text messages were largely additional
to voice calls rather than a substitute for them. Furthermore, text messaging would not necessarily put pressure on call termination charges, since the operator providing the termination for
the text message would be the same operator as was providing termination of voice calls, and
who could therefore set termination charges for text messaging at such a level as to prevent its
putting pressure on voice termination charges. Evidence from Vodafones survey of mobile
phone users, carried out by NOP, also suggests that most people do not regard text messages as
a satisfactory substitute for voice calls to a mobile phone. Only 30 per cent of those questioned
thought that text messages were a good total substitute for calls to a mobile or a good substitute
for many such calls; 23 per cent thought that they were a poor substitute and 36 per cent
thought that they were a good substitute for only some such calls (see paragraph 6.101).
2.104. For these reasons, we consider that neither calling a fixed line nor sending a text
message is an adequate substitute for reaching the called party on their mobile phone.

Competitive pressures on termination charges

2.105. We consider a number of possible competitive pressures on termination charges in
the following paragraphs (see paragraphs 2.106 to 2.131) before looking at the available
evidence of consumer knowledge and awareness of termination charges (see paragraphs 2.132
to 2.146).
35

2.106. The four MNOs put it to us that the key question was not whether call termination
constituted a separate market on the network of each of the MNOs. They maintained that, even
if there was no alternative to termination on the network of the called party, that did not mean
that the network in question faced no competitive constraints in delivering termination.
2.107. Three of the MNOs, O2, Orange and T-Mobile, submitted that the competitive
pressures they faced prevented them from setting termination charges above a competitive rate.
Vodafone took a different approach. It said that competition from other MNOs indeed prevented it from generating excess profits from call termination charges. But, in a market in
which MNOs competed to sell mobile subscriptions and outgoing calls, each MNO would be
subject to pressures to maintain call termination charges for calls from fixed networks above
efficient levels (by which it meant levels which would achieve what it regarded as optimal consumption of each service provided via mobile networks) with a view to using the excess
revenues earned to fund competition to win subscribers. Vodafone argued that, even if the CC
were to reach the conclusion that there was a separate market for call termination on each
MNOs network, it would be quite wrong to infer that an MNO would be abusing a dominant
position if it were to set its call termination charges above efficient levels. This was because the
MNOs were operating in a market in which they competed among themselves to win subscribers and to sell outgoing calls, and the nature of competition in the market would lead them
to maintain termination charges above an efficient level. Vodafone thought that these features
of the market, and the context in which the MNOs operated, called into serious doubt the utility
and correctness of regarding call termination on each MNOs network as a separate market.
2.108. The MNOs cited a number of retail pressures which they said had the effect of constraining the level of termination charges. We discuss these in the following paragraphs and
then examine whether the evidence suggests that these are indeed effective to constrain the
level of termination charges.
2.109. There are a number of possible ways in which customers of the MNOs might put
pressure on the level of mobile call termination charges. For example, if customers of MNOs
valued receiving calls as part of the bundle of services they received when they purchased a
mobile phone, then their initial choice of network might depend at least partly on the level of
termination charges levied by that network. They might switch networks for the same reason.
However, (on the assumption that the benefit of high termination charges is not retained by the
MNO making the charges, but is passed on as a result of competition to its own customers via
lower prices) they would take either action only if they valued the expenditure that others incur
to call them on their mobile phone above their own expenditure for their own calls. The consistent evidence from surveys is that the reverse is the case (see paragraphs 2.133 to 2.135).
Moreover, such behaviour would presuppose a sufficient degree of awareness of, and concern
about, termination charges on the part of the MNOs customers to induce them to take such
actions. Those customers would need to know, when making an outbound call, whether or not
they were calling a mobile phone (which in practice most of them would know) and if so which
network they were calling, the level of that networks termination charges and the effect on
their own call charges; as well as caring sufficiently about the price of incoming calls.
2.110. We discussed with the industry various mechanisms for increasing pressure on call
termination prices based around optional RPP, whereby the calling or called party could trigger
a mechanism to transfer the cost of the call to the mobile subscriber either at the outset of the
call (for example, by dialling a prefix before dialling the mobile number or based on some preset pattern) or during the call (by the mobile subscriber accepting the charges by pressing keys,
in the same manner as is adopted to put calls on hold). Vodafone told us that any such solution
would require substantial development work for the network, billing system and handsets. BT
told us that no major new technological developments were required to implement this solution,
though there might be some practical issues and cost implications.
36

2.111. At a late stage in the inquiry, O2 suggested a solution along the following lines. The
calling party could call the mobile directly and thus pay the standard termination charge, but he
could also call a national rate (087xxx) number, which would trigger the following options:
If the mobile subscriber had preselected this, the call would be diverted straight to that
subscriber, the subscriber paying for the mobile element of the call (effectively a form
of partial RPP). We note that both Orange and T-Mobile currently offer similar services.
Failing this, the calling party could opt to have a text message sent to the mobile subscriber containing a call-back number. The mobile subscriber could then return the call,
paying the appropriate rate for outgoing calls. We note that this can already be done
between mobile networks or by a fixed network subscriber using the free Internetbased SMS services to request a call. However, at present, for mobile-to-mobile calls
the cost of the SMS may be higher than the savings made, and in any case it may be
more cost-effective to make a short voice call to request call-back.
Alternatively, the calling party could leave a voice-message for the mobile subscriber,
who would then be notified of the awaiting message. Again, we note that there is
nothing to stop this taking place at the moment, if the mobile subscriber opts to divert
all calls to voice-mail.
2.112. Whilst the price for outgoing calls would remain lower than for incoming calls, the
system as described would have the potential to turn incoming traffic into outgoing traffic and,
if the two prices equalized, there would in our view be no incentive to use the alternative
mechanism. We noted that O2s suggested approach was similar in concept to the optional RPP
approach that we had previously discussed with the industry (see Appendix 3.1). The main
objection to optional RPP was that it required action by mobile subscribers, and would be complex to understand, control and implement unless it were introduced for all subscribers. It
would, therefore, seem likely to increase the burden on callers to mobiles. In paragraph 2.504
we consider whether a proposal on these lines could form the basis of a remedy.

Closed user groups

2.113. Oftel told us that it distinguished two types of closed user group. The first was a
narrow group, typically a family in which the mobile owner was also the person who paid for
the calls to their mobile and so had a direct interest in the cost of incoming calls, or a business
which chose a single MNO for all its employees and then paid for all the calls made from the
business premises or from employees mobile phones. In these situations, Oftel said, the calling
party not only paid but was also the customer of the called partys MNO, and a connection was
made between the payment of the termination charge and the mobile owner. The second type of
closed user group was a rather wider circle of people with a mutual financial interest in keeping
call costs down, and largely consisted of loose collections of friends and families. Individuals
might be members of more than one such group.
2.114. Orange and T-Mobile argued that the effect of the CPP principle in keeping charges
high (as seen by Oftel) was undermined by the growth of closed user groups. They argued that
the sensitivity of these groups of mobile phone customers to incoming call charges placed a
competitive constraint on the level of termination charges that the MNOs could impose. Closed
user groups took advantage of the fact that on-net calls were priced much lower than off-net
calls.
2.115. O2 told us that it thought the term closed user group was misleading and preferred
the concept of repeat calling relationships. It said that closed user groups were far less common than repeat calling relationships, which involved calling pairs who called each other regularly, whether or not they fell into the category of closed user group. Such calls occurred both
37

on-net and, very frequently, off-net. It said that analysis of its customer call data showed that a
large proportion of traffic was generated within repeat calling relationships, most people having
such relationships with around five other people. It was easy for a pair of individuals in such a
relationship to exert pressure on charges by deciding jointly to switch call direction on a regular
basis in response to price differences. This ability to switch call direction accordingly constrained the MNOs from introducing any imbalance between incoming and outgoing call
charges, O2 argued. Orange and T-Mobile took the view that closed user groups were widespread and that call recipients were likely to be concerned about the cost of calls to them.
Vodafone, however, told us that in its view closed user groups did not effectively constrain call
termination charges.
2.116. We looked at a number of surveys carried out by the MNOs for evidence of consumer behaviour and attitudes on this matter.
(a) According to Vodafones NOP survey, carried out in March/April 2001, 39 per cent of

respondents told NOP that all the mobile phones for which they or the other members of
their families paid were on the same mobile network. But 56 per cent said that this was
not the case. About half of those who were on the same network said that this was
because it reduced the costs of each calling the others, while 22 per cent said that their
being on the same network was coincidental.
(b) A survey commissioned by O2 in February 2002, also carried out by NOP, asked those

owning mobile phones whether the network used by people they were likely to be
communicating with was an important factor when deciding which network to join. For
around one-third of respondents this was important but for over half it was unimportant.
A high percentage of these groups could not say why this factor was, or was not,
important to them.
(c) Vodafones NOP survey (see (a) above) also asked respondents whether they took into

consideration when choosing their mobile network those networks of which their
friends, family or other regular contacts were customers. 58 per cent did not consider
this matter.
(d) Another survey, carried out by Martin Hamblin GfK (GfK) for O2, exploring why

people changed their mobile network provider, found that very few respondents said
they had changed network so as to be on the same network as family and friends (see
Table 6.44).
(e) A survey for Oftel carried out by Taylor Nelson Solfres plc (TNS) in November 2001

found that of those who had ever changed network provider (29 per cent), price was
given as the reason by most respondents, with wanting to be on the same network as
family and friends being a reason for far fewer respondents.
2.117. This survey evidence (set out in paragraphs 6.210 to 6.217) suggests that the
majority of people do not consider it important for their mobile phone to be connected to the
same network as that of the people they call most often.
2.118. These findings were confirmed in the first of our own surveys (see Appendix 6.2),
which we commissioned from BMRB during the inquiry. 81 per cent of respondents who paid
for a mobile phone told BMRB that they had never chosen, or changed to, a network in order to
be on the same network as someone they spoke to often. Of those who had chosen to do so,
88 per cent said that it was to save money on outbound call charges. This evidence should be
seen in the light of other findings from our survey; these showed that, for a high proportion
(79 per cent) of those who paid for mobile phones, the groups of mobile users that they called
most often on their mobile phone were family, friends or partner and that such calls accounted
for some 60 per cent of all the respondents mobile-to-mobile calls. However, of the 79 per cent
who paid for mobile phones, only 9 per cent of their calls were on-net calls.
38

2.119. While this evidence suggests that a small minority of residential mobile users is
aware of, and interested in, the reduction of call charges among their circle of family or friends,
manifested in behaviour such as the formation of closed user groups (or repeat calling relationships) or tactics such as call-back, we do not consider that it shows a widespread concern on
their part to keep incoming call costs down.
2.120. The position is slightly different for corporate customers. Thus, Oftels research in
the business sector showed that about 19 per cent of small and medium-sized enterprises
(SMEs) had taken steps to reduce the cost for them to call mobile phones through methods such
as private wires and adaptations to convert calls from fixed lines into on-net mobile-to-mobile
calls. However, the research found that most SMEs cared more about other considerations, such
as the prices of outgoing calls and network coverage. Large organizations, which are likely to
be a more price-sensitive segment of users, have sufficient negotiating strength and expertise to
secure favourable terms which help to reduce their costs of using mobile phones.
2.121. Before reaching our conclusions on the extent to which closed user groups or repeat
calling relationships constrain termination charges, we need to examine the purpose and effects
of the large differential between on-net and off-net prices, since the MNOs claim that on-net
calls have been a particular focus of competition and are the means by which frequent callers to
each other on mobile phones can minimize the costs of doing so, thereby putting pressure on
call termination charges.

On-net and off-net calls

2.122. In order to obtain a broad comparison of the prices of on-net and off-net calls, we
calculated the average retail revenue per minute (excluding subscription revenue) received by
the four MNOs in 2000/01 (see Table 6.10). Average revenue was about 25 ppm from off-net
calls and about 6 ppm from on-net calls. We noted a corresponding relationship between the
contribution per minute (after network costs and termination charges paid) from these two call
types (see Table 5.22 and also paragraph 2.200).
2.123. Table 6.5 shows the total number of outgoing call minutes from mobile phones for
different call types. In 2001/02, on-net calls accounted for 30 per cent of all outgoing minutes,
off-net calls for 15 per cent and calls from mobiles to fixed lines for 55 per cent. If we compare
the prices of these types of call with the number of call minutes for each, it is clear that, broadly
speaking, off-net calls account for the lowest proportion of traffic (15 per cent) and have the
highest prices (25 ppm), while mobile-to-fixed calls account for the highest proportion of
traffic (55 per cent) and, at 7 ppm, have much lower prices than off-net calls and broadly
similar prices to on-net calls (6 ppm).
2.124. Vodafone and Orange offered commercial explanations for the price differences.
Vodafone said that, as the mobile market had begun to expand dramatically, the MNOs had
started to focus attention on ways in which they could offer attractive new packages to new
subscribers and help to differentiate their offer from those of their competitors. Vodafone told
us that it had accordingly started to offer a category of relatively lower call charges that could
change the widely held perception that all mobile calls were expensive; for new subscribers,
joining one of the established networks would be particularly attractive, because of the large
number of existing subscribers whom they would be able to contact at cheap on-net rates.
Orange told us that lower on-net charges had been its strategy for building share when it had
first entered the market.
2.125. Vodafone gave us two further reasons for the upward pressure on off-net charges:
first, off-net calls incurred higher costs than on-net calls (see paragraph 6.118) and second, the
downward pressure on on-net charges had meant that revenues had to be recovered from other
39

charges. This rebalancing had to bite on off-net calls, Vodafone said, because any adjustment
to other charges would have resulted in a further divergence from what it called the optimal
pricing structure and would thereby have placed it at a competitive disadvantage.
2.126. We were concerned to establish whether the price differences indicated that off-net
calls were priced significantly above cost; and, if so, why this should be. We received no evidence that the resource costs of off-net calls were so much higher than those of on-net calls as
to justify the very large price differential between them. In its review of competition in the
mobile sector (see paragraph 2.11), Oftel said that there was no clear competitive pressure on
off-net call prices, and no evident differences in demand for on-net and off-net calls, or in the
costs of each type of call, which might justify the price differences between them. On-net and
off-net calls should incur approximately the same costs.1 In any case, Oftel said, even if there
were some small cost difference, this would not account for the wide price differential between
them.
2.127. We suggested to the MNOs that the price difference might relate to cost-allocation
made by reference to different price-elasticities, as envisaged under demand-led pricing (for
example, Ramsey pricing (see paragraphs 2.212 and 2.213)). Vodafone told us that the differential between on-net and off-net pricing did not reflect Ramsey principles. Orange said that it
had no direct evidence on the relative magnitudes of demand elasticities between on- and offnet calls on which to base an assessment of whether on-net prices were consistent with Ramsey
principles.
2.128. Vodafones and Oranges explanations of the commercial rationale for the price
difference are credible. However, Oftel put it to us that on-net calls were in effect another
opportunity which the MNOs offered price-sensitive customers to reduce their bills. It argued
that the on-net/off-net differential was a natural response by the MNOs to the more pricesensitive elements of the mobile phone population. Businesses in particular were attracted to
schemes by which their costs could be kept down; the same would be true within some families, especially where one person was paying for all the calls. In Oftels view, the MNOs had
recognized the constraints on their ability to set high prices where there was high pricesensitivity of this kind and had isolated such price-sensitive customers from the generality of
mobile phone users by specifically targeted tariff schemes. Oftel thought that, in this way, the
degree of market power which MNOs exercised over the remaining, less price-sensitive, customers was increased. The Consumers Association (CA) told us that it would expect the market to be highly segmented given that the MNOs possessed extremely accurate information
about both consumer usage and pricing sensitivities.
2.129. Vodafone and Orange both told us that the price difference between on-net and offnet calls had recently been declining. Vodafone told us that it was clear that on-net and off-net
charges were gradually being rebalanced, to reflect both consumer demand and the fact that, as
the four UK networks were now of roughly equal size, the impetus for the larger MNOs to
differentiate themselves through low on-net prices had largely disappeared. It said that it had
made no reduction in on-net charges over the past two years, during which period it had
reduced other outgoing call charges. Orange said that its Pay Monthly tariffs had increasingly
included both on-net and off-net mobile calls within the bundled inclusive call minutesan
example was Oranges Everytime, which had removed the differentiation except for calls
made outside the bundle. Orange said that, by bringing this type of call within the bundle of
inclusive minutes offered to mobile subscribers, the price of such calls had been reduced and in
due course this should increase the competitive pressure on MNOs to reduce their call termination charges.
1
Off-net calls of course have to bear termination charges, while on-net calls do not. However, if call terminations are priced at
cost, this fact should not invalidate the comparison. In any event, off-net calls are far more expensive than on-net calls, even after
deducting termination charges paid from the price of off-net calls (see Table 5.22).

40

2.130. Figure 6.8 shows the trend in on- and off-net prices from 1999 to 2002. Until the
first quarter of 2000/01, the price of on-net calls was falling while that of off-net calls was
rising. Since then, the trends have been broadly similar. Moreover, as will be seen from the call
tariffs of the MNOs set out in Table 6.32, on-net and off-net prices offered to prepay customers
in May 2002 still show a very significant disparity. This disparity is not diminished by the
inclusion of off-net minutes in inclusive minutes, as prepay customers are not generally offered
inclusive minutes either for on- or off-net calls.
2.131. In summary:
(a) if there is no difference in the relevant elasticities, then off-net charges at least must be

above cost. The commercial rationale for the price differences, as explained by the
MNOs, provided no justification why this should occur and be sustainable in competitive conditions; and
(b) if on the other hand there is a difference in the relevant elasticities, then the difference

would need to be substantial if it were to explain the price difference in terms of Ramsey
principles. Moreover, the Ramsey price models presented by the MNOs, none of which
differentiates between on-net and off-net calls in terms of price elasticities, could not be
relied upon to give an accurate depiction of optimum relative prices even in their own
terms, and would be likely to understate the proportion of fixed costs to be borne by the
relatively inelastic off-net call origination charges and therefore to overstate the proportion to be borne by termination charges.

Customer awareness and behaviour

2.132. We examined the available survey evidence in relation to four aspects of customer
awareness and behaviour with respect to the use of mobile phones: first, the sensitivity of the
mobile subscriber to the price of incoming calls; second, how aware people were that they were
calling a particular mobile network; third, how aware mobile customers were of the relative and
actual prices of on-net and off-net calls; and fourth, customer behaviour when calling a mobile
phone.

Sensitivity to the price of incoming calls

2.133. We consider first how sensitive the mobile subscriber himself is likely to be to the
price of an incoming call when choosing a network. Oftel told us that there was no evidence
that, in general, mobile customers responded to the high prices charged for inbound calls by
switching to another network where the inbound call price was lower. It told us that market
research carried out for Oftel in February 2002 had found that only 15 per cent of those surveyed said that they had found out how much it would cost others to call them before choosing
their network, and only 9 per cent had said that this cost had been a significant factor in their
choice of one network over another. These levels were similar, Oftel said, to those revealed in
its corresponding survey in February 2001. For the majority of respondents, reception quality,
geographical coverage and outgoing call costs were the most significant reasons for choosing a
network. Even for those respondents who had suggested that incoming call costs were a factor
in their choice of network, it was costs in general (which were likely to be mostly outgoing
costs) that represented the most compelling factor in their choice of one network over another.
2.134. Three main points emerged from our own BMRB survey on the question of how
sensitive mobile phone customers are to the price of incoming calls.
(a) The cost of incoming calls (expressed in the survey as the price others pay to call you)

was not a highly rated factor by respondents who paid for the use of a mobile phone,
41

being ranked tenth in a list of 14 factors offered to respondents as likely to be relevant in

their choice of MNO. The most important factor for respondents was the price you pay
to call others, followed in descending order of importance by the quality of the network service, the ease of understanding of prices and the geographic network coverage (see Table 6.57). If the price others pay to call you is regarded as a fair proxy for
the level of termination charges, then the level of those charges is not a factor to which
mobile subscribers are sensitive; many other factors were more important to them. It
will be seen from Table 6.57 that 49 per cent of those surveyed rated the price others
pay to call you as either very or fairly important to them. However, much higher
percentages of respondents rated as very or fairly important each of the nine factors
that were more important to them overall. Thus, 83 per cent thought that the price they
paid to call others was very or fairly important to them; 84 per cent thought this of
the quality of network service they received; 76 per cent thought it of the ease of understanding of prices; and higher percentages (than 49 per cent) were recorded for the six
other factors that respondents ranked above the price you pay to call others.
(b) Over two-thirds of respondents were not concerned about the costs incurred by their

main group of phone contacts in calling their (the respondents) mobile phone.
(c) Nearly two-thirds (61 per cent) of respondents were more concerned about their own

costs of telephoning one of their main contacts than they were about the costs that those
contacts incurred in calling them. Only 9 per cent said that they were more concerned
about their contact groups costs than their own costs and 28 per cent were equally concerned about both sets of costs.
2.135. These findings are consistent with those of surveys of residential customers
commissioned by two of the MNOs. O2s NOP survey found that it was the cost of making
calls from their mobile phone and the overall value for money of the packages available that
were more important to respondents than the cost to others of calling them. Nearly three-quarters of respondents said that the cost to other people of calling them on their mobile phone was
an unimportant factor when they decided which mobile network to join. Under one-fifth said
that it was important. A high proportion (85 per cent) of both categories were unable to say
why they took the view they did. The results of the NOP survey commissioned by Vodafone
showed that the price paid by respondents for making calls and the overall value for money of
the packages on offer were much more important to them than the costs that others incurred to
make calls to them. In our view, all these surveys demonstrate that there are low levels of concern on the part of mobile owners about the costs that others incur in making calls to them.

Awareness on part of customers of calling a particular mobile network

2.136. We looked next at levels of awareness on the part of those mobile customers who
called other mobile networks of the particular network they were calling in each case. Our own
survey indicated that only 28 per cent of respondents who paid for the use of a mobile phone
were likely to know whether they were calling a mobile phone on the same network as themselves. Hence, 72 per cent of mobile users did not know this. The survey carried out for O2 by
NOP found that, in response to the question whether respondents knew which network they
were calling when they rang a mobile phone, there was a difference in awareness between those
who had and those who did not have a mobile phone (see Table 6.55). While 57 per cent of
those owning both a mobile and a fixed line claimed to know the mobile network they were
calling when using their fixed line, only about 30 per cent of those who did not own a mobile
phone were aware of this. When NOP put the same question to mobile-only customers, a similar level of awareness was claimed as that of the owners of both fixed and mobile phones. We
note that MNP will make it progressively more difficult for those calling a mobile to know
what network they are calling, unless they are calling a person with whom they are in a repeat
calling relationship.
42

Knowledge of relative and actual prices

2.137. It is also important, if the arguments advanced by the MNOs that mobile customers
put pressure on them to set termination charges at competitive rates are to carry weight, that
enough customers are shown to be broadly aware of the relative prices of on-net and off-net
calls, of fixed-to-mobile calls and of calls made to a mobile phone at different times of day. We
have already noted evidence indicating low levels of awareness and concern on the part of
mobile owners about the costs of incoming calls. It is also important to consider how knowledgeable mobile phone owners and fixed-line users are about actual costs of calling mobiles
and the relative costs of different call types.
2.138. Mobile phone customers have a large number of different tariffs available to them
for outgoing calls. In part because of this, in constructing questions for our own consumer survey about levels of knowledge of relative and actual prices, we thought it would be difficult to
elicit reliable responses about mobile-to-mobile prices. Instead, we considered it preferable to
ask respondents what they thought it would cost them to make a fixed-line call to a mobile
phone. We found that 77 per cent of customers did not know. (In the example we put to them,
the actual cost of a 2-minute daytime weekday call from a fixed line to a mobile phone would
have been between 38 and 47 ppm at full tariff rates. We set the correct range for responses at
between 30 and 50 ppm. 18 per cent suggested prices that were lower than this and 43 per cent
suggested prices that were higher than this. 16 per cent said that they did not know and 2 per
cent said that it would depend.) Thus, only 21 per cent had even an approximate idea of the
true cost of such a call. (These findings are set out in paragraphs 6.221 to 6.226.)
2.139. O2 conducted a survey which tested consumers knowledge of relative rather than
actual prices, in particular how prices of fixed-to-fixed calls compared with those of other types
of call. This revealed that about 48 per cent of respondents thought that fixed-to-mobile calls
were a lot more expensive than fixed-to-fixed calls (findings which correspond to actual price
relationships); 21 per cent thought they were a little more expensive, 7 per cent thought they
were anything between the same price and a lot less expensive, while 22 per cent did not know.
36 per cent said that off-net calls were a lot more expensive than fixed-to-fixed calls (again, this
corresponds with the actual position); of the remaining 64 per cent, 22 per cent thought they
were a little more expensive, 8 per cent thought they were the same or a little less and 33 per
cent did not know. (We also note that fewer respondents (36 per cent) thought that off-net calls
were a lot more expensive (than fixed-to-fixed) than thought that fixed-to-mobile were a lot
more expensive (48 per cent).) Given that the difference between fixed-to-fixed and fixed-tomobile is actually smaller than the difference between fixed-to-fixed and off-net prices, this
tends to confirm the finding that there is a lack of widespread awareness of the relative prices
of different types of call.
2.140. About half the respondents to O2s NOP survey correctly said that on-net prices
were lower than off-net prices and roughly the same proportion correctly said that the price of
fixed-to-mobile calls varied by time of day. 60 per cent correctly said that mobile-to-mobile
calls varied by time of day. 44 per cent correctly said that the price of calling mobile phones
from fixed lines varied by the network being called; the remaining 56 per cent either did not
give a view or did not know. Oftels February 2002 survey found that only 18 per cent of fixedline customers thought they knew how much it cost to call a mobile from a fixed-line telephone. Finally, Vodafones survey found that 57 per cent of respondents correctly thought that
fixed-to-mobile prices were a lot more expensive than fixed-to-fixed prices, while 24 per cent
thought that they were a bit more expensive; of the remaining 19 per cent, 11 per cent did not
know, 4 per cent thought they were about the same, 2 per cent thought they were a bit cheaper
and the final 2 per cent thought they were a lot cheaper. Vodafones survey also found that
43 per cent of respondents correctly said that the off-net price of calls was a lot more expensive
and 22 per cent a little more expensive, than on-net calls. Of the remaining 35 per cent, 21 per
cent did not know, 9 per cent thought they were about the same, 3 per cent thought they were a
little cheaper and 3 per cent thought they were a lot cheaper.
43

2.141. In our view, these findings indicate that there is some degree of consumer awareness of relative levels of charges for different types of call, but that there is a much lower level
of knowledge of actual prices. Further, and more tellingly, they show that a large number of
consumers have little knowledge of either, and that many peoples beliefs as to both relative
and absolute prices are significantly mistaken. Taken overall, we consider that these findings
reveal a degree of awareness on the part of consumers which is insufficient to enable them to
make an appreciable impact on prices or to drive termination charges down to competitive
levels.

Customer behaviour when calling mobile phones

2.142. The final aspect of the survey evidence relevant to our consideration of the competitive pressures on termination charges is the behaviour of customers when calling mobile
phones. (This evidence is set out in paragraphs 6.235 to 6.241.) Vodafones survey found that
78 per cent of respondents making fixed-to-mobile calls agreed with the proposition that they
tried to keep the length of their calls to a minimum. Over one-quarter of those questioned in
NOPs survey for O2 said that they considered ringing the mobile phone to let the recipient
know that they wanted to talk and to request that the recipient call back. 38 per cent of mobile
customer respondents said that they had offered, or had been asked, to call back in order to save
money but 62 per cent said that they had not.
2.143. Those (a minority) who knew and cared about the cost of making fixed-to-mobile
calls either kept calls short or requested call-back to save money. In both cases, to the extent
that people are using tactics to keep calls short or avoid charges, then the social value and
utility of mobile phone use are lowered. The remainder, who were unaware, pay higher
charges.
2.144. O2, through NOP, asked respondents how they would change their pattern of calling
mobile phones if the cost of calling fell by 25 per cent. Broadly, two-thirds of respondents said
that they would not change their behaviour, while one-third said that they would. But signifycantly more non-mobile owners said this than mobile owners. O2s NOP survey also found that
the main reason by far why respondents called mobile phones (either from a fixed or mobile
phone) was because they wanted to contact the person straight away. It is significant in our
view that price did not feature at all highly in the list of factors suggested to respondents in this
survey as reasons why they called mobile phones. Only 8 per cent said that they rang another
mobile phone on the same network because it was cheaper and a mere 2 per cent said that they
did so because of the free minutes offered.
2.145. We also explored the behaviour of mobile phone customers in our BMRB survey
(see Appendix 6.2). The main finding from this survey was that over 90 per cent of respondents
said that they would not change the way they contacted the person they wished to call in
response to a 10 per cent price increase in call charges. This reveals a marked lack of price sensitivity among those paying for the use of both fixed and mobile phones. The findings in this
and the previous paragraph suggest that it is the ability to reach someone straight away that is
the most important factor for many people in calling a mobile and that price does not feature
highly in their list of priorities.
2.146. In summary, so far as residential mobile phone users are concerned, the cumulative
evidence set out in paragraphs 2.132 to 2.145 suggests that, for most of them, incoming call
charges are a relatively low priority. Moreover, we find low awareness among people as
regards which mobile network they are calling, little evidence that many people try to join the
same network as those people they call most often, low levels of preparedness to adopt strategies that would save the calling party high charges (for example, call-back) and fairly low
levels of knowledge of actual prices. We conclude that there is insufficient knowledge and concern among callers to mobile phones to induce them to change their behaviour to such a degree
as to put pressure on call termination charges and force these down to competitive levels. At the
44

same time, we consider that for those consumers who are aware of the relatively high cost of
making fixed-to-mobile calls and who take steps to reduce the length of their calls, this reduces
the value of their mobile usage.

Conclusion on market definition

2.147. Taking all the evidence set out in paragraphs 2.94 to 2.146, we think it is clear that
there are currently no adequate substitutes for termination of calls on the network of each
operator. There are currently no practical technological means of terminating a call other than
on the network of the MNO to which the called party subscribes and none that seems likely to
become commercially viable in the near future. At the retail level, there are no ready alternatives to making a call to a mobile (such as calling a fixed line or using SMS) and no adequate
consumer strategies (such as call-back) that are easy to use and effective enough to put
sufficient pressure on termination charges. For most mobile users, incoming call charges are a
relatively low priority. There is evidence of a degree of price-sensitivity on the part of a
minority of mobile users, including particular groups such as large businesses. The MNOs
have, however, in our view largely neutralized the pressure which these groups would otherwise have been able to exert on termination charges by offering such users more favourable
terms than the generality of mobile customers, usually in the form of differential tariff rates. By
segmenting the market in this way, the MNOs have been able to sustain higher termination
charges for the generality of customers than would otherwise have been the case. For all these
reasons taken together, we conclude that each MNO has a monopoly of call termination on its
own network.

Other market definition considerations

2.148. We also need to consider, in the context of market definition, whether our terms of
reference require us to include termination charges in respect of voice calls carried over 3G as
well as 2G networks. As we have already noted (see paragraph 2.40), in his September 2001
statement, the DGT said that in his view there should be no regulation of voice calls provided
using 3G technology and spectrum, for three reasons.
2.149. For the purposes of this report, the term call has the meaning set out in Annex A
of our terms of reference. In substance, the definition set out in that Annex refers to voice calls
intended to terminate on a GSM mobile handset using the GSM air interface for the conveyance of that speech call. In our view, this definition means that we are required to make a public interest finding on voice calls delivered using GSM technology. This will include:
(a) voice calls delivered solely using GSM; and
(b) for calls which are delivered using both GSM and 3G technologies, that proportion of

call minutes which is delivered over GSM technology; but neither

(c) calls that are carried and terminated using only 3G technology; nor
(d) for calls that are delivered using both GSM and 3G technologies, that proportion of call

minutes which is delivered over 3G technology.

In our view, this definition means that we are required to make a public interest finding on
voice calls carried over GSM but not 3G technologies.
2.150. The MNOs and Hutchison 3G, however, argued that because handsets would be
developed with the capacity to handle both 2G and 3G calls and, in the early years of 3G, calls
45

would be passed backwards and forwards between the two technologies, it would in practice be
very difficult to regulate only 2G termination. Indeed, customers with such dual-mode handsets
would be unaware of whether their voice calls were being transmitted via the new or existing
technology. Oftel told us that under its proposals it would be possible for MNOs to set an
unregulated charge for 3G and a different regulated charge for 2G. Even if customers were
unaware of the network underlying their calls, Oftel said, the terminating MNO would know
whether the call was 2G or 3G, or could set up an inter-network or wholesale billing system to
enable it to differentiate. The MNOs disputed this, saying, first, that it was meaningless to
describe a call as 2G or 3G, because it could be a mixture of both; and second, they could
not tell, even in retrospect, how much of each technology any particular call had used. Oftel
envisaged a single charge that would be a weighted average of the 2G and 3G elements of the
call, the MNOs accounting to Oftel as to the relative proportions, based on engineering estimates at an aggregate level of how much of each type of network was used by calls. This
appeared to us, based on evidence from the MNOs, to be feasible (see paragraphs 3.102 to
3.105).
2.151. The definition of a call in our terms of reference means that we are not required to
consider termination charges made in respect of text messaging; for the same reason, international roaming also falls outside our terms of reference.
2.152. We conclude that the scope of our inquiry is voice call termination on the GSM
networks of O2, Vodafone, Orange and T-Mobile and that there is a separate market for termination of voice calls on the network of each of these four operators. We address the practical
issues surrounding 3G later, in the context of remedies (see paragraph 2.545 to 2.547).

Competitive pressures at the retail level

2.153. The four MNOs have laid great emphasis on competition between the MNOs at the
retail level which, in their view, is the source of much of the competitive pressure that leads to
the dissipation of any excess profits which may arise from high termination charges. They have
argued that, even if termination charges were excessive, any excess profit would be competed
away in prices at the retail level. It is therefore necessary for us to examine what evidence there
is of competitive activity in the retail market, in order to form a view both on the extent to
which this competitive activity is likely to be effective in keeping termination charges down,
and what might happen to call origination and access charges if termination charges were to
change.
2.154. The MNOs pointed to a number of factors which, they said, demonstrated the effective competitiveness of the mobile sector. These included, in O2s submission, low and falling
prices, operators that were improving their efficiency, a dynamic market with shifting market
shares, a high degree of switching, a high degree of innovation and some new entry in the shape
of Hutchison 3G. Orange drew our attention to falling prices, rapid service innovation, tariff
charges that were offering customers increasing value for money, low barriers to switching,
high consumer satisfaction and high consumer penetration as evidence of the intensity of
competition in the mobile market. T-Mobile pointed to the long-term decline of mobile prices,
high consumer satisfaction, competition in quality and lack of evidence of anti-competitive
behaviour as evidence that the UK market was extremely dynamic and competitive. Vodafone
said that the best evidence of the effectiveness of competition in the market for mobile services
was the speed at which Orange and T-Mobile had been able to gain market share.
2.155. Although we acknowledge that these factors are likely to be indicators of competition at the retail level, they are not by themselves a guarantee of full competitiveness in the
sector. As we have already noted, Oftel considers that the retail market is not yet effectively
competitive. Oftel told us that it was reasonable to consider competition in the mobile sector in
the context of a single retail market for all outgoing services, because the competitive con46

ditions for particular outgoing retail services, for example, subscription and call tariffs, were
likely to be the same. The sector still fell short of effective competition, Oftel said, because of
the room for improvement in consumer awareness of different prices and tariffs, evidence of
the existence of some barriers to customers switching networks and poor levels of consumer
information. Prices were above the level that would be found in an effectively competitive
market; and Vodafones return on capital employed (ROCE) had consistently, and substantially, exceeded the cost of capital. Oftel noted that, in coming to its view that the mobile
sector was prospectively competitive, it had placed weight on the long-term trend of declining
prices in the mobile market. It told us that, since publishing its 26 September 2001 review
statement, however, there had been little movement in mobile prices which (in the early part of
2002) were at about the same level as in May 2001; this was evidence, it said, that the declining
trend of retail prices had flattened out.
2.156. In this section, we look first at the profitability of the MNOs (see paragraphs 2.157
to 2.162) and then go on to consider market shares (see paragraphs 2.163 to 2.167) and entry
conditions (see paragraphs 2.168 and 2.169). We then look at indicators of competition in
relation to access and call origination issues (paragraphs 2.170 to 2.173). We discuss access
first (paragraphs 2.174 to 2.197) and then call origination (paragraphs 2.198 to 2.210). Finally
in this section, we give our conclusions on competitive pressures at the retail level (see paragraph 2.211).

Profitability
2.157. We looked at the profitability of each of the four MNOs. It appeared that Vodafone
at least had been making profits in the UK in excess of its cost of capital over the period 1998
to March 2001. Oftels September 2001 Effective Competition Review showed O2s returns
over the same period declining to a level below Oftels estimate of cost of capital, while
Oranges returns available up to December 2000 had not reached Oftels estimate of cost of
capital. T-Mobile was loss-making throughout the period. Thus, none of these three MNOs was
found by Oftel to have profits that persistently and significantly exceeded the cost of capital.
2.158. Oftel put it to us that a finding of effective competition implied an absence of operators with market power, as an operator with market power could raise prices above the competitive level. Judged largely by its high levels of profitability, Oftel said that Vodafone appeared
to be pricing above the competitive level. Oftel told us that, as regarded overall profits in the
mobile businesses, it was appropriate to regard prices that reflected a reasonable ROCE as a
proxy for the competitive level of prices. This ROCE would relate to an MNOs efficiently
incurred costs, adjusted to allow for economies of scale and the costs of a hybrid GSM 900/
1800 MHz network operator compared with an 1800 MHz operator. On that basis, Oftel had
found that Vodafones ROCE was persistently and substantially in excess of the cost of capital
over the period 1998 to 2002.
2.159. Vodafone put a number of arguments to us as to why its profitability levels, and the
trend in those levels, were not as Oftel believed them to be, and proposed various adjustments
to Oftels calculations. Also, Vodafone said that its profit levels should not be regarded as indicating that the mobile sector was not competitive. Vodafone said that the DGT had placed too
much weight on its profitability, not only overstating its return on capital but incorrectly concluding that a high rate of return on capital was evidence of a lack of competition. It told us that
its return had been achieved because of its advantages over the other MNOs in terms of customer mix and cost efficiency.
2.160. Oftel stated in its 26 September 2001 Effective Competition Review (see paragraph
A8.9 of that review) that supernormal profits were consistent with competitive market conditions under particular circumstances. For example, high returns might reflect relative
efficiency and high levels of innovation by an operator. These factors might give rise to excess
47

profits in any single period. Oftel said, however, that it was clear that persistent high returns
were difficult to reconcile with a competitive sector, because over time competitors should be
able to replicate efficiencies or produce rival innovations. We broadly concur in this view.
2.161. However, when deciding whether persistently high profit levels are an indicator of
ineffective competition, it is necessary to consider the circumstances in which such returns are
earned. Vodafones returns have been earned in a period when the mobile phone market has
been expanding extremely rapidly. In our view, the circumstances in which persistently high
profits become an indicator of ineffective competition is a matter of judgement, about which
contrary views may legitimately be held. In the circumstances, we do not conclude that
Vodafones high profit levels, whether they have been declining (as Oftel believed) or remain
approximately constant but at a lower level (as Vodafones evidence indicates) demonstrate, in
themselves, ineffective competition. These conclusions are therefore not affected by whether
Oftels or Vodafones version of Vodafones ROCE, or some other version, is the right one to
examine. Our view is that the profitability of each MNO over the past few years is not critical
as an indicator of competition in any particular part or parts of the wholesale or retail market.
2.162. Whatever our views on the matter, it is obvious from its submissions that Vodafone
believes that the adjustments it has made to its published accounts are necessary, in both principle and practice, for the valid computation of its ROCE, and Oftel will no doubt wish to consider them carefully.

Market shares
2.163. We calculated the MNOs market shares using the following measures: number of
subscribers, value of outgoing revenue and the number of outgoing call minutes.1 These are set
out in Table 2.4 (and in more detail in Tables 6.39, 6.41 and 6.42).
TABLE 2.4* Percentage share held by MNOs, 2001/02

Vodafone
O2
T-Mobile
Orange
Total
Volume

Subscribers
year ending 2001

Outgoing revenue
2001/02

24.6
24.7
23.2
27.6
100.0
44.9m

34.8
22.4
16.6
26.2
100.0
7.1bn

Outgoing call minutes

2001/02
30.8
20.7
19.7
28.7
100.0
46.3bn

Source: CC calculations on data provided by Oftel.

*There are discrepancies between the share figures in Table 2.4, which were derived from data provided by Oftel and those in Table 2.1, which were based on information from the MNOs.

2.164. Each of the four MNOs had broadly the same proportion of subscribers in the year
to 2001: Vodafone had 24.6 per cent; O2, 24.7 per cent; T-Mobile, 23.3 per cent; and Orange,
27.6 per cent. The picture has changed significantly since 1997 when Vodafone and O2 had
nearly 80 per cent of subscriber numbers between them.
2.165. In terms of revenue from outgoing calls, fixed charges and connections earned by
each of the MNOs between 1996/97 and 2001/02, Vodafone still had the largest share in
2001/02 (34.8 per cent), even though its share fell by some 11 percentage points over the period
1
We also calculated shares according to the number of text messages each MNO sent and received but it is not necessary for us to
consider these here, as we have decided that text messages do not fall within our terms of reference.

48

from 1996/97. O2s share also fell and was lower (at 22.4 per cent) than that of Orange
(26.2 per cent). T-Mobiles share was the lowest at 16.6 per cent, but both its and Oranges
shares rose markedly over this period.
2.166. In 2001/02, Vodafone and Orange had the highest percentage of outgoing call
minutes (30.8 per cent and 28.7 per cent respectively) and T-Mobile the lowest at 19.7 per cent.
Total outgoing call minutes rose from 6.8 billion in 1996/97 to 46 billion in 2001/02.
2.167. No single dominant or dominating player in the retail market emerges from these
measures of market share, although some MNOs are clearly stronger than others. The fact that
there are currently four more or less evenly-sized businesses is unusual by European standards1
and gives all the MNOs a similar, broadly common interest in events affecting the sector as a
whole.

Entry
2.168. When Hutchison 3G starts operating, it will bring the number of MNOs in the UK
to five. No other new entry is expected in the foreseeable future because of a shortage of spectrum, although it is always possible that the ownership of particular MNOs may change hands.
2.169. So far as its high-speed mobile data transmission is concerned, Hutchison 3G will
want to build up market share rapidly in those limited parts of the country in which it intends
initially to concentrate its services. Oftel told us that mass market take-up of 3G services was
not expected before 2004 at the earliest. So far as voice traffic is concerned, in the initial
phases, Hutchison 3G will be unable to offer its customers many other customers to call on an
on-net basis. The incumbent MNOs will have the incentive, therefore, to raise their termination
charges and lower their on-net charges, and if they did, this would make it more difficult for
Hutchison 3G to establish itself, at any rate in mobile voice calls.

Access and call origination issues

2.170. We distinguished earlier (see paragraph 2.91) three activities involved in the take-up
and use of mobile phones by consumers: access, call origination and call termination. We have
already concluded that call termination forms a separate market at the wholesale level. We
noted that access and call origination are retail activities. We now examine the degree of
competition at the retail level in the context of each of these activities, beginning with access.
2.171. The MNOs rejected any analysis of the retail market which separated services for
access from the making and receiving of calls. For example, O2 said that subscription charges
and outgoing charges were part of a package, so that broadly speaking a consumer could obtain
attractive outgoing rates (that is, many free minutes) in exchange for a higher subscription
charge.
2.172. T-Mobile told us that it was in the business of selling calls to and from its subscribers, and that around 65 per cent of its customers (that is, prepay customers) did not pay
subscription charges. It said that the existence of subscription charges for a minority of customers reflected a two-part tariff structure where the subscription charge provided for a certain block of minutes. In addition, handsets were of value because they were part of the means
(along with the network infrastructure) by which calls were made and received; the degree to
which the cost of handsets was recovered upfront or in call prices would also reflect optimal
1
We note that the structure of the GSM sector in the UK is rather different from that of most other EC countries. With four
established operators, each with roughly equal market shares, and a fifth about to enter the market, the UK has a larger number of
MNOs and these more evenly matched than is the norm on the Continent, where the pattern is more typically that of a single
dominant operator and one or two smaller players.

49

two-part tariff considerations. Orange told us that the monthly payment made by post-pay or
contract customers was not a subscription charge (that is, a fee paid just for being connected to
the network) but a service charge for elements such as inclusive minutes, answerphone service,
customer care and itemized billing.
2.173. We believe, nevertheless, that there is an important distinction to be made between
the competitive activity which takes place among the MNOs to acquire and sign up new customers to their networks, in respect of which the MNOs incur various subscriber acquisition
costs, on the one hand, and ongoing competition among the MNOs in respect of call charges on
the other. We therefore propose to examine access and call origination separately for the purposes of establishing where competitive pressures are felt in the retail market. We begin with
access and look at the sales and marketing of mobile services (see paragraphs 2.174 to 2.176),
before examining the nature of the subsidies that the MNOs offer customers to induce them to
join the network (see paragraphs 2.177 to 2.187). Finally, under the heading of access, we look
at switching (see paragraphs 2.188 to 2.196) before reaching our conclusions on access (see
paragraph 2.197).

Access
Sales and marketing and customer care
2.174. The MNOs incur various categories of expenditure in attracting customers to their
networks and encouraging them to make calls. The main categories are customer acquisition
and retention costs, sales and marketing costs, and customer service (including customer care
and billing services). We discuss sales and marketing, and customer care first, then go on to
consider customer acquisition and retention.
2.175. The MNOs each spent, an average, 185 million on selling and marketing their services in 2001. Average expenditure on customer care for the four MNOs was 154 million in
2001 (see Table 7.10). We calculated that the average sales and marketing costs per subscriber
were a little over 17, while average customer care costs per subscriber were 14.50.
2.176. The MNOs argued that their sales and marketing expenditure was by no means
excessive. As a proportion of total voice and SMS revenue, their spend was, they said, reasonable, ranging from 5.6 per cent to 8.8 per cent. Vodafone said that there was nothing discretionary about its marketing expenditure: it was a sign of how effective competition was that
each MNO was forced to incur ever greater costs to win and retain subscribers, up to the point
at which any excess profits were competed away.

Customer acquisition and retention

2.177. An important part of the MNOs competitive strategy is the funding of initiatives to
encourage customers both to join, and remain on, their networks. Such investment is worthwhile for the MNOs, because customers generate the MNOs future revenue streams through
their subscription and other tariff payments and through the termination charges they generate
by receiving calls. Within customer acquisition and retention costs we have included net handset costs (ie handset costs less handset revenues), discounts and incentives to retailers, and sales
and marketing. In their last financial year for 2001/02, the four MNOs spent, on average, 682
million on these categories (see Table 7.11).
2.178. Our primary concern in this section is to consider the implications of the MNOs
strategy for acquiring and retaining customers through their expenditure on customer acquisition and retention. In this connection, we look at what the MNOs spend on acquiring cus50

tomers and at how, and to what extent, customers are given a financial incentive to join or stay
on the network. We have already considered sales and marketing costs and customer care above
(see paragraphs 2.174 to 2.176).
2.179. Prices for mobile phone subscribers comprise a number of components. Customers
require a handset, a SIM card and a tariff package before they can make and receive calls. They
have the choice of either a prepay or post-pay package, the latter involving a contractual
arrangement between the MNO and the mobile customer under which the customer makes
periodic (usually monthly) payments in return for a handset and SIM card (or a SIM card alone
if the customer wishes to retain his existing handset) and a specified allowance of call minutes
(the number of minutes allowed varying with the amount of the subscription). Within each of
the prepay and post-pay categories there is a choice among various tariff plans. Each tariff plan
has many different pricing components, reflecting for example the time of day and day of week
when the call is made, and whether the call is on-net, off-net or mobile-to-fixed. (Tables 6.31
and 6.32 illustrate the range of prepay and post-pay tariffs on offer.) All of these prices may
have an effect on the willingness of particular customers to join the different networks.
2.180. A key element in the MNOs marketing strategy is the sale of a bundle or package
of services which includes, at the very least, the sale of a SIM card which connects the customer to a particular network. Prepay customers purchase connection to the network outright
and also a handset if they need one. When contract (post-pay) customers buy a new mobile
handset, they sign a contract (typically for 12 months), which commits them to paying a
periodic (usually monthly) subscription charge, known as a line rental. They receive a number
of inclusive minutes per month and in some cases a number of inclusive text messages. As
some contract customers do not use all their inclusive minutes, at least part of the monthly subscription charge might be thought of by the customer as a way of paying for the mobile handset. It is possible to purchase a handset without a network connection, but this would be very
unusual (and indeed would be more expensive in the short term than buying a handset with a
network connection). The purchase of a mobile phone therefore normally consists of two transactions, namely the purchase of the mobile handset itself and the establishment (or continuation) of a connection to a network.
2.181. The parties stressed the important role that subsidization of the handset played in
their marketing of the bundle of mobile services to customers. O2 told us that post-pay handsets were subsidized (that is, sold below incremental cost) and would remain subsidized for the
time being. It was still commercially important, O2 said, to provide incentives for customers to
take up post-pay packages. Post-pay customers tended, once established, to make and receive
more calls than prepay customers. It added that, although prepay handsets had been subsidized
in the recent past, increasingly the sale of such handsets was making a full contribution to costs.
2.182. In order to encourage retailers to recruit more customers to their networks, the
MNOs may purchase handsets from manufacturers which they then sell on to retailers, often at
below cost, to enable the retailers to sell the handsets to their retail customers at a subsidized
price. Some MNOs sell handsets to the retailers at cost and make incentive payments to
retailers when the handset is sold as part of a customer joining the network. All the MNOs
incentivize retailers to recruit customers to their networks.
2.183. In their most recent financial year for 2001/02, each MNO spent, on average,
122 million on net handset costs, and 381 million on related discounts and incentives (see
Table 7.11). The MNOs told us that it was impossible to know with any precision how much of
the sums paid to retailers by way of discounts and incentives went into handset subsidies. We
accept this. However, we have been told that some of the funds they make available to retailers
are used towards subsidizing handsets. Although we were unable to quantify how much of the
MNOs costs for discounts and incentives was used in this way, we believe that this category of
spend can properly be regarded as representing a type of subsidy. These expenditures amounted
to some [  ] per new subscriber to the networks in the year 2001/02 (see Table 7.12).
51

2.184. The figures in the previous paragraph represent the up-front handset subsidy (that is,
the subsidy given at the time of sale). To calculate any offset to the subsidy, we considered
whether some proportion of the contract customer acquisition costs should be offset by part of
the periodic subscription revenue that the MNOs receive from those customers. The MNOs
agreed that part of the subscription revenue from contract customers could be regarded as call
revenue and that this could be calculated by valuing the bundled free minutes offered as part of
those packages. The MNOs differed in their view as to what proportion of the subscription
revenue could reaonably be allocated to call revenue and hence what residual amount was left.
The allocations of subscription revenue to call income by the different MNOs left a residual of
between zero and 73 per cent of subscription revenue which we consider should be offset
against customer acquisition costs. It was clear to us that any consideration of call charges and
customer acquisition costs necessitates an allocation of periodic subscriptions between these
two elements. In the absence of any consensus among the MNOs as to either the correct basis
of allocation, or the proportion to be allocated to outgoing call revenues or to offset customer
acquisition costs, we took the view that 50 per cent of subscription revenue should be attributed
to calls, and 50 per cent to offsetting customer acquisition costs (see paragraphs 7.156 and
7.157 and Table 7.13).
2.185. Making this adjustment resulted in average customer acquisition costs (ie net handset costs, discounts and incentives and sales and marketing) being reduced in 2001/02 from an
average for the MNOs of 682 million, to 425 million as shown in Table 7.13. We thereby
calculated that the average net cost of acquiring a new customer was around 100 for the most
recent completed financial year.
2.186. The MNOs produced various figures on different bases to quantify the average level
of subscription/handset subsidies. All the figures that they provided indicated that the average
subsidies per customer were greater for post-pay customers than for prepay customers. MNOs
were prepared to invest more up-front to attract post-pay customers, as such customers tended,
on average, to have a higher lifetime value.
2.187. Several of the MNOs drew to our attention the reduction in handset subsidies that
would be necessitated if termination charges were reduced, and the extensive loss of their customer base which they said would result if this were to occur. We discuss the impact which we
would expect any reduction in termination charges to have on customer recruitment when we
discuss possible remedies, later in this chapter (see paragraph 2.567).

Switching or churn
2.188. Finally, under access, we look at the extent of switching or churn in the mobile
sector. The MNOs put it to us that switching or churn rates in the mobile market were an
indicator of competition among them. They monitor churn rates closely. The usual measure of
churn is the number of disconnections from an MNOs network in any period, expressed as a
proportion of the average number of customers in that period. However, there is no industry
standard for calculating churn; and the term churn is sometimes used to describe both disconnection from a network and disconnection from a particular tariff (which does not necessarily
mean that the MNO has lost a customer). Churn rates therefore need to be treated with caution.
2.189. We sought to estimate the extent of churn (by which we mean disconnection from
an MNOs network), and to obtain some idea of what proportion of customers are new to the
network on the one hand and what proportion are simply churning between networks on the
other. We are not interested, for these purposes, in those who switch tariff but remain on the
same network.
2.190. For residential customers and some SMEs, switching mobile networks appears to be
quite straightforward and relatively inexpensive. The Large Business Users Panel, however,
52

told us that large customers found switching more difficult as there could be logistical problems
in getting new handsets or SIM cards to employees. Contract customers seeking to leave a
network before the end of the contractual period will normally be obliged to pay all outstanding
monthly subscriptions. Customers also incur a charge to purchase a new SIM card and some
customers have to pay to have their SIM cards unlocked.
2.191. Oftels November 2001 survey found that there was an appreciable level of awareness among consumers that it was possible for them to switch networks by changing their SIM
card (63 per cent of mobile customers were aware of this) (see paragraph 6.203). However,
only 6 per cent of these aware customers claimed to telephone on more than one network by
using more than one SIM card in their mobile handset, while 11 per cent of them had switched
networks while keeping their original handset.
2.192. Number portability (see paragraph 2.34) provides a way for mobile customers to
switch networks without the associated costs of changing stationery or business signage, or of
contacting colleagues, family and friends about a change of number. Oftel found, however, that
number porting was not widely used (see paragraph 6.204); its August 2002 survey of residential customers found that only 18 per cent of respondents who had ever changed networks had
ported their existing number; 82 per cent had changed their number. Customers informing their
MNOs that they intend to move to other operators may well be offered incentives to stay and
this might be thought to lower switching rates.
2.193. A number of surveys carried out by Oftel and the MNOs have sought to establish
the rate of churn between networks (inter-network churn) in the sector (see paragraphs 6.178 to
6.182). Three surveysby Oftel in February 2002, GfK for O2 and TNS for Oftel in November
2001all found broadly similar churn rates, the percentage of people saying that they had
changed networks or service provider at some time in the past ranging from 22 to 29 per cent.
The MNOs own surveys produced annual inter-network churn figures of [] to [] per cent
for O2, [] to [] per cent for Vodafone and [] to [] per cent for T-Mobile. The MNO
figures suggest that an individual MNO is losing up to one-quarter of its customers to other
networks every year. However, the churn figures may be higher than they otherwise would
have been, as a result of retailers receiving higher rewards from the MNOs for signing up new
customers than from switching, or upgrading the handsets of, existing customers. The true rate
of churn between networks appears to be less than 20 per cent.
2.194. Using Oftel and MNO data, we estimated that, of the total number of subscribers in
March 2002 (as measured by handset sales), 10 per cent were genuine new subscribers to
mobile telephony during 2001/02. The proportion of new to existing subscribers rose up to
March 2000, and then fell sharply to March 2002 (see Table 6.52). We also sought to estimate
the proportion of new residential customers, this time using survey data. The latest survey data
(for August 2002) shows a year-on-year increase in new mobile customers of 1.5 million, or
about 5 per cent of all mobile customers (see Table 6.53). Previous survey data (for February
2002) had shown the proportion of new to existing customers as being 9 per cent (see paragraph 6.199). These findings indicate a sharp drop in the rate of new customers over a
comparatively short period.
2.195. We estimated that the percentage of genuinely new subscribers (that is, first-time
subscribers who have not previously owned a mobile phone before signing up) is a modest
fraction of the total number of customers who churn between networks. This means that the
MNOs are very largely taking customers from each other.
2.196. The MNOs submitted that a high rate of inter-network churn would normally be
regarded as an indication of intense competition in a sector and that this was indeed the case in
the mobile sector. They said that we were therefore inconsistent in appearing in our Remedies
53

Statement (see Appendix 2.2) to regard a high level of churn as excessive. We accept that the
movement of customers between the different MNOs is indicative of competition between
them. To that extent, we think the mobile market is competitive. However in our view, the
degree of churn in this sector is symptomatic of the structure of retail incentives that the MNOs
have adopted. This tends to encourage frequent changes of network and upgrading of handsets
by mobile users rather than the expansion of call activity.

Conclusion on access
2.197. We conclude that competition among the MNOs to attract and sign up subscribers
to their networks is vigorous. Moreover, a range of retailers compete to sign up customers.
However, in our view, the nature of the competition between the MNOs at the access level,
involving the subsidization of customer acquisition, has led to a structure of retail incentives
that leads to a higher degree of churn between networks and a lower level of call activity by
consumers than would otherwise have been the case.

Call origination
2.198. We now turn to consider evidence of the degree of competition at the retail level in
respect of call origination. We discuss the margins that the MNOs make on their outbound calls
(see paragraphs 2.199 to 2.201) and the transparency of tariffs, including the number of minutes
that remain unused in mobile packages (see paragraphs 2.202 to 2.209). We then reach our
conclusions on call origination in paragraph 2.210, and our overall conclusion on competitive
pressures at the retail level in paragraph 2.211.

Margins
2.199. Each of the MNOs offers contract customers a wide variety of monthly subscriptions, ranging from 1015 to over 75. Most packages are configured so that the higher the
monthly subscription charge, the greater the number of free inclusive minutes in the subscription price and the lower the call charges. MNOs have recently begun to offer packages with
inclusive minutes which can be used on any network (that is, for off-net, as well as on-net and
mobile-to-fixed, calls) and at any time of day. O2 offers nine such packages, T-Mobile six,
Orange four and Vodafone two.1 This is a further development of the move by MNOs to widen
the coverage of inclusive minutes on their networks. Very few packages now restrict inclusive
minutes to off-peak usage, although some still exclude off-net calls.
2.200. We examined the margins that the MNOs make on different call types. For all the
MNOs, the largest unit contribution is made by off-net calls. Table 5.22 shows that the average
contribution across the four MNOs (including a proportion of subscription revenue) made by
this type of call in the year 2001/02 was 11.7 ppm, compared with just 3.4 ppm for on-net calls
and 5.4 ppm for mobile-to-fixed calls. To a large extent, these margins help to meet non-network costs not recovered in termination charges or subscription revenues.
2.201. We discussed the difference between the prices of on- and off-net calls earlier, in
paragraphs 2.122 to 2.131, and expressed our view that this difference threw doubt on the
MNOs claim that their pricing approximated to a Ramsey structure of prices. Although the
MNOs have begun to offer pricing packages which allow their customers a certain number of
either free on-net or free on-net and off-net call minutes, we consider that the marked differences in average contribution made on these two call types indicate a less than fully competi1
Vodafone told us that it had seven any network tariffs, the number depending on how tariffs were classified (see paragraph
6.133).

54

tive market in call origination. These differentials vary considerably from one MNO to another
[
Details omitted. See note on page iv.
], so that the average
contribution of 11.7 ppm for the four MNOs for off-net, compared with 3.4 ppm for on-net,
calls disguises very high contribution relative to on-net calls in the case of two of the MNOs. It
is significant, in our view, that the MNOs choose to offer tariff packages that include on-net
and off-net minutes, rather than bringing the prices of these calls closer to each other (although
Vodafone told us that its on-net and off-net prices had moved closer together by virtue of a
reduction in its off-net price). We acknowledge that the MNOs are offering more of these
packages. Nevertheless, in a fully competitive market, we might have expected to see the net
price differential between on-net and off-net call prices (that is, contribution after taking
termination charges into account) competed away across the board.

Transparency of tariffs
2.202. Oftel put it to us that the proliferation of tariff structures offered by the MNOs had
evolved to appeal to distinct user profiles or market segments, the most obvious split being
between business and residential consumers. Other sub-segments related to particular customer
characteristics, for example, heavy or light users. In Oftels view, the many different tariff
structures might have the effect of confusing consumers and result in prices being less transparent and hence more difficult to compare. This might lead to higher switching costs for
mobile customers.
2.203. Oftel also submitted that the range of tariff structures and tariffs was a form of price
discrimination, in which products were designed in such a way that customers could self-select
themselves on to the relevant tariff structure according to their demand characteristics. The
welfare outcome of this was, in Oftels view, ambiguous. Price discrimination could be
welfare-enhancing if it had the effect of increasing the number of mobile customers who used
mobile services, or their usage of mobile phones. Equally, it could reduce welfare if the result
was that overall volumes of purchases were reduced or resulted in consumers making wrong
choices. Oftel said that its research showed that there was evidence of low consumer awareness
of the prices of different kinds of call and it was reasonable to conclude that, despite the
considerable amount of information on offer, consumers had difficulty in making effective
choices between the wide range of tariff packages available to them.
2.204. We asked the MNOs whether the number and variety of pricing packages available
to customers made comparisons between the prices of one MNO and another more difficult for
mobile customers than they needed to be.
2.205. Vodafone agreed that there was a wide range of tariffs available to customers and
argued that this was the consequence of having a competitive market. MNOs, MVNOs and service providers all competed to offer attractively-tailored tariffs designed to suit every kind of
customer. That did not mean that consumers made confused or bad choices; each consumer
readily identified a small number of tariffs likely to be suitable for his usage pattern, and could
readily make price comparisons among the tariffs available. Vodafone said that survey findings
had shown that the majority of customers found it easy or fairly easy to find the best package.
Moreover, there were numerous sources of information for consumers and as they became
more experienced in using mobile phones, consumers could predict their own usage better and
hence make a better-informed choice of tariffs. Vodafone drew our attention in this connection
to the fact that some [] per cent of its subscribers could be expected to switch to a different
plan each year; this might involve changing tariff or selecting an option such as an off-net
bundle to be added to the existing tariff.
2.206. O2 and Orange made very similar points. O2 told us that it was aware of no evidence
that competition was muted by virtue of the MNOs multiple tariff offers. Indeed, the range of
tariffs on offer meant that customers preferences would be met more precisely. Moreover,
55

there were numerous sources of consumer advice from the MNOs themselves, specialist
retailers, specialist web sites and specialist magazines. O2 said that consumers became more
knowledgeable about prices with repeat purchases and would be aware of charges from their
monthly bill or, in the case of prepay, the frequency with which they topped up their credit.
Survey evidence showed that consumers were adapting their behaviour in response to price and
were not making ill-informed decisions. Orange told us that, while there were certainly a large
number of tariffs, all fell into common patterns (for example, number of inclusive minutes or
monthly subscription charge), so that comparisons were relatively easy to make. For most consumers, it said, only a small proportion of tariffs were relevant to their personal usage, further
simplifying the choice.
2.207. We asked the MNOs what proportion of inclusive call minutes in their subscription
packages were unused. O2 told us that over [] per cent of its call minutes were unused.
Oranges unused call minutes for all but one of its tariffs (Everyday 50) amounted to less than
[] per cent; however, the unused minutes on Everyday 50 were [] to [] per cent.
T-Mobile had just under [] per cent of unused minutes, while Vodafone had between just
over [  ] and [  ] of unused minutes for individual tariffs, with a weighted average
of [] per cent for 2000/01. These figures (set out in paragraph 6.135) indicate a substantial
proportion of unused minutes in inclusive packages.
2.208. The validity of the benefits to customers claimed by the MNOs for the proliferation
of tariffs depends, in our view, on three factors: first, the ability and willingness of customers to
forecast their usage patterns; second, the existence of tariffs reflecting such usage patterns and
third, the degree to which customers are either aware of or concerned about the existence of the
tariffs that would best suit them. The high proportion of unused free minutes indicates to us
that at least one of these factors is likely to be absent. High proportions of unused minutes such
as these could be evidence of a lack of competition between the MNOs, since if competition
were intense, we would expect mobile customers actual usage to correspond more closely with
the packages they acquire. Persistent unused call minutes would seem to represent overpayment
by customers for the services they purchase.
2.209. We accept that there is information available to help consumers choose the best
package and that mobile phone users are likely to become more knowledgeable the more they
use their phones. However, in our view the bundled features of the packages on offer may make
it more difficult for consumers to get what they want than if the features were unbundled.
Moreover, the extent to which consumers make the right choices from among the numerous
packages on offer should be seen in the light of the evidence of consumer knowledge and
awareness that we discussed earlier (see paragraphs 2.132 to 2.146). We believe that this
knowledge and awareness are at lower levels than would be effective fully to constrain mobile
prices. In our view, therefore, there is a lack of transparency in the pricing of mobile packages
which makes it difficult for mobile customers to compare different packages, make the right
choices and bring downward pressure to bear on prices. Indeed, to the extent that, as Vodafone
said, the MNOs offer attractively tailored packages designed to suit every customer, such packages represent a type of segmentation of the market that facilitates price discrimination and
localizes price competition in a few groups of consumers.

Conclusion on call origination

2.210. We conclude that competition in call origination is less intense than competition for
access. This is evidenced in high margins for off-net calls, a substantial level of unused free
minutes in mobile packages, and the bundling and complexity of call tariffs. The MNOs appear
to display at least some power to set call origination price structures that suit them, and to set
prices for at least some calls well above any reasonable estimate of the incremental costs
involved in making those calls (we discuss costs in paragraphs 2.217 to 2.333).
56

Conclusion on competitive pressures at the retail level

2.211. We have concluded that each of the MNOs has a monopoly of call termination on
its own network (see paragraph 2.147). We have further concluded that, while there is intense
competition among the MNOs to attract and sign up subscribers to their networks (see paragraph 2.197), there is less effective competition in call origination (see paragraph 2.210), as
evidenced by high margins for off-net calls, a substantial level of unused free minutes in mobile
packages, and the bundling and complexity of call tariffs. All this indicates, in our view, less
than effective competition at the retail level.

The basis of the MNOs pricing

2.212. As we have already noted, our terms of reference require us to investigate and
report on whether termination charges would, in the absence of a charge control mechanism on
them, be set at levels which operated, or might be expected to operate, against the public interest. Before we can reach a conclusion on that question, it is first necessary to describe what the
MNOs have put to us as their approach to pricing and the allocation of costs in the mobile sector. (We consider costs in paragraphs 2.217 to 2.333). So far as pricing is concerned, the MNOs
told us that their pricing was in effect demand-led rather than cost-plus, and reflected the need
to recover the substantial fixed and common costs which they incurred. They said that this was
a fundamental aspect of the case that they were putting before us.
2.213. The MNOs told us that, when they acquire new customers, they regard themselves
as effectively buying a revenue streamthat is, the revenue they expect to derive from customers, including that from subscriptions, from call charges when their customers make outbound calls and from termination charges when other people call customers on their networks.
The handset subsidy which the MNOs give their customers (formerly both prepay and post-pay,
but since last year predominantly post-pay, customers) may be thought of as part of the MNOs
investment in this revenue stream. Other payments are also part of that investment, but handset
subsidies represent the most significant element. Part, but not necessarily all, of the MNOs
payments to retailers may be seen as an indirect way of granting handset subsidies. The MNOs
told us that, in competing to win new mobile customers, they had greater incentives to reduce
charges to mobile customers than to reduce call termination rates (in line with cost reductions).
The right way to maximize the revenue stream, they said, was therefore to price broadly in line
with Ramsey principles, which would result in lower charges for call origination and higher
termination charges. We were told that the levels of the MNOs two main categories of retail
prices, subscription prices and overall call prices, were broadly consistent with Ramsey pricing.
2.214. Economic theory states that, in competitive markets, prices will be set equal to marginal cost. This is described as cost-reflective pricing and is generally regarded as the method
by which overall economic efficiency is best achieved. However, economic theory recognizes
that, where there are fixed or common costs to be recovered, setting price equal to marginal
cost will leave firms making losses. To cover these costs, prices need to rise above marginal
costs and one approach is to require less price-sensitive customers, or products for which
demand is less elastic, to bear a greater proportion of the common costs, and more price-sensitive customers, or products for which demand is more elastic, to bear a smaller proportion of
those costs. This structure of pricing is sometimes referred to as Ramsey pricing, and it is
widely accepted as being conducive to the recovery of costs in a way that minimizes distortions. However, Ramsey pricing may be expected to produce a more efficient outcome than
cost-reflective pricing only if it is applied in conditions where revenue overall is cost-reflective
(that is, where no excess profits are made), and where either demand for some services is more
price-sensitive than for others or where some customers are more price-sensitive than others.
Further, Ramsey pricing may, in some circumstances, lead to outcomes which seem inequitable, or which disadvantage entrants. Ramsey pricing is a structure of pricing and in the mobile
57

sector may occur within a single service or across different services. Thus, Ramsey prices may
occur within call termination, these being set at different levels at different times of day or for
different days of the week. Alternatively, Ramsey prices may be set across different services,
such as access, call origination and call termination.
2.215. If prices in the mobile sector were to be set by Oftel at Ramsey levels, that is, at
levels which reflect customers price sensitivity to different products or services, then the
relative price elasticities of these services would need to be established. Some of the MNOs
produced their own estimates of elasticities for this inquiry and these are discussed in
Chapter 8. We received a considerable volume of econometric and other evidence on this topic,
which is also discussed fully in Chapter 8. Oftel did not produce its own econometric elasticity
estimates, but told us that it had seen no evidence to suggest that any sufficiently robust estimates of demand elasticities existed, and certainly none that could be relied upon. We consider
the questions whether the MNOs currently set Ramsey prices, and whether they would, in the
absence of a charge control on termination charges, set Ramsey prices, later in this chapter (see
paragraphs 2.429 to 2.446).
2.216. The MNOs put it to us that their pricing had to reflect the need fully to recover fixed
and common costs, which were significant in the mobile sector. It would, they said, be inappropriate to use cost-reflective termination charges as the basis for determining whether those
charges were set at levels that operated against the public interest, as this would result in overall
losses. We now turn to a consideration of the MNOs costs, in particular the costs involved in
call termination.

Costs of call termination

Introduction
2.217. Our findings on the market and on the competitive pressures felt at the retail level
do not, by themselves, answer the question whether the termination charges of O2, Vodafone,
Orange and T-Mobile operate against the public interest. There are a number of other matters
we need to consider in deciding this question. In the following section, we address the question
of what the charges for call termination would be if they were cost-reflective. We look first at
the appropriate rate of return for the termination of incoming calls with particular reference to
the cost of capital (see paragraphs 2.218 to 2.243). We then turn to the network costs of call termination (see paragraphs 2.244 to 2.319). We next examine whether certain non-network costs
should also be included in the costs of call termination (see paragraphs 2.320 to 2.333). Finally,
we consider whether any allowance should be made for externalities (see paragraphs 2.334 to
2.386), before reaching our overall conclusions on costs (see paragraph 2.387).

Cost of capital
2.218. The DGT estimated the weighted average cost of capital (WACC), which adds
together the cost of debt and equity finance, each weighted by the proportion of debt and equity
respectively in the MNOs financial structures. He estimated the cost of equity of the MNOs by
using the Capital Asset Pricing Model (CAPM). The CAPM in effect generates a measure of
how much investors need to be rewarded for holding the risky equity of the particular company
rather than of shares generally. The MNOs also adopted this approach, although T-Mobile
made use of a two-factor risk model as well (see paragraphs 7.252 to 7.258). Oftel used
Vodafones cost of capital as a benchmark for the other MNOs, although certain aspects of its
calculation utilized data relating to other operators (see paragraph 7.201). Oftel told us that the
appropriate cost of capital for the purposes of our inquiry was the cost of capital for 2G termination.
58

2.219. As will be seen from Table 7.21, Oftel estimated the MNOs nominal cost of capital
as between 13.3 and 16.9 per cent, averaging 15 per cent, while the MNOs estimated their cost
of capital as ranging from an average of 16.3 per cent to an average of 26.8 per cent (also in
nominal terms). The Competitive Operators Group (COG) estimated a cost of capital for the
MNOs of 9 per cent.
2.220. The cost of capital can be expressed in real terms (that is, after adjusting for inflation) or in nominal terms. Oftel and most of the MNOs estimated the nominal cost of capital.
However, Oftel used a real cost of capital in its forward-looking LRIC model. In this section of
the chapter, we use and discuss both nominal and real rates. We refer to pre-tax rather than
post-tax rates.
2.221. In estimating the cost of capital a degree of judgement is required. This is because
the different components (or inputs) that contribute to the WACC (discussed in the following
paragraphs) are subject to changenot only because of movements in financial markets but
because of evolving work and views by financial and academic analysts on interpreting data
and there can be considerable uncertainty over the appropriate level for some inputs. In the
light of these uncertainties, we have adopted our normal procedure in calculating the cost of
capital, which is to identify, in respect of each of its components, what we consider to be
reasonable ranges, and to take as our base case the mid-point of each component. The MNOs
told us that they disagreed with this approach and said that we should exercise judgement to
decide the value of each component, rather than simply adopting the mid-point in the range in
each case.
2.222. The MNOs told us that they disagreed with many aspects of our cost of capital estimations, which they said seriously understated their costs of capital.
2.223. We now examine each of the elements or inputs in turn. We begin with the risk-free
rate (see paragraphs 2.224 to 2.226), then look at equity risk premium (see paragraphs 2.227 to
2.230), equity beta (see paragraphs 2.231 to 2.234), debt premium (see paragraphs 2.235 to
2.237), gearing and taxation (see paragraph 2.238) and an alternative pricing theory model to
CAPM (see paragraphs 2.239 and 2.240), before setting out our overall findings in paragraphs
2.241 to 2.243.

The risk-free rate

2.224. Oftel and the MNOs agreed that the nominal risk-free rate was 5.2 per cent. We
assessed the real risk-free rate by looking at the redemption yield on index-linked gilts, which
are generally considered as having negligible default risk and inflation risk. Index-linked gilt
yields have been on a downward trend since the early 1990s (as shown by Figure 7.3). At June
2002, the real yield for long-dated (20 years) index-linked gilts was 2.2 per cent while the
yields for medium-term (ten years) and short-term (five years) index-linked gilts were both
about 2.3 per cent. All these figures are below their averages for the whole period (about
3.5 per cent) but above the long-run real return on government securities (about 1.3 per cent
from 1900 to 2000). There appears to be widespread recognition that gilt yields have been
reduced by special factors, for example the introduction of the minimum funding requirement
for pension schemes.
2.225. The MMCs and CCs published reports in the last four years which considered the
cost of capital were those on Cellnet and Vodafone1 and on two water companies.2 In those
reports, the MMC and CC, respectively, based their ranges for the risk-free rate on both recent
1

See footnote 4. Towards the end of our inquiry, the CCs reports on two airports inquiries were published.
CC reports, Mid-Kent Water plc: a report on the references under sections 12 and 14 of the Water Industry Act, August 2000.
Sutton and East Surrey Water plc: a report on the references under sections 12 and 14 of the Water Industry Act, August 2000.
2

59

and longer-term evidence. In the Cellnet and Vodafone inquiry in 1998, a range of 3.5 to
3.8 per cent was used, but this was at an early stage in the decline in yields on index-linked
gilts. In the water inquiries in 2000 the CC took account of more recent evidence of the failure
of yields to recover to their earlier levels and a risk-free rate of 3 per cent was used in both
inquiries. This level for the risk-free rate was above the rates then current for index-linked gilts
of around 2 per cent, but took account of the downward trend in these gilts.
2.226. Bearing all these factors in mind, and reflecting the continuing downward trend in
the underlying rates, we have used a range for the real risk-free rate of 2.5 to 2.75 per cent.
While this is below the range of 2.75 to 3.25 per cent used by the CC in the water inquiries, it is
above the current spot rates, which have recovered slightly to around 2.2 per cent from their
level in 2000 of around 2 per cent. Using a rate of inflation of 2.5 per cent, our range in nominal terms is 5.1 to 5.3 per cent. Both we and the MNOs have made broadly the same estimates
but ours is based on nominal rates, using different inflation factors. Our real rate is 2.6 per cent,
compared with 3 per cent for the main parties.

Equity risk premium

2.227. We turn now to the equity risk premium, which represents the additional return that
investors require to compensate them for the higher risk associated with investing in equities
rather than risk-free securities. As the future returns from equities are uncertain, the main
methods used to estimate the equity risk premium are first, to take historical averages of
realized equity returns over the risk-free rate and second, to use surveys and other evidence of
investors current expectations over the short term.
2.228. O2, Vodafone and Orange all estimated an equity risk premium of 5 per cent, while
T-Mobile suggested a range of 5.6 to 6.1 per cent, its estimate being more heavily influenced
by the long-run average of the equity risk premium. A major disadvantage of using historical
estimates is that they vary markedly depending on the period used, as shown in Table 7.15.
2.229. We also considered forward-looking estimates of the equity risk premium. Dimson,
Marsh and Staunton (see footnote to paragraph 7.213) estimated expected equity risk premia
ranging from 2.4 to 4 per cent. We noted that the CCs 2000 water reports (see footnote to
paragraph 2.231) cite a number of surveys of UK investors, and other similar evidence, which
gave or implied premium levels of 4.5 per cent or less. However, Vodafone and T-Mobile supplied us with details of other surveys, some of which showed higher estimates. We considered
whether a possible indicator of the right level of the equity risk premium might be the longterm rate of return on equities for companies defined benefit pension schemes, which companies are required to publish. Using various assumptions (see paragraph 7.220), these showed
estimates of the equity risk premium ranging from 2 to just under 3 per cent. However, we
noted the objections which Vodafone and T-Mobile put forward to the use of these estimates.
2.230. It is clear that no certainty can attach to any equity risk premium figure and we
believe that a best estimate must draw both on the historical evidence and the evidence of market expectations. Our view is the premium lies in the range 2.5 to 4.5 per cent. The mid-point of
that range, at 3.5 per cent, is below the 4 per cent adopted by the CC in its water reports to
which we have referred, and below the mid-point (4.25 per cent) of the range 3.5 to 5 per cent
used in the earlier Cellnet and Vodafone reports. In its report on the water companies, the CC
noted that its best estimate of 4 per cent for the equity risk premium was somewhat below the
historical average but above current estimates of market expectations. The range 2.5 to 4.5 per
cent would reflect both the continuing downward trend in historical data and recent academic
opinion. Using an inflation rate of 2.5 per cent, our range in nominal terms is 2.6 to 4.6 per
cent.
60

Equity beta
2.231. We now consider the equity beta of the MNOs UK operations. Equity beta
measures the risk of investing in a particular companys shares relative to the average for all
equities. Oftel said (and we agree) that theoretically we should be estimating the beta of 2G call
termination. However, this is not observable. We have therefore measured the beta of the
MNOs UK operations overall and then considered whether any adjustment was necessary to
represent the beta of 2G termination. Table 7.17 sets out the estimates of equity beta made by
Oftel, O2, Vodafone, Orange, T-Mobile and The Brattle Group (consultants commissioned by
Oftel). It will be seen that the range of estimates is wide: at gearing of 10 per cent, Oftels average estimate is 1.22, while T-Mobiles is 2.32. The COG told us that there were strong a priori
reasons to expect the beta for the MNOs 2G business to have declined as the market
approached saturation and as mobile phones became a basic necessity of business and social
life rather than a discretionary purchase, and it estimated beta at 0.58 (see paragraph 7.224).
2.232. As beta is not measurable directly from market data, statistical estimates are made
by regression analysis (in which total returns from holding a particular share are regressed
against total returns from the market portfolio). In carrying out such analyses, it is necessary to
decide first, whether estimates of beta should be based on daily or monthly returns, second,
which time period should be used and third, whether there is any need to adjust betas to take
account of overseas activities of companies or business mix and if so, how this should be done.
2.233. A further consideration was the practicability of assessing betas for the operations
of the MNOs (that is, termination services) that are the subject of our investigation. COG
thought that we should be concerned with estimating betas for the 2G operations of the MNOs.
This view was, however, disputed by the parties. We nevertheless considered whether BT, notwithstanding that it is a fixed line rather than a mobile network operator, might be regarded as a
suitable proxy for the MNOs in estimating beta, as part of its operations are regulated by Oftel.
Oftel estimated a beta for BT ranging from 1.16 (based on 20 per cent gearing) to 1.45 (based
on 40 per cent gearing) (see paragraph 7.236). We also estimated equity beta for UK utilities
based on a National Audit Office report (Pipes and Wires, NAO, 2002). This showed equity
betas of between 0.61 (electricity) and 1.1 (gas) (see Table 7.19).
2.234. The range of beta that we have chosen necessarily involves a degree of judgement.
Taking into account all the uncertainties, we have concluded that the range of beta should be
1 to 1.6. The range is our estimate of the beta of the 2G termination business. It is similar to the
beta of a quoted 2G MNO because on balance we did not find enough evidence indicating how
far the beta of the quoted company should be adjusted to exclude non-2G activities, or 2G
access and origination activities, or overseas activities. The lower end of the range is based on
monthly data and takes account of the fact that the operations with which we are concerned
would be regulated, while the upper end of the range is based on daily data for the UK activities
of the MNOs. To avoid the difficulties caused by overseas ownership, our upper estimate is
based on mmO2 rather than Vodafone, because mmO2, of all the parent companies of the
MNOs, derives the biggest proportion of its turnover from the UK.

Debt premium
2.235. We now turn to the question of debt premium, which is the interest that an MNO
would have to pay over and above the risk-free rate when borrowing. Oftels and the MNOs
percentage estimates of the debt premium are set out in Table 7.20. Again, the range of estimates among the MNOs is wide, ranging from 1.2 to 5.5 per cent. Oftels estimated range is 1.0
to 1.75. Charles Rivers Associates (CRA) told us that a higher rate should be applied to
T-Mobile and Orange, as less well-established MNOs, and a lower rate to Vodafone and O2.
2.236. Based on recent debt premia for water and electricity companies (2007 to 2010),
COG thought that the debt premium should be 0.81 per cent for the MNOs. We thought this
61

estimate was too low, as we believed that the MNOs would be considered as somewhat more
risky than the utility companies.
2.237. Taking all relevant factors into account, our range for the debt premium is 1.0,
based on Oftels low estimate, to 4 per cent (the average of the premium incurred by O2 and
Orange).

Gearing and taxation

2.238. The MNOs provided us with gearing figures ranging from 10 to 30 per cent. Those
of Vodafone, at 10 per cent, and O2 (10 to 13 per cent), were based on their actual financial
positions. We considered it right to give more weight to these actual figures (as opposed to
estimates), and have therefore adopted a figure for gearing of 10 per cent as opposed to
choosing the mid-point between 10 and 30 per cent. So far as tax is concerned, we have
adopted the standard 30 per cent rate of corporate tax.

Alternative to CAPM
2.239. T-Mobile argued that the CAPM was not empirically robust and did not provide a
reliable estimate of the cost of capital, particularly when used for forecasting purposes or for an
assessment of actual outcomes, because of its lack of predictive power. A more appropriate
model, in T-Mobiles view, was a two-factor risk model used by its consultants, CRA. This
model added a so-called value component to the CAPM and was in CRAs view accordingly
more appropriate for a company such as T-Mobile which was susceptible to distress and
likely to perform less well in recession.
2.240. As T-Mobile is not listed on any stock market, CRA estimated the impact of the
value component using mmO2 as a proxy for T-Mobile; it said that mmO2 was likely to be a
more realistic comparator for T-Mobile than Vodafone because mmO2 is a quoted MNO which
derives the biggest proportion of its turnover from the UK. CRA estimated that mmO2 had a
post-tax cost of equity of between 1.2 and 2 per cent a year higher than the estimate from the
CAPM. We considered whether we should adopt the CRAs suggested model. We were aware
of the debate that was taking place on the relative merits of CAPM and multi-factor models, but
we concluded that there was insufficient evidence of the superiority and greater reliability of
the two-factor model to cause us to depart from use of the CAPM for the purposes of this
inquiry. Indeed, some experts believe that the two-factor model has no foundation in finance
theory and is merely a statistical model that summarizes the empirical regularities that have
been observed in US stock returns. We noted, moreover, that the CAPM was used by four of
the five main parties to our inquiry and that it is widely used across the private sector, finance
institutions and utility regulators. We therefore adopted the CAPM to estimate the cost of
capital.

Overall findings on the cost of capital

2.241. Orange put it to us that it would be inappropriate to apply a single cost of capital to
all MNOs, as each had different funding and risk profiles. Oftel and Vodafone told us that in
their view the same cost of capital should be used for all the MNOs; Oftel took the view that
this should be based on an MNO whose financing costs were efficient, that is, as low as practically possible. Vodafone believed that the efficient cost of capital did not vary among the
different MNOs and that the CC should not take account of differences that arose from individual MNOs having an inappropriate capital structure. In reaching our view on the appropriate
cost of capital, we have taken account of various estimates of company-specific inputs. We do
not believe that the variation in these inputs is sufficient to justify our using different costs of
capital for different MNOs.
62

2.242. We set out our own ranges of each of the elements that make up the cost of capital
in Table 2.5. We acknowledge that these are much lower than the estimates of either Oftel or
the MNOs. The mid-point of our estimates is just under 14 per cent in nominal terms and
11 per cent in real terms.
TABLE 2.5 Illustrative range of the cost of capital
per cent

Risk-free rate
ERP
Equity beta
Cost of equity
Debt premium
Cost of debt
Gearing
Taxation
Pre-tax WACC
Pre-tax WACC (real)

Low
case

High
Case

5.1
2.6
1
7.6
1
6.1
10
30
10.4
7.7

5.3
4.6
1.6
12.7
4.0
9.3
10
30
17.3
14.4

Source: CC.

2.243. The exact extent to which the appropriate level for the equity risk premium has been
moving downwards in recent years is uncertain and, if market conditions altered, the trend
could change. In view of this uncertainty, we would wish to be cautious over implementing in
full the decline represented by our range of 2.5 to 4.5 per cent (in real terms). We consider that
a degree of smoothing of the downward trend in the equity risk premium would be appropriate,
an approach which would also help to prevent volatility in the short term. In our view, the most
appropriate way of recognizing this factor is not by modifying our judgement of the range for
the equity risk premium, but by an increase of 0.25 per cent in the overall level of the WACC
for the MNOs. We conclude that the cost of capital figure we should use is 11 per cent, to
which we add 0.25 per cent (in real terms), giving a final figure of 11.25 per cent.

The network cost of call termination

2.244. To estimate the cost of call termination, it is necessary either:
(a) to determine the stand-alone cost of operating only a call termination service; or
(b) to determine the overall cost of the mobile network and then to find a way to allocate

part of this cost to the termination service.

2.245. Stand-alone cost includes costs that would otherwise be shared or common across a
larger group of services. The sum of the stand-alone costs of all services exceeds the total cost
of the operator whenever economies of scope may be achieved in providing a larger group of
services. Call termination would not, in practice, be provided in isolation. We did not consider
the stand-alone cost to be the most appropriate way to estimate the cost of this service, because
it would take no account of economies of scope.
2.246. Determining the overall cost of the mobile network can be done using either
bottom-up or top-down cost analysis. Bottom-up models aim to estimate the cost of building an
efficient network from its component parts using engineering, economic and accounting principles and assumptions. Bottom-up models are useful when there is limited data available on
operators actual costs or where one wants to know the costs that a hypothetical efficient
entrant could be expected to incur, but it can be difficult to check whether the underlying
assumptions are realistic. Top-down models are based on the cost of a complete network
derived from existing operators accounts, and therefore provide a more reliable measure of the
63

costs at existing levels of activity and efficiency. It can be easier to confirm the accuracy of
top-down numbers. They do not, however, provide any information on whether existing operators are efficient. Given the advantages and disadvantages of the two approaches, to reach a
reliable estimate of the cost of an efficient mobile network it is useful to compare the results of
bottom-up and top-down analyses.
2.247. Termination of incoming voice calls is only one of a number of services provided
by MNOs, alongside outgoing voice, incoming and outgoing data, and SMS. Most equipment
in a GSM network is used by more than one service; therefore, in order to derive the cost of
incoming calls alone, having determined the overall cost of the mobile network some basis then
has to be found for dividing the cost of the network among the individual services.
2.248. There are two main approaches for determining the cost of an individual service:
(a) A fully allocated cost (FAC) approach, sometimes also referred to as fully distributed

cost, allocates all of the costs between the various services, with shared costs usually
being allocated to services based on the factor that causes the cost to vary; and
(b) LRIC considers the additional cost that the firm incurs in the long run by providing a

service (alternatively, the cost that the firm would avoid in the long run if it decided not
to provide a service), and then allocates the costs that are not incremental to any individual services in an appropriate way.
2.249. The DGT told us that the most appropriate and economically efficient basis for
regulatory charges was LRIC, as this provided a cost base for future pricing that mimicked the
effects of a competitive market.
2.250. A report for the four MNOs by LECG, an economic consulting company, in
February 2001 concluded that LRIC could be used to determine a regulated rate for termination
on a mobile network if it included an appropriate share of the network operators joint and
common costs.
2.251. We agreed with the DGT that LRIC was the appropriate basis for estimating costs
because it identifies costs that are directly caused by a particular service. We also agreed that
the LRIC should be based on a 2G-only network, as the proposed regulation only covers 2G.
Beyond the choice of the basis for estimating costs, it is important to consider the detailed
methodology behind the LRIC model. Depending on how the costing systems are designed, a
LRIC model can be very similar to, or very different from, an FAC model.

Oftels LRIC model

2.252. The DGT used a bottom-up LRIC model that estimated the cost of building what he
called a reasonably efficient 2G mobile network. The DGT told us that he had consulted the
MNOs when developing his LRIC model. An industry working group chaired by Oftel, which
included the MNOs and several fixed-line operators, first met in July 2000 and the group considered a number of versions of the model during 2000 and 2001.
2.253. The MNOs told us that the process of the development of the Oftel model, in
particular their involvement in that process through the Working Group, was not satisfactory.
They told us that concerns over costing principles were ignored or discounted with little feedback from Oftel, and the versions of the models supplied to them were difficult to use with
inadequate documentation, and they were not given an opportunity to review the final version
of the model.
2.254. The DGT told us that, at each stage of development of the model, a draft had been
circulated to the MNOs, and that he had seriously considered comments and had taken these
into account where the arguments and data provided had seemed valid. However, he told us that
64

it was only during our inquiry that the MNOs had made detailed and constructive comments on
the LRIC model. He said that he had made Oftels consultants available to MNO staff to assist
them with explanations of how the model worked and to help deal with any problems running
the model. He said that he had believed it appropriate to issue the final version of the model at
the same time as his proposals, and that there had been only one material change to the model
since the versions previously released to the members of the working group.
2.255. We noted the issues raised by the MNOs in relation to how Oftels LRIC model was
developed. It was apparent to us that the parties had found little common ground at the meetings of the industry working group. However, we decided that our main focus should be on the
output of the process rather than on the problems encountered in the process leading to its
release.
2.256. During our inquiry, in April 2002, the DGT made available a new version of the
LRIC model. The later version of the model excluded data services, and thus modelled a hypothetical voice-only network, whereas the September 2001 model had covered all services.
There were changes to the way that depreciation was calculated which the DGT described as
improvements to the methodology. The model also included corrections to some minor errors
identified by the MNOs. These changes, taken together, resulted in lower costs of terminating
calls in every year. We recognized that it was inconvenient for the MNOs to have to consider
another version of the model, and that in removing the data service from the model it would be
more difficult to compare the overall results of the model to the actual networks. We considered, however, that there was sufficient time for parties to consider the new version and that
it would be possible to uplift the results of the voice-only April LRIC model to make it comparable to the MNOs networks. Because we wanted to take into account all new evidence put
to us, we based our analysis on the April 2002 version of the LRIC model. The cost per minute
of terminating calls from the two versions of Oftels LRIC model is summarized in Table 2.6.
TABLE 2.6

Comparison of Cost per Minute for Terminating Calls in Oftels Sept 01 and April 02
models (LRIC cost including mark-up for common cost) 2001 prices

Pence per minute for average operator with 25% market share in 2000/01 falling to 21.8% by 2005/06
and 20% by 2009/10
Model

2000/
01

2001/
02

2002/
03

2003/
04

2004/
05

2005/
06

Combined 900/1800 MHz

Sept 01
Apr 02
Change

5.95
4.96
0.99

5.51
4.83
0.68

4.85
4.20
0.65

4.38
3.98
0.40

4.18
3.77
0.41

4.00
3.58
0.42

1800 MHz

Sept 01
Apr 02
Change

6.89
6.07
0.82

6.36
6.03
0.33

5.58
5.17
0.41

5.08
4.88
0.20

4.74
4.60
0.14

4.51
4.35
0.16

Source: CC based on data from Oftel.

2.257. In designing the April 2002 Oftel LRIC model, the DGT made important decisions
in five areas. These were:
(a) the length of the time period over which cost behaviour would be considered;
(b) the definition of common costs and how these should be recovered;
(c) the definition of the increment;
(d) the level of efficiency to be assumed; and
(e) the depreciation method to be used.

We reviewed each of these decisions in the light of views submitted by the MNOs.
65

The time period

2.258. The DGT told us that the relevant time period for LRIC was the long run, which he
defined as the period over which all assets are replaced. The DGT told us that using short-run
costs was impractical because of their volatility, because they would be unlikely to promote
efficient decisions by consumers, and because they might not allow the operators to recover
their costs.
2.259. T-Mobile told us that the economic definition of the long run focused on the ability
to change inputs, and did not imply that all costs could be varied in the long run. The only time
when all costs could be altered in most businesses, it said, was prior to entry, when (almost) all
costs could be avoided. In any case, T-Mobile told us that the relevant time period for the
model should be the period of regulation rather than the long run.
2.260. In our view the long run is the appropriate period for considering costs. By long run,
we mean the period over which the MNO has complete flexibility with respect to how it configures its network. Flexibility here is from both an operational viewpoint (that is, the network
reconfiguration should be technically feasible), and from an economic viewpoint (that is,
MNOs should be able to achieve and recover an efficient level of cost, including their cost of
capital). We believe that this definition is broadly consistent with that of the DGT. With regard
to T-Mobiles point that the only time that all costs could be altered was prior to entry, in our
view, the MNOs have considerable flexibility to redesign their networks over a period of
several years. We therefore think it is reasonable to assume that an MNO has complete flexibility with respect to how it configures its network in the long run. We agreed with the DGT
that over a shorter time period costs could be volatile as capacity constraints were reached and
extra equipment was put in place. Selecting any period less than the long run would therefore
be arbitrary and unsatisfactory.

Common network costs

2.261. The DGT told us that the only common costs in a mobile network were site acquisition and lease costs of the minimum number of base station sites needed to provide coverage
across the UK, together with the cost of the network management system. He called this the
cost of minimum coverage presence. He said that these costs should be recovered by equi-proportional mark-up.
2.262. The MNOs told us that the starting point for a consideration of common network
cost was the equipment needed to allow a call to be made anywhere on the network. T-Mobile
told us that the cost of providing this single call was much greater than just the cost of the network management system and site costs, and this demonstrated that fixed and common costs
were substantial. Common network costs, T-Mobile told us, included the cost of building a
national network with the capability to make a single call to or from any location in the network. Vodafone told us that it considered Oftels approach to be theoretically unsound because,
even with very low volumes of traffic, there would be a need for equipment on all base stations.
2.263. In addition to the equipment on the minimum number of base stations needed to
provide coverage across the UK, the MNOs told us that other equipment that was shared
between services over and above the minimum network was also common cost. The total common cost should be recovered, the MNOs told us, using Ramsey pricing (see paragraph 2.214).
2.264. In our view, the fact that equipment is shared between services now does not
necessarily mean that the cost of the equipment is common among services in the long run
because, if one service ceased, the amount of equipment needed couldand wouldbe scaled
down to the level needed to run the other activity. Shared equipment that is deployed on the
sites that are needed for coverage could be defined either as common or as incremental. However, even if such shared equipment was found to be common, such apparently common costs
66

can be allocated across the services in a reasonable manner on the basis of the extent to which
each service makes use of the equipment. In the long run, therefore, any equipment cost should
be treated as variable. We agreed with the DGT that site acquisition and lease costs for the
coverage network, and the network management system, were strictly speaking common costs,
because these costs could not be scaled down in the event that one service ceased.

The increment
2.265. The service that we need to cost is that of terminating calls. This could be achieved
in one or other of two main ways:
(a) define the increment as the terminating calls service and calculate the long-run cost of

the service in isolation; or

(b) define the increment as the traffic on all services and then, as a second step, allocate

some of the cost of the increment to the call termination service.

2.266. The DGT told us that in the long run, and starting from current traffic volumes, it
was not necessary to define the increment as a single service because the choice of the increment should not significantly affect the cost calculation. This was because, even if equipment
was shared between services in the short run, the quantity of equipment required was related to
the demand for particular services. That is, if a service did not exist, less equipment would need
to be deployed.
2.267. The MNOs told us that the choice of increment was important and that it did matter
whether the increment was defined as a single service or total volume. Orange told us that the
increment should be the voice termination service, with the full overlap between the cost of the
hypothetical stand-alone termination only and origination only networks (that is, the difference between the sum of the two separate stand-alone networks and the cost of the combined
origination and termination network) being treated as a common cost. T-Mobile told us that the
increment should be less than the whole service so as to estimate, as closely as was practical,
the marginal cost of the service, with the remainder of costs being allocated according to
Ramsey pricing principles. Vodafone told us that the relevant increment was the termination of
all incoming calls, because that was the service whose price we were seeking to determine. In
order to identify and calculate the LRIC of call termination, we should consider what costs
would be avoided if Vodafone were not to offer an incoming call service at all. Similarly, we
should look at costs which would be avoided if Vodafone were not to offer outgoing calls at all.
Costs that would not be avoided in either of these cases were to be regarded as common costs.
Vodafone argued that, on this basis, the full costs incurred in providing a network with the
capability to make or receive a call anywhere within the coverage area should be treated as
common costs, and should be marked up on top of LRIC, using Ramsey pricing.
2.268. We considered the MNOs arguments on the definition of the increment. As we
stated in paragraph 2.247, we recognize that most equipment in a mobile network is shared
among services. This could lead to a conclusion that, if a service were removed, the change in
the total cost of the network would be small. That might be the case in the short to medium run.
However, for the reasons given in paragraphs 2.258 to 2.260, we are basing our analysis on the
long run and we consider it reasonable to assume that in the long run all equipment costs can be
varied. We therefore agreed with the DGTs view that the choice of whether to define the
increment as all traffic (and then allocate a share of this to terminating calls on a cost causation
basis) or just as the terminating call service was not important.
2.269. Within the Oftel LRIC model, the cost of the increment, being the cost of all traffic,
is allocated to the different services using what the DGT described as routing factors. These are
measures of the relative use that each service makes of each type of network equipment.
67

2.270. With regard to the routing factors, the MNOs challenged the DGTs position that
the cost of the home location register (HLR) should not be allocated to terminating calls. The
cost of the HLR is around 1 per cent of total network cost. Vodafone argued that location updates were needed because it was necessary to locate customers only for the purposes of call
termination. The DGT told us that the cost of location updates was driven exclusively by the
number of active subscribers within a mobile network, rather than by the volume of incoming
calls. Therefore he considered that the most economically efficient way to cover these costs
was by means of a per subscriber charge.
2.271. We noted that HLR updates take place all the time that a handset is switched on,
whether or not any incoming calls are actually received by that handset or any other. In that
sense, the cost of the HLR and updating it is not incremental to the volume of incoming calls.
On the other hand, the purpose of HLR updates is to enable an incoming call to reach the
intended mobile handset more economically than if the whole network had to be paged each
time a call arrived. Therefore incoming calls are also a cost driver. On balance, we concluded
that the DGTs case for excluding the cost from terminating calls was not wholly persuasive: in
the absence of call termination there would be no need for location updates. Hence, the fairer
approach would be to allocate the cost across terminating calls including on-net calls.

The level of efficiency

zReal-world efficiency
2.272. The DGT based his bottom-up model on a reasonably efficient operator. He told
us that this was based on the network investment that a hypothetical new entrant would incur.
New entrants would not have to recreate the design of an existing operators network if that
were less than fully efficient, but in the absence of evidence to the contrary, they could be
expected to suffer practical difficulties in rolling out their networks similar to those that had
been experienced by existing operators.
2.273. We were told by MNOs that the Oftel LRIC model assumed a level of efficiency
that was unrealistic. For example, Vodafone told us that assumptions made in the model did not
accurately reflect the reality of providing mobile services in the UK, and Orange told us that the
optimal network assumed in the model was unachievable in practice.
2.274. We agreed with the DGT that the costs should ideally be based on a reasonably efficient operator. However, we recognized that the Oftel LRIC model was, in a sense, groundbreaking as it was probably the most complex model of its type that had been developed for a
mobile operator anywhere. It has also necessarily been developed without access to all the data
that the DGT would have wished to take account of. There was, therefore, a real risk that the
model had created a hypothetical network that could be unrealistic. Whilst the DGT had not
had access to financial and operating information from the MNOs to enable him to do so, we
were in a position to compare this information with the outputs of the LRIC model and so
identify whether efficiency assumptions appeared unrealistic.
zMarket share
2.275. Although the Oftel LRIC model is capable of calculating costs for operators with
various traffic volumes, the DGT based his recommendations on the costs of an average
operator with a 25 per cent market share of call minutes in 2001 declining to 20 per cent by
2010 with the entrance of the fifth operator, Hutchison 3G. This has important implications
because the costs per minute of the two operators with above-average market shares (Orange
and Vodafone) would be expected to be lower than the average LRIC output cost, and the costs
of the two operators with below-average market shares (O2 and T-Mobile) would be expected
68

to be higher than the average LRIC output cost because the higher the volume, the greater the
potential to handle volume using the sites and equipment needed to provide coverage.1
2.276. T-Mobile told us that not accounting for variations in market share would punish it
for its small size relative to the larger networks.
2.277. In principle, we agreed with T-Mobile that the cost of terminating calls should, in
the short term, take into account the extra cost of an MNO with a market share of total traffic
lower than the average. This should ensure that even a relatively small MNO receives enough
income to finance its termination business. The appropriate cost, in the short term, for an
operator with a lower than average market share is the cost of an efficient operator with that
actual market share. However, over a period of two to three years we think that an MNO with a
lower than average market share has the opportunity to capture at least an average share of the
market. Therefore, by 2006, we would expect there to be no need for any extra cost due to low
market share, and the extent to which any extra cost would be relevant in the earlier years of
any price control would depend on decisions taken on the glide path between current prices and
the cost projection for 2006.
2.278. It would be wrong, however, to penalize an MNO with a greater than average traffic
market share for its success in winning customers. Rather than taking into account market share
only for T-Mobile and O2 and basing our cost calculations for Orange and Vodafone on the cost
of an operator with a 25 per cent market share, we therefore decided that the appropriate cost
for all operators, in the short term, should be based on a 20 per cent market share, being the
approximate share of T-Mobile and O2 in 2002. By 2006, the appropriate cost for all operators
would be based on the DGTs original estimate of the share for that year of an average existing
MNO following the launch of Hutchison 3G, being a 22 per cent market share.
2.279. We considered two alternative ways of calculating the cost of terminating calls at
20 per cent traffic market share:
(a) Assume that the network was designed to support only a 20 per cent market share.

Using the April 2002 Oftel LRIC model, before any adjustments, this increased the ppm
for terminating calls in 2001 relative to an operator with a 25 per cent market share by
0.3p for a combined 900/1800 MHz operator or 0.8p for an 1800 MHz operator, using
the DGTs estimate of real cost of capital at 12.5 per cent. The difference between 0.3p
and 0.8p is due to the higher initial cost of the coverage network for 1800 MHz operators, and hence the greater importance of economies of scale to those operators.
(b) Estimate the cost of a network that could support a 25 per cent market share but, so far,

actually handles only 20 per cent of total market traffic. This would involve taking the
network cost of a 25 per cent operator and dividing by the volume of a 20 per cent
operator. This increases the ppm for terminating calls in 2001 relative to an operator
with a 25 per cent market share by 1.4p for a combined 900/1800 MHz operator and by
1.7p for an 1800 MHz operator at a real cost of capital of 11.25 per cent.
2.280. We found this a difficult decision. The April 2002 Oftel LRIC model assumes that
an MNO with a 20 per cent share of traffic would need a smaller network as compared to an
MNO with a 25 per cent share but, looking at the units of equipment in use by the MNOs, we
did not see such a clear relationship between market share and network size. [  ] had less
equipment and less traffic in 2001 than the other MNOs, but has since then invested heavily in
its network to improve quality. This indicated to us that the Oftel LRIC model might be overstating the degree to which network size varies with traffic volume at volumes below 25 per
cent of total traffic in the UK at present. This may be an area where further LRIC modelling
work is needed. In addition, in order for an operator to be able to maintain or grow its market
1
This economy of scale is more significant at lower traffic volumes; once an MNO has captured between 20 and 25 per cent of
the current total market volume, there are only very limited remaining economies of scale.

69

share, it would be reasonable that that operator would design its network to provide a level of
coverage and quality of service commensurate with its competitors. As such, we do not see the
fact that the operators with smaller market shares have similar amounts of equipment to the
larger operators as necessarily reflecting inefficiency. We concluded, therefore, that the second
approach was more appropriate, as there was a risk that relying on the Oftel LRIC model could
lead to an understatement of the extra cost per minute of an MNO with a 20 per cent market
share.
zQuality of service
2.281. Two MNOs told us that we should take into account their costs of providing a high
quality of service (QoS) on their networks, as they had designed their networks to a higher
specification than that assumed in the Oftel LRIC model. We considered that it would be unfair
to expect callers to mobiles to have to pay for a higher level of quality above a reasonable level,
and we considered that a 2 per cent blocking probability as assumed in the Oftel LRIC model
was suitable. If callers enjoyed a higher QoS, then the MNO could expect to benefit in any case
through longer calls or higher traffic levels.

Economic depreciation
2.282. The depreciation approach selected by the DGT for the LRIC model was economic
depreciation. This matches the cost of equipment to its actual and forecast usage over the long
term. As a consequence, there is relatively little depreciation in years where utilization is low
and relatively high depreciation in years of full equipment utilization. By contrast, most forms
of accounting depreciation are relatively simple, taking the actual price paid for equipment (or
its replacement cost) and dividing by the expected equipment life to reach a depreciation charge
for the year. The timing of cost recovery under economic depreciation varies from that under
accounting depreciation, and between 2001 and 2006 the use of economic depreciation by the
DGT added between 0.62p and 2.4p to the per minute cost of terminating calls. In years prior to
2001, economic depreciation would have resulted in lower ppm costs compared to an equivalent calculation based on accounting straight-line depreciation.
2.283. In our view economic depreciation is the appropriate method to use because it most
accurately matches the costs incurred in order to carry traffic to the periods in which that traffic
is carried.

Concerns over accuracy of the Oftel LRIC model

2.284. During the course of the inquiry we received a great deal of information concerning
Oftels LRIC model, including suggested amendments to that model as well as alternative
models from Orange and Vodafone. We considered carefully each one of the issues raised.
Some appeared to reflect a misunderstanding of the Oftel LRIC model (or to apply to the
September 2001 version but not to the April 2002 version of the Oftel LRIC model), but the
large majority appeared, at face value, to be reasonable insofar as the suggested amendment to
the Oftel LRIC model might make it more relevant to a particular MNOs network. We asked
the DGT to consider all of the proposed amendments, and our staff met Oftel on several occasions to explore how the amendments could be incorporated into a revised version of the
model. In parallel to this process we collected the information which we would need in order to
carry out a comparison between the calculations of the Oftel LRIC model and the MNOs
actual data.
2.285. The DGT responded on a point-by-point basis to the MNOs concerns and proposed
amendments. He argued that most of the proposed amendments were either wrong in principle,
immaterial (because the changes were small or because proposed changes cancelled each other
70

out), or reflected inefficiency in an existing MNO network. Other points, he said, should only
be accepted if they could be demonstrated to improve the overall accuracy of the results. The
DGT told us that the effect of the suggested model amendments, taken together, would lead to
lower costs for incoming calls. He suggested that the MNOs might be cherry picking by
focusing only on changes that would increase costs. He did not, therefore, prepare a revised
LRIC model that was acceptable to the MNOs.
2.286. The MNOs rejected the DGTs responses. Vodafone, for example, told us that it
believed it had carried out a thorough analysis and that Oftel had overestimated some costs and
underestimated others, but that the overall picture was that Oftel had underestimated costs. It
said that the DGTs response was in danger of suffering from systematic bias by recognizing
only some of Vodafones suggested amendments as improvements to the model, and by continuing to err on the side of over-optimization and cost omission.
2.287. It became clear to us that the MNOs and Oftel could not reach agreement on a LRIC
model, not least because the DGT wanted evidence to show that suggested changes would
improve the overall accuracy of the model, while the MNOs would not provide him with
accounting data that might form a basis for such evidence. We considered whether there were
any suitable alternative models that could be used to help determine the cost of terminating
calls for a reasonably efficient operator. Vodafone had submitted an amended version of the
September version of Oftels LRIC model to us, and Orange had submitted its own model.
Both companies argued that their models were more accurate than the unadjusted Oftel LRIC
model. We considered using these models, but we decided that they were less suitable than the
Oftel LRIC model because they were designed to reflect these operators specific individual
network configurations and so would not enable us to compare different operators on a like-forlike basis. At best, these models have represented a single operator only, and using them would
have led to different models for each operator, each developed based on different assumptions
and methodologies. Further, the Oftel LRIC model had the important advantage of providing an
estimate of the cost of an efficient operator to compare with the MNOs costs. We also decided
that attempting to develop our own LRIC model was wholly impractical given that many
experts in the industry had already been working on the model for over two years and had
failed to reach agreement. Instead we decided to start with the April 2002 version of the Oftel
LRIC model, to test whether the results of the model were consistent with top-down FAC estimates and other data on actual network equipment and value, and then to adjust the results of
the Oftel LRIC model if appropriate. In focusing on the outputs of the model, we deal with the
effects of the detailed issues raised by the MNOs in an objective and comprehensive manner.
2.288. The April 2002 version of the Oftel LRIC model was less complex than the
September 2001 version but it was still a large model by any standards. There is a risk of formula errors in large models, over and above any issues regarding the underlying logic in the
model. We took steps to ascertain that the DGTs testing of the model had been significantly
stringent. We satisfied ourselves that it was. We did not test every part of the model ourselves.
However, we did not identify any significant problems in the course of our use of the model.
The small number of calculation errors that were identified by the MNOs (and their specialist
consultants) in their use of the model were not sufficient to persuade us that our use of the
LRIC model was inappropriate.
2.289. Three MNOs told us that our approach was wrong. Vodafone told us that our
approach was superficial, flawed and wholly inadequate and that a thorough review of its
critique of the Oftel LRIC model was needed. T-Mobile told us that our approach did not test
the validity of the Oftel LRIC model, but rather simply produced a broad estimate of the gap
between the model outputs for a particular year and actual figures. Orange expressed its concern at our approach and said that it appeared that, rather than undertaking a thorough, detailed
study of the Oftel LRIC model to gauge its suitability, we had merely performed a high level
review of a limited number of parameters. O2 welcomed our approach of testing the accuracy of
Oftels LRIC model by comparing it with actual data from the operators, and said that the tests
that we had carried out on the model were reasonable.
71

2.290. We consider that basing our conclusions on both a bottom-up LRIC model and topdown FAC costs and operational data is entirely logical and fair. Comparing the two sets of
numbers is the only way of gaining confidence that the numbers that will be relied upon reflect
both a reasonable degree of efficiency and are achievable.

Comparison of outputs of the Oftel LRIC model with MNOs data

2.291. As we said above, in considering the areas of disagreement we decided that it was
important to compare the key results of the Oftel LRIC model with the actual costs of the
operators in order to reach a view on the overall accuracy of the model. The comparisons that
we made were based on September 2001 for units and value of equipment, and the financial
year ending in 2001 for operating costs, to avoid including substantial investment in 3G. The
comparisons were based on the size of the network in terms of overall quantity and value of key
items of equipment (base stations, base station controllers and mobile switching centres), and
overall operating costs. We then went on to examine the differences between the costs of combined 900/1800 MHz operators and 1800 MHz operators, and the cost trends beyond 2001.
2.292. We also compared the FAC ppm for terminating calls with the LRIC equivalent as a
check that the LRIC outputs were broadly accurate. We asked the MNOs to calculate their cost
per incoming voice call minute for their financial years ending in March or December 2001
based on their published financial results. We adjusted these calculations to ensure as far as
possible that they were comparable with the Oftel LRIC model, for example by excluding nonnetwork costs and 3G costs. The aim of this was to achieve a like-for-like comparison; we consider non-network costs later (see paragraphs 2.320 to 2.333).
2.293. Before making comparisons between the outputs of the Oftel LRIC model and the
data from the MNOs, it was necessary to take account of two important differences between the
two sets of data.
2.294. First, when looking at the size of the network the April 2002 Oftel LRIC model considered a hypothetical voice-only network whereas the information submitted by the MNOs
was for their total network and included both voice and data services. It was clear there was a
fairly wide possible range of values for the uplift needed to reconcile the voice-only network in
Oftels LRIC model to a real voice and data network. MNOs estimates of the proportion of cell
site capacity that was dedicated to data in 2001 varied from zero to 14 per cent. The operator
that had no capacity dedicated to data did, however, tell us that it believed around 5 per cent of
its network cost should be allocated to data. The DGT told us that the data uplift should be
much higher than 14 per cent to take into account the need to deploy equipment in advance of
the expected future growth in data traffic.
2.295. We noted that the data uplift assumption was critical to the comparison of top-down
FAC costs to the costs calculated by the April 2002 voice-only Oftel LRIC model. A higher
uplift factor would explain a greater difference between the size of the network that was estimated by the model and the MNOs actual networks. (This would, in turn, reduce the discrepancy between the output of the Oftel LRIC and the MNOs actual equipment, and any
subsequent adjustment that we might make to the outputs of the Oftel LRIC model.) We considered that the more reliable basis for the uplift was the operating data from the MNOs, rather
than the DGTs estimate, which was based on what had turned out to be a very high forecast of
data traffic. It was possible that the MNOs had originally invested in equipment with a view to
using it for data traffic, but had since decided to use it for voice, but we found no reliable basis
for resting our data uplift on anything other than the existing network configurations. We also
decided that the impact of SMS on equipment requirements was minimal. We decided to base
our comparisons on a data uplift factor of 5 per cent. In the light of a range of estimates of
equipment that was dedicated to data, we considered that taking a figure slightly below the
average was reasonable because it would be unlikely to lead to an eventual understatement of
the ppm of terminating calls.
72

2.296. Second (see paragraph 2.299), the Oftel LRIC model used economic depreciation
whereas the MNOs FAC estimates, being based on their financial results, used straight-line
accounting depreciation. The differences between historical cost accounting depreciation and
economic depreciation were calculated using an adapted version of the Oftel LRIC model, and
based on this we added 0.62p and 2.16p to the FAC ppm cost of terminating calls for combined
900/1800 MHz operators and 1800 MHz operators respectively. Orange told us that this
economic depreciation adjustment was understated, and the DGT told us that this adjustment
might have been overstated. We considered that the adjustment was reasonable, because it was
based on a modelled network that, at least up to 2000/01, was broadly in line with the actual
size of the two MNOs for which the DGT had been able to use publicly available data. Further,
we noted that the effect of economic depreciation was particularly high in the year in which we
made our comparison and this made it very unlikely that the adjustment had been understated.
2.297. Having identified the adjustments that could be made to ensure that the outputs of
the Oftel LRIC model were broadly comparable to the information from the MNOs, we then
reviewed the differences between the two sets of ppm results and the factors that could explain
these differences. The areas that we considered were:
(a) whether the Oftel LRIC model might have incorrectly estimated the quantity and value

of equipment needed to operate a network;

(b) whether the Oftel LRIC model might have incorrectly estimated the annual cost of run-

ning a network; and

(c) whether the Oftel LRIC model might have incorrectly estimated the cost differential

between combined 900/1800 MHz operators and 1800 MHz operators.

Quantity and value of equipment

2.298. On the quantity and value of equipment needed to operate a network, our tests
showed that the Oftel LRIC model had underestimated cell site capacity by 20.1 per cent, transceivers by 24.0 per cent and number of cell sites by 12.4 per cent. We also compared the total
gross historical cost asset values of the MNOs with those suggested by Oftels LRIC and this
also showed an understatement of equipment of around 24.1 per cent, or 645 million per
operator in 2001/02. We calculated that correcting this understatement of quantity and value of
equipment would add around 0.87p or 1.01p to the LRIC cost per minute of terminating calls
for a combined 900/1800 MHz operator and an 1800 MHz operator respectively at 11.25 per
cent real cost of capital.
2.299. Only one MNO, [  ], had quantities of equipment lower than that shown in the
LRIC model. This MNO argued that using its data from September 2001 was misleading
because since then it had invested heavily in extra equipment in order to improve its quality of
service.

Operating cost
2.300. On the operating cost of running a network, we found that the Oftel LRIC model
overstated operating costs for 2001 by 13.1 per cent. Adjusting for this would deduct around
0.24p from the LRIC cost per minute of terminating calls for combined 900/1800 MHz
operators, and 0.32p for 1800 MHz operators.

Combined 900/1800 MHz vs 1800 MHz costs

2.301. The Oftel LRIC model calculated lower costs for a combined 900/1800 MHz
operator (as exemplified by O2 and Vodafone) compared with an 1800 MHz operator (as
73

exemplified by Orange and T-Mobile). We reviewed whether the evidence from the MNOs
supported the differences between the costs of the two types of network. We concluded that the
two types of network were broadly similar in terms of both the quantity and gross book value of
network equipment by September 2001. However, under economic depreciation, because the
1800 MHz operators had lower utilization than combined operators in earlier years, they need
to recover more cost in later years when volumes are higher.
2.302. The DGT told us that the main driver of the difference in LRIC costs between the
two types of network was not the different amounts of equipment in use in 2001 but rather the
low utilization (that is, volume of traffic carried compared with equipment in place) achieved
by 1800 MHz operators in previous years. The DGT told us that Orange had more cell sites in
place between 1996 and 1999 than Vodafone but less traffic for those years, based on publicly
available data for these companies. He said that the available data strongly supported his view
that 1800 MHz operators had experienced lower equipment utilization, and that this explained
the higher costs in the years to 2006 of the 1800 MHz operators under economic depreciation.
2.303. The 1800 MHz operators agreed with the DGT that it was important to consider the
higher costs they had experienced in earlier years. T-Mobile told us that early deployment of
1800 MHz technology had resulted in higher network equipment costs and more expensive
handsets. It also said that 1800 MHz operators had suffered significant disadvantages in terms
of their ability to offer roaming services. As a result, T-Mobile said, 1800 MHz operators had
enjoyed lower earnings to repay their initial investment. Orange told us that the cost differential
between the two types of network equipment had fallen in recent years, and that the prices of
new equipment were now very similar. However, Orange said that the 1800 MHz operators had
an installed base of more expensive equipment that should lead to higher depreciation charges
compared with the combined operators.
2.304. O2 told us that it considered that the Oftel LRIC model overstated the difference
between combined 900/1800 MHz operators and 1800 MHz operators. In addition, it did not
agree that differing profiles of economic depreciation that were based on cost differences in the
past was a justification for maintaining a cost difference between the two types of network in
the future. O2 said that the impact of moving to economic depreciation would require extensive
backwards reconciliation in order to identify the degree of cost recovery that each operator had
already achieved. We should also take into account, O2 told us, the higher prices for termination that the 1800 MHz operators had charged over the previous regulatory period. Vodafone
told us that its amended version of the Oftel LRIC model showed a much smaller difference
between the costs of the two networks than the DGTs version, with costs very similar at
current volumes.
2.305. Although there is some controversy over the exact advantages and disadvantages of
combined 900/1800 MHz networks as compared to 1800 MHz networks, for the technical
reasons described in Chapter 3, we believe that the 1800 MHz operators faced cost disadvantages in earlier years due to higher equipment prices and lower equipment utilization (although
the costs of the two types of operators had more or less converged by 2001). As described in
paragraph 2.289, we chose economic depreciation because it most accurately matches the costs
incurred in order to carry traffic to the periods in which the traffic is carried. Hence, we agree
that an average 1800 MHz MNOs costs can be expected to be higher than a combined
900/1800 MHz MNOs costs for the years to 2006.
2.306. We did not accept O2s view that the economic depreciation calculations should
take into account the fact that the 1800 MHz operators had been able to charge higher termination charges in the past few years. We need to ensure that future revenues compensate the
operators for their relevant costs, calculated using economic depreciation. Historic levels of
revenue and return are not relevant to this calculation.
2.307. It was clear that the two types of network had a similar amount of equipment by
2001, whereas the Oftel LRIC model assumed that some differences still existed in that year.
74

The effect of increasing the number of cell sites for the combined 900/1800 MHz operators in
2001 to the same level as the 1800 MHz operators added 0.2p to the outputs of the Oftel LRIC
model. As we had already adjusted the outputs of the Oftel LRIC model to reflect the amount
of equipment needed by the average MNO in 2001, we decided that an upward adjustment of
0.1p for combined 900/1800 MHz operators, and a downward adjustment of 0.1p for
1800 MHz operators in each year was appropriate.

Comparison with FAC estimates for 2001

2.308. Having made the relevant adjustments to the Oftel LRIC model costs and to the topdown FAC cost estimates, the result of our comparison is shown in Table 2.7.
TABLE 2.7

Comparison of incoming calls network cost per minute 2001 at 11.25 per cent real
cost of capital
ppm
Simple
Average
Average
average
900/1800 1800 MHz

Network cost
Cost of capital
FAC
Economic depreciation adjustment
Adjusted FAC

3.8
1.4
5.3
1.4
6.7

Comparison with LRIC results for incoming minutes

Oftel LRIC outputs 2000/01
Add: Market share adjustment (see paragraph 2.286)
Add: Equipment adjustment (see paragraph 2.304)
Deduct: Operating cost adjustment (see paragraph 2.306)
Add/deduct: 900 MHz adjustment (see paragraph 2.313)
Adjusted Oftel LRIC
Percentage difference between adjusted FAC and adjusted LRIC
Percentage difference between adjusted FAC and adjusted LRIC
excluding the market share adjustment

5.6
1.5
0.9
0.3
0.0
7.8

5.0
1.4
0.9
0.2
0.1
7.1

6.2
1.7
1.0
0.3
0.1
8.4

15

7

Source: CC based on information from the MNOs.

Note: Totals may not add due to rounding.

2.309. Table 2.7 shows that the average MNO had an FAC cost per minute for incoming
calls in 2001/02 that was 15 per cent lower than the adjusted Oftel LRIC model after our market share adjustment and at 11.25 per cent real cost of capital. All the MNOs FAC costs were,
on this basis, lower than the adjusted LRIC, with the smallest difference being 3 per cent.
Before the market share adjustment, the average MNOs FAC was 7 per cent above the
adjusted LRIC. One operators cost was 3 per cent below adjusted LRIC and three were above.
Details of these results are shown in Table 7.8. We regard the lowest of the differences between
FAC and adjusted LRIC as being well within the range of variations that might be caused by
slight differences in methods of calculation, timing, measurement error or approximations in
the calculations.
2.310. From our comparisons between the results of Oftels LRIC model and the MNOs
FAC estimates, we are satisfied that the Oftel LRIC model outputs for 2001/02, as adjusted in
the ways that we have described, form a suitable starting point for estimating costs in the period
from 2002/03 to 2005/06. We believe there is very little risk that the adjusted results underestimate the cost of a reasonably efficient operator. If anything, the adjusted LRIC result for
2001/02 may overstate costs, first, because of the low data uplift factor that we used (if we had
used a higher data uplift, we would have needed a lower adjustment to the LRIC outputs, and
this would have given a lower adjusted LRIC result); second, because of the market share
75

adjustment that has been applied to all four MNOs; and third, because of the relatively large
economic depreciation adjustment in our base year as compared to other years, which the DGT
told us could have been overstated.

Cost trends 2002 to 2006

2.311. Having decided that the Oftel LRIC model results in 2001/02 should be adjusted to
reflect more closely the actual results of the MNOs, we turned to consider the estimates of costs
for terminating calls in the following years up to 2006. We found that the most significant factor affecting the estimates was the DGTs expectation that the cost of new equipment would
continue to decrease. The forecast changes in traffic volumes, which for a single MNO were
small due to the market growth being offset by the introduction of Hutchison 3G, had very little
impact on the cost of calls. We therefore decided to focus on whether the equipment cost trends
forecast by the DGT appeared to be reasonable.
2.312. Vodafone said that, if the Oftel LRIC model had been unable to replicate what had
already happened, it was astonishing that we believed that the model would be able to predict
what was still unknown. We considered this view but decided that, provided we had confidence
in an adjusted cost for 2001, and were in a position to assess why the cost changed in the years
after 2001, there was no reason not to use the Oftel LRIC model as our starting point for
reviewing the cost trends between 2002 and 2006.
2.313. The outputs of the Oftel LRIC model from 2002 onwards depend, therefore, on the
assumptions made about decreasing prices of equipment in the future. For most types of network equipment in the LRIC model, unit costs were assumed to decrease by 10 per cent a year,
whereas the cost of site acquisition and preparation was held constant, and site rental and lease
costs were assumed to [

] a year in real terms.
2.314. Vodafone told us that some costs were escalating, including site rentals and infrastructure support costs. Vodafone also provided details of its own forecast of equipment price
trends for the equipment in the Oftel LRIC model. T-Mobile also told us that its costs might
increase rather than fall, due to the need to improve network quality and to roll out 3G. The
DGT told us that the cost trend in the Oftel LRIC model for 2001/02 to 2005/06 of 7.2 per
cent a year was within normal bounds of forecasting sensitivity when compared to 5.7 per cent
a year which he calculated using Vodafones suggestions for modern equivalent asset (MEA)
prices for the period until 2010. The DGT told us that this suggested to him that the forecast of
7.2 per cent was reasonable.
2.315. We considered that the DGTs forecast, which reflected general industry trends for
a reasonably efficient operator without being specific to the equipment that had been selected
by any particular MNO, formed a reasonable basis for our view of costs looking forward. Much
of the recent increase in site costs, we believe, is due to the high number of sites needed for 3G
and in our view the increasing site rental and lease costs in the Oftel LRIC model adequately
reflect more general property cost trends. As the decreasing trend in equipment prices is the
most significant driver of the cost trend of terminating calls in Oftels April 2002 LRIC model,
we therefore concluded that this trend appeared reasonable.
2.316. Vodafone also argued that cost trends of equipment should continue beyond 2010,
rather than being assumed to be flat from 2010. The DGT told us that Vodafones suggestion
that equipment prices would continue to decline beyond 2010 was less plausible than his
assumption that prices would fall until 2010 and then would remain constant in real terms.
2.317. Clearly, equipment prices beyond 2010 are difficult to forecast, but given that by
2010 the forecast will have been for up to 20 straight year-on-year cost reductions, we believed
it more reasonable and conservative to assume that equipment prices would remain constant in
real terms rather than carry on falling.
76

2.318. The MNOs expressed general concerns about the use of volume forecasts in the
LRIC model. For the existing operators, annual traffic volume is assumed to increase by less
than 5 per cent a year in the period to 2006, because most of the growth in voice calls overall is
assumed to be offset by a decreasing market share for each of the existing MNOs as Hutchison
3G enters. We think this is a reasonable forecast, although we recognize that these matters are
very uncertain.
2.319. Starting with the result for 2001, we estimated the cost trend from 2002 to 2006. As
shown in Table 2.8, we first adjusted the 2001/02 Oftel LRIC cost of terminating calls for the
equipment understatement (see paragraph 2.298) and the operating cost overstatement (see
paragraph 2.300). We projected this total forwards using the cost trend in the April 2002 Oftel
LRIC model. We then made our 900 MHz adjustment (see paragraph 2.307) that was fixed
across all years. Finally, we multiplied the total by the market share adjustment (see paragraph 2.280) that was based on the market share of one MNO as set out in the Oftel LRIC
model, divided by the market share of one MNO as defined by us.
TABLE 2.8 Derivation of CC-adjusted LRIC of terminating calls by year at 11.25 per cent cost of
capital at 2000/01 prices
ppm
Year ending
Mar 02
Combined April 02 Oftel LRIC
900/
Equipment adjustment
1800 MHz (31% 0.57 Oftel LRIC)
Operating cost adjustment
(11.0% 0.43)
900 MHz adjustment
Market share adjustment
CC-adjusted LRIC
1800 MHz April 02 Oftel LRIC
Equipment adjustment
(31% 0.53 Oftel LRIC)
Operating cost adjustment
(11.0% 0.47)
900 MHz adjustment
Market share adjustment
CC-adjusted LRIC

Mar 03

Mar 04

Mar 05

Mar 06

5.0

4.0

3.8

3.6

3.4

0.9

0.2
5.6
0.1
5.7
1.4
7.1

4.5
0.1
4.6
0.8
5.4

4.3
0.1
4.4
0.4
4.8

4.1
0.1
4.2
0.2
4.3

3.9
0.1
4.0
0.0
4.0

6.2

4.9

4.6

4.4

4.1

1.0

0.3
6.9
0.1
6.8
1.7
8.4

5.4
0.1
5.3
0.9
6.3

5.1
0.1
5.0
0.5
5.5

4.9
0.1
4.8
0.2
5.0

4.6
0.1
4.5
0.0
4.5
per cent

Memo

Market share of one MNOOftel

LRIC model
Market share of one MNOCC

24.9

24.1

23.1

22.4

21.8

20.0

20.5

21.0

21.5

21.8

Source: CC based on data from MNOs and Oftel.

Non-network costs
2.320. In order to reach the total cost of the incoming calls service it is necessary to add
any appropriate share of non-network costs to the network cost that has been calculated. Nonnetwork costs are the costs of all activities that are not directly associated with enabling calls to
be made. The main types of cost that we considered in this category were incurred for the purposes of:
77

(a) acquiring customers, including a share of the costs of the MNOs own high-street shops

and telephone sales centres, commissions paid to third party retailers, and discounts on
handsets provided to new customers;
(b) retaining customers, including a share of the costs of the MNOs own high-street stops

and telephone sales centres, commissions paid to third party retailers, and discounts on
handsets provided as upgrades to existing customers;
(c) providing services to existing customers, typically including call centres to deal with

any questions or problems that the subscriber may experience;

(d) charging customers for calls, including billing, credit control, sales ledger and debt

collection for customers on contracts;

(e) encouraging higher call levels, including advertising, product development and market-

ing new services; and

(f) managing the business, including head office support functions such as human

resources, building costs and rent, and general IT costs, which can be grouped together
as administration costs.
2.321. We grouped the various types of non-network costs into three broad categories so as
to provide the information necessary for our purposes on a reasonably comparable basis:
(a)

Customer acquisition, retention and service costs (CARS)comprising advertising

and marketing, handset costs, discounts and incentives; customer care; billing; and bad
debts.

(b)

Administration coststo include general overheads, which MNOs could not allocate
more directly to cost categories such as network, or CARS.

(c)

Other costs requiring separate consideration, such as 3G amortization costs.

2.322. Table 2.9 summarizes the non-network costs in the three categories noted above,
based on the average across the four MNOs (details are shown in the tables in Appendices 7.2
to 7.5). The CARS costs shown are before taking account of offsetting revenue, such as the
proceeds of handset sales and any part of periodic subscriptions from contract customers. Cost
of capital on non-network assets is not shown, because we found that the MNOs had negative
working capital which was roughly equal to the level of their non-network assets.
TABLE 2.9 The four MNOs: summary of non-network costs for 2001*
million
Simple
average
CARS costs
Administration costs
Other
Total

1,276
159
61
1,496

Source: CC based on information from the MNOs.

*Years to 31 March 2001 for Vodafone and O2, and years to 31 December 2001 for Orange and
T-Mobile. Appendix 7.15 shows the full table with confidential information from the MNOs.
Before taking account of offsetting revenue.
Comprises costs that were irrelevant to a consideration of the costs of incoming calls. These do not
include 3G amortization costs, which were identified, as appropriate, in Appendices 7.2 to 7.5 for the
respective MNOs.

78

2.323. On this basis, non-network costs were between 51 per cent and 66 per cent of the
total accounting costs of MNOs and between 64 per cent and 77 per cent of total cost less interconnect and roaming costs.
2.324. None of these costs was clearly incremental to terminating calls. In deciding which,
if any, non-network costs should be allocated to the cost of terminating calls, we therefore considered the following criteria:
(a) Is the cost common to two or more services in the long run? If so, on what basis should

it be allocated to services?
(b) If not common, should it be allocated to terminating calls for either of the following

reasons:
(i)

callers to mobiles cause the MNOs to incur the cost; or

(ii)

callers to mobiles benefit from the cost incurred by MNOs?

2.325. The MNOs argued that most non-network costs should be recovered in part from
terminating calls because they were common, although in many cases they also benefited
callers to mobiles. O2, T-Mobile and Orange all said that most non-network costs should be
treated as common and recovered on Ramsey principles from all services. Vodafone told us that
customer care and billing should be treated as fixed common costs and recovered from
incoming calls on Ramsey principles. Vodafone considered that the customer acquisition and
retention costs were specific or incremental to acquiring and retaining subscribers. Further
details of the MNOs views are set out in paragraph 7.190.
2.326. The DGT told us that he did not believe that any customer acquisition costs were
incremental to the terminating calls service. The operators chose to acquire customers, he said,
and the fact that calls were then made to those customers did not result in any additional customer acquisition costs. The number of subscribers, the DGT told us, drove customer acquisition costs. Similarly, other non-network costs including marketing and advertising costs and
customer service were caused by services other than call termination.

zCommon cost
2.327. On the question of whether non-network costs should be considered a common cost
in the long run, we looked first at customer acquisition and retention costs. In our view, the
level of customer acquisition and retention costs was closely linked to the expected revenue
that the subscriber would generate: more money is spent attracting and retaining those
customers that generate the greatest revenue (see paragraph 7.154 and 7.155). Although we
received no evidence that more money was spent attracting customers who received the most
calls, we would accept that there is probably some correlation between the intensity of call
making and that of call receiving. However, if, in the long run, expected call revenues from any
service decreased then MNOs would be expected to scale back their expenditure. Therefore, we
did not consider that customer acquisition and retention expenditure was common to termination and other services in the long run.
2.328. We then examined whether customer service costs could be considered as a common cost in the long run, including the cost of resolving problems, billing, bad debts, and distribution of prepay cards. We decided that a very small element of this cost could be described
79

as common across traffic services (specifically, the customer call centre to the extent that staff
help to resolve technical problems that affect incoming calls). However, most customer service
costs varied with either traffic volume or number of subscribers, and so were not common.
2.329. Finally, we looked at administration overhead. This included the costs of head
office functions, including corporate IT, human resources, and property costs. We decided that
this was a common cost that should be allocated across all areas of the business, including network functions and customer acquisition, retention and service functions. On this basis, and
assuming a reasonably efficient administration overhead of 130 million a year based on the
average of three MNOs (we excluded one MNOs cost figure from our average because it was
significantly higher than the other three, suggesting a difference in how costs had been allocated or an inefficiency), we calculated an addition to the LRIC ppm of terminating calls of
0.3p in each year. To reach this figure we allocated the administration overhead across the
different areas of the MNOs business in proportion to the direct cost of each area, because we
considered the purpose of the administration overhead to be to support all areas of the business.
The element attributed to the network was then allocated by traffic volumes to individual services, and the element attributed to CARS was then considered alongside other CARS costs as
described below.

zCaller causes cost

2.330. We then considered the extent to which non-network costs were caused by callers. It
was apparent that little, if any, non-network cost was directly caused by callers to mobiles.
Callers to mobiles might contact a customer service centre of a mobile network if they experienced a problem, but we were told that this was unusual. Callers cannot be expected to, and
usually do not, know the telephone number of other networks service centres.

zCaller benefits
2.331. Finally, we considered the extent to which non-network costs led to benefits to
callers. To the extent that customer acquisition and retention costs led to a mobile user either
changing network or staying on a particular network for longer than they might otherwise have
done, we saw little if any direct benefit to the caller. The caller from a fixed line is unlikely to
know what network the receiving party is on before or after switching network, nor would the
caller care if the termination rates were the same. For a caller from a mobile, in some cases
when the receiving party has switched networks the caller might benefit from a cheaper on-net
call, but it is equally likely that the caller would suffer because a previously on-net call could
have become an off-net call. For many off-net callers (those subscribing to networks other than
the winner and the loser involved in the switch) the switch would make no difference. To the
extent that customer acquisition costs led to a new mobile phone subscriber (a small minority of
customer acquisitionssee paragraph 2.195), and thus to a greater volume of calls, we believe
that this is better captured by means of an externality adjustment.
2.332. We concluded that it would not be appropriate to add any non-network cost to LRIC
on the basis that callers either caused or benefited from the expenditure, over and above the
non-network cost that we considered common.
2.333. We separately considered the question of whether handset subsidies to encourage
the introduction of new technologies may have led to lower network expenditures. Most of this
investment in handsets had taken place by 2001 and needs to be considered against the investment in handset subsidies that would have been given regardless of the operational benefits
(which, we have seen, is not relevant to call termination). We decided that it would not be
appropriate to increase the LRIC of call termination to include this expenditure.
80

Externalities
Introduction
2.334. In telecommunications, there are network effects: the more people who join a particular network, the more valuable membership becomes, because people on the network are
able to contact, and be contacted by, more people than they were before. Existing subscribers
benefit when new customers make the decision to become subscribers because there is then a
larger number of people whom they can call and from whom they can receive calls. In the case
of mobile phones, the benefit would mainly relate to the ability to call or be called at times
when this would not otherwise be possible (rather than at all), but the principle remains the
same. For this reason, at least some of the expenditure by MNOs to stimulate growth in the
number of subscribers is also of benefit to existing subscribers of all interconnected networks.
The benefit is an externality because it does not generally enter into the decision-making of the
customer deciding whether or not to become a subscriber at a particular price.
2.335. A number of different types of externality have been identified in mobile networks.
The principal type is the so-called network externality, described in the previous paragraph.
Closely related to the network externality is the option externality, which refers to the benefit
to existing subscribers from having the ability to contact the new subscriber, even if they do not
do so. These types of externality are generally regarded as positive, but negative externalities
also exist, for example, the nuisance caused by mobile phones ringing, and people using mobile
phones, in public places. Other types of externality are described in Chapter 8; however, we are
concentrating here on the network externality and it is not necessary for our purposes to consider the other types, because these were either agreed to be small (for example, the option
externality) or were not clearly related to the volume of subscribers, which is the focus of our
interest. The network externality was also the only kind of externality that we were confident of
being able, broadly, to measure.
2.336. All the MNOs and Oftel told us that call termination charges should include an
allowance to reflect the network externality. This would be consistent with current regulation
established following the MMCs report in 1998, which recommended a 0.5 ppm externality
surcharge.
2.337. The DGT told us that, when he was considering the target average termination
charge in the proposed charge control for the four MNOs, he had added a mark-up of 2 ppm to
take account of network externalities. However, he told us that Oftel had undertaken further
analysis since the start of our inquiry, and he now took the view that 2 ppm was more likely to
overstate than understate the appropriate externality surcharge. Later in our inquiry, the DGT
said that a figure of 0.5 ppm, at most, was more appropriate.
2.338. BT told us that it agreed in principle with the proposition that we had canvassed in
our Remedies Statement, which was that any subsidy justified by economic theory should relate
to the benefit to existing mobile phone users, both fixed and mobile, brought about by marginal
customers being induced thereby to join, or stay on, a network. BT said that it also agreed with
the modelling work carried out on behalf of Oftel, which estimated the externality mark-up for
fixed-to-mobile calls at between 0.06 ppm and 0.25 ppm.
2.339. In the following paragraphs, we begin by considering how the network externality
might be defined (see paragraphs 2.340 and 2.341) and measured (the Rohlfs-Griffin (R-G)
factor) (see paragraphs 2.342 to 2.346). We then address in turn a number of related issues concerning externalities, as follows: internalization (see paragraphs 2.347 to 2.350) and whether
the R-G factor changes with penetration levels (see paragraphs 2.351 to 2.354); funding of the
subsidy (see paragraphs 2.355 to 2.357); the effectiveness of the surcharge in increasing mobile
penetration (see paragraphs 2.358 to 2.363); and the consequences of a reduction in the subsidy
81

(see paragraphs 2.364 to 2.371). Finally, we give our estimate of the externality mark-up that
should be allowed (see paragraphs 2.372 to 2.384) and our conclusions on the externality surcharge (see paragraphs 2.385 and 2.386).

The concept of a network externality

2.340. Central to the notion of a network externality is the relationship between the private
benefits that new subscribers obtain by joining the mobile network on the one hand and the
external benefits which existing fixed and mobile subscribers obtain by virtue of that new subscribers joining, on the other. All new subscribers willing to join are deemed to derive a private benefit from joining at least as great as the private cost of joining. For people who are not
willing to pay the marginal cost of joining, it is assumed that the private benefit to them of
joining is less than the cost of doing so, and so these people do not subscribe. In other words,
for these people, the value they themselves would derive from joining is not worth the price.
2.341. The purpose of an externality surcharge is to provide a subsidy to encourage marginal non-subscribers on to the mobile network and to encourage marginal existing subscribers
to remain there. That society as a whole generally benefits from each new take-up of a mobile
phone, because this enlarges the pool of people who can contact each other and be contacted in
turn, is consistent with section 3(1)(a) of the Act, which imposes a duty on the Secretary of
State and the DGT to secure that there are provided throughout the UK such telecommunication services as satisfy all reasonable demands for them, but regard must be had to costs. In
considering whether an externality surcharge is justified in the context of the allowable call termination costs of the MNOs, the questions for us are:
(a) how the size of any surcharge should properly be measured;
(b) how any surcharge should be funded;
(c) how effective the resulting subsidy would be likely to be in increasing mobile penetra-

tion and maintaining current levels of mobile phone use;

(d) what the benefits and detriments are of a subsidy funded by a surcharge on call termina-

tion; and, in the light of the answer to those questions

(e) whether the benefits of an externality surcharge would outweigh the detriments in public

interest terms.
Finally, we give our estimate of what the externality surcharge should be.

Measurement of the network externality

2.342. We considered how the network externality might be measured; that is, how we
might establish the extent of the external benefit that someone joining the network generates in
relation to their own benefits from joining. The DGT put it to us that the external benefit generated by additional subscribers could be quantified using the so-called R-G factor. The R-G
factor (sometimes also referred to as the gross externality factor) is defined as the ratio of the
marginal social benefit of an additional mobile subscriber to the marginal private benefit. The
marginal social benefit is defined as the sum of (a) the private benefits obtained by an additional subscriber and (b) the external benefits which existing fixed and mobile subscribers
obtain from the addition of that subscriber. The R-G factor is then the ratio of (a) plus (b)
to (a).
2.343. The R-G factor is generally thought of as having a lower bound of one and an upper
bound of 2. Thus, the benefit would not be much above one for a person who valued making
82

and receiving calls but whom existing subscribers did not much value contact with. An upper
limit of two is proposed on the grounds that it is unlikely that existing members of the network
would generally benefit by a greater amount in aggregate than the new subscriber when the
latter joins a network.
2.344. The R-G factor may be estimated either by a priori reasoning or by inference from a
system of demand equations. The MNOs constructed systems of demand equations during our
inquiry and, through their consultants, mounted a number of arguments attacking the robustness
of Rohlfss1 model; these are set out in detail in paragraphs 9.106 to 9.116 and Appendix 9.1.
2.345. We looked carefully at the various methods of modelling the call termination
charge, which were presented to us in connection with estimating the mark-up on fixed and
common costs, and at the effects of various levels of call termination charges on consumers.
(These models are described and analysed in Chapter 9 and Appendix 9.1.) We arranged for
these models to be disclosed to the other main parties and to Oftel, and invited the main parties
views on them. Many submissions were made, exchanged and commented on as the result of
this initiative but no sort of consensus on the models was reached among the parties.
2.346. Dr Rohlfs told us that the value of the gross externality factor implied by the initial
estimates of Frontier Economics (commissioned by Vodafone) was over 7, and that such an
extreme outcome was not credible. It meant, Dr Rohlfs said, that other fixed and mobile
customers derived an implausible six or seven times as much value from communicating with
mobile subscribers as mobile subscribers derived from communicating with them. As a result
of changed assumptions, Frontier Economics subsequently produced an implied R-G factor of
2.7, but Dr Rohlfs told us that that figure was not much more credible.

Internalization
2.347. The R-G factor is reduced if the benefits accruing to others by virtue of someone
joining the network are taken into account in that decision to join. For example, two or more
subscribers might choose to subscribe to mobile networks primarily and explicitly to communicate with each other by mobile phone. Or subscribers might contribute towards the cost of telephone services (both subscriptions and calls) for others with whom they have a substantial
community of interest. Examples are people paying for childrens or parents mobile phone
services; and businesses acquiring mobile phones for their employees. In these cases, the network externalities are said to be to a greater or lesser extent internalized and there is accordingly less justification for a surcharge. The more that the external benefits to others are taken
into account in someones decision to join, the fewer are the external benefits that are not taken
into account, and which it is the purpose of the surcharge and subsidy to stimulate.
2.348. BT put it to us that it was implausible to suppose that many users would not take the
benefits to their family and friends of being contactable into account in their decision whether
or not to buy a mobile phone, and thus doubted that uninternalized externalities could mean an
R-G factor as large as 1.7, as suggested by Oftel.
2.349. The MNOs, however, questioned the levels of internalization. T-Mobile, for
example, argued that the derivation of a low termination mark-up value in Dr Rohlfss work
was driven by unrealistic assumptions about the degree of internalization of network externalities and cost levels in the industry. Lexecon, for Orange, said that the argument underlying
Dr Rohlfss assumption that network externalities benefiting mobile subscribers were largely
internalized by MNOs was both empirically inaccurate and internally inconsistent. These arguments are discussed further in Chapter 8.
1
Dr Jeffrey Rohlfs. An academic economist whose work led to the establishment of the R-G factor and who was engaged by Oftel
as a consultant in relation to this inquiry.

83

2.350. The survey evidence in our view shows that a significant proportion of network
externalities have been internalized. The relevant survey evidence is set out in paragraphs 8.116
and 8.117, and consists partly of research by Oftel, partly of research by Vodafone, and partly
of evidence from our own August 2002 survey (the August 2002 survey results being set out in
Appendix 8.5). Although the evidence as to the extent of internalization is mixed, we believe it
suggests that the R-G factor is below 2. To the extent that this is so, then the appropriate externality allowance would be correspondingly lower. This is also consistent with all the evidence
presented by the MNOs on closed user groups and repeat calling relationships (see paragraphs 2.113 to 2.121).

Whether the R-G factor changes with penetration levels

2.351. We considered whether there was any relationship between the R-G factor and the
extent of mobile phone penetration.
2.352. BT argued that, during the period when the mobile network had been growing, there
had perhaps been justification for an imbalance in the charges paid respectively by fixed and
mobile customers. Given, it said, that the mobile market would be virtually saturated by the end
of the period of the proposed price cap, any such justification had now largely disappeared. For
that reason, BT believed that the externality surcharge represented little more than a tax on
callers from fixed lines to mobiles.
2.353. T-Mobile argued that there was no reason why late joiners to the mobile network
should bring lower benefits to existing subscribers than early joiners. With additional existing
(fixed and mobile) subscribers, both the social benefit and the new subscribers private benefit
from joining should increase, it said, leaving the R-G factor constant.
2.354. In a market in which most people who want to make and receive mobile calls
already have a mobile phone, the value which new subscribers place on calling and being called
and the additional value which existing subscribers will place on calling and being called by
each new marginal subscriber are both likely to be lower than was previously the case, when
subscribers who joined were keen to communicate. This is because new marginal subscribers
will tend on average to make fewer calls to mobiles themselves and generate fewer incoming
calls from others, compared with longer-standing subscribers who make many calls on a
regular basis. Hence, they are likely also to generate little additional benefit to existing subscribers. Overall, although there will inevitably be individual exceptions to the general pattern,
we take the view that the R-G factor is likely at most to have remained fairly constant over time
and may have declined somewhat; at any rate, we see no reason why it should have increased in
recent years.

Funding of the subsidy

2.355. To the extent that the MNOs were to be permitted a mark-up on their allowed costs
of call termination, in order to fund a subsidy to encourage new subscribers on to the mobile
network, we would need to consider how such a subsidy should be funded. The MNOs have
some incentive to subsidize subscriptions, especially for their highest-spending customers; it is
clear that the higher the surcharge, the more the MNOs have the incentive to make subsidies
available.
2.356. Some of the FNOs put it to us that we should take into account the disbenefits as
well as the benefits of a surcharge. Thus BT told us that high termination charges discouraged
calls from fixed telephones and that, as most people now had a mobile phone, the public interest would be better served by encouraging customers to make full use of networks that most
customers had already joined.
84

2.357. To the extent that, as we have seen, it is the FNOs who contribute by far the largest
share of the MNOs net revenues from call termination, then it is the FNOs customers who
bear the greatest part of the funding of the subsidy. As we have already seen, this means that
some fixed-to-mobile calls which would otherwise have been made are not in fact made and all
calls from fixed lines to mobiles cost more than they otherwise would. It is true that customers
of the FNOs obtain some benefit from the expanded network brought about by the externality
surcharge. In our view, however, if a class of consumer is to bear the cost of a subsidy, then
those consumers interests might be prejudiced if they do not benefit from the existence of the
subsidy. In so far as the subsidy does not have its intended effect, for example, because it is
spent on encouraging mobile customers to churn between networks, then the FNOs customers
who bear the cost of it cannot benefit from its existence. In this instance, the sole purpose of the
subsidy (the network externality allowance) is to induce people, who would otherwise not do
so, to become mobile subscribers, to the benefit of others who might call, or receive calls from,
them on the mobile phone they thus acquire.

The effectiveness of a surcharge on call termination in increasing mobile penetration

2.358. As penetration increases, the value of each additional customer to the MNOs is
likely to diminish, because (other than those reaching an age when they begin to have sufficient
use for a mobile, or those new to the UK) these customers tend to be lighter users of phones
and hence to be less profitable. At the same time, potential marginal customers derive less
benefit from mobile ownership than those who have already joined the network, and so require
a higher level of subsidy to induce them to join.
2.359. Some of the surcharge has been competed away in other ways than subsidy, for
example, in sales and marketing spend. Although some of this spend is directed at potential
new subscribers, some of it (and, in terms of the effect, most of itsee paragraphs 2.188 to
2.196 on churn) is also directed at existing subscribers, to induce them to upgrade their handsets, change tariffs or make more calls, or to encourage or discourage churn.
2.360. The MNOs told us that there were severe limits on the extent to which they could
price-discriminate effectively between marginal and non-marginal subscribers. It was not
possible, they said, to identify marginal subscribers individually, so as to target subsidies on
them. Marginal customers did not comprise a homogeneous group with distinguishing characteristics; they represented a significant part of the market and were very diverse. Vodafone said
that it was not possible to introduce new tariffs targeted at the marginal subscriber, which
would not also be available and attractive to the non-marginal subscriber. Commenting on
Vodafones view that, in order to provide subsidies to marginal subscribers, subsidies had to be
offered to all subscribers, Oftel said that the MNOs offered different tariffs appealing to different types of customer. Perfect targeting might not be feasible, but in Oftels view, Vodafone
appeared to be making the untenable argument that absolutely no targeting was possible. That
was, in Oftels view, clearly not the case.
2.361. We are not persuaded that the MNOs could not direct any subsidy specifically at
marginal subscribers. However, the MNOs may not consider it to be in their interests to do so.
Subsidies go, in part, to non-marginal subscribers who do not need them, to persuade them to
stay on the network. While it is true that some part of the subsidy funded by call termination
charges currently finds its way to all subscribers, it remains the case that marginal subscribers,
who by definition are not so profitable for the MNOs, obtain a smaller proportion of that subsidy than non-marginal subscribers.
2.362. There is another reason which tends to weaken the MNOs argument that they cannot target subsidy at marginal subscribers. As BT and Oftel put it to us, both the FE and
DotEcon models have assumed constant per unit pricing. But the reality is that the MNOs have
(as we have seen earlier) a wide variety of pricing schemes which allow price discrimination by
85

virtue of subscribers self-selecting themselves on to alternative tariffs. BT suggested to us that

the more disaggregated the MNOs approach to pricing, and the more non-linear the tariff systems, the lower the optimal mark-up on termination would need to be.
2.363. Consistent with our conclusion that the MNOs already effectively segment the retail
market and price-discriminate in their offers to customers (see paragraph 2.209), we believe
that the MNOs could, if they chose, target individuals who thought it worthwhile to have a
mobile phone but did not use it very much, and who were concerned above all to minimize the
amount they spent on it. The most appropriate package would be one offering a basic inexpensive handset and high call charges. Such a package would be attractive to marginal subscribers
but not to high-volume users. Another possible package could offer an initial tranche of calls at
relatively low prices but successive calls at relatively high prices, again making the package
unattractive to high-volume users. The MNOs can identify and target marginal subscribers
already on their networks (since they know their usage pattern), but it is harder for them to
identify and target people who are not on any network. Whatever the precise position, it would
in our view be disproportionate to allow an externality adjustment if we thought that there was
no ability to target on the MNOs part. Oftel said that the MNOs ability to price discriminate
implied that to some extent the MNOs were able to internalize external benefits which accrued
to the population of mobile subscribers and was a further reason for a reduction in the size of
the surcharge on mobile termination from the 2 ppm it had previously recommended.

Consequences of a reduction in the subsidy

2.364. We considered the likelihood of substantial numbers of marginal existing subscribers leaving the mobile network if the subsidy for the externality surcharge was reduced.
The MNOs argued that reduced levels of subsidy would increase this risk. Vodafone, for
example, told us that, if prepay handset prices rose by 20, as many as 10 million to 15 million
customers could leave the network over a matter of a few years. In Vodafones view, this fall in
subscriber numbers would result in underutilization of the network, an economically inefficient
outcome because efficiency consisted, in its view, in the maximization of the aggregate number
of minutes on telecommunications networks. Orange, too, told us that, while a Ramseyadjusted externality approach would achieve optimal use of the network and of the number of
subscribers, a reduction of the subsidy would lower efficiency and welfare.
2.365. In assessing the likelihood of existing subscribers leaving the network in response to
reductions in handset subsidies, we think that factors such as the high levels of mobile usage,
the utility of mobile ownership, and the perceived disadvantages of giving up a mobile phone
after having used one would militate against subscribers leaving the network if subsidies were
reduced.
2.366. The level of handset subsidy would generally be expected to affect a marginal existing subscribers decision whether or not to continue his subscription only when his handset was
lost, stolen or broken. From our consumer survey (see Appendix 8.5), we estimate the total
number of marginal subscribers to be around 12 million, and would expect them to replace their
handsets in the ordinary way only every four years on average.
2.367. Another important determinant of whether marginal current subscribers would leave
the network if call termination revenue to MNOs was reduced is the willingness of the MNOs
to retain such customers. [  ] told us that MNOs could reintroduce time-limited vouchers for
prepay customers to compensate them for reduced call termination revenues. We note that the
incremental cost to MNOs of maintaining low users on their networks is very small[  ]
told us that it amounted to a few pence a month, which represented the cost of maintaining
them on the HLR.
86

2.368. More fundamentally, and addressing the view that reducing levels of subsidy would
be economically inefficient and would reduce welfare, we believe that there are efficiency
losses associated with subsidization of marginal customers by way of surcharges on mobile call
termination and that these have to be set against the benefits of increased overall mobile
membership. High termination charges (embodying high levels of surcharge) make fixed-tomobile and off-net calls more expensive and tend to result in fewer and shorter calls, thus distorting the volume and pattern of network use. The subsidy also encourages the replacement of
handsets at the expense of network use.
2.369. Finally, we do not accept that, as the MNOs contend, the claimed overall efficiency
brought about by the subsidy at its present level more than outweighs the detriments resulting
from it. These detriments are higher charges for all callers to mobile phones from other
networks, unnecessary upgrading and switching of handsets and of customers between one network and another, fewer and shorter calls to mobiles from other networks and enforced subsidization by FNOs customers of a form of competition which is becoming less and less
effective in bringing more subscribers on to the mobile network.
2.370. Having considered all these matters, we take the view that the MNOs could, if
required to do so, target with a broad degree of accuracy any subsidy funded through a surcharge on call termination to existing marginal subscribers and potential new subscribers. The
evidence, however, suggests that, to the extent that the surcharge is currently being used to subsidize handsets, this is having a limited impact on the recruitment of new, or the retention of
existing, marginal subscribers. Rather, it is either encouraging switching of existing subscribers
between the different mobile networks or simply funding upgrades of handsets for subscribers
who are not, in any event, considering leaving the network and some of whom are high
spenders on mobile calls. This means that the customers of the FNOs, who are ultimately providing the subsidy by funding the surcharge on call termination, are very largely funding an
ineffective subsidy.
2.371. It follows from this that, in principle, any externality mark-up should be either very
small or not allowed at all. On the grounds that a targeted subsidy could have the effect of
bringing marginal customers on to the network and helping to retain marginal customers, we
take the view that there should be an externality mark-up on LRIC but that it should be very
modest.

Estimate of the externality surcharge

2.372. We now turn to consider how we might calculate the amount of the surcharge which
would, if it were used for the purposes of bringing marginal customers on to the network and
helping to retain them there, be effective in generating external benefits. We considered two
alternative methods. The first approach involves the construction of a system of simultaneous
demand equations and calibration using estimates of various price elasticities. The problems
associated with these models are discussed fully in Chapter 9. As we show in that chapter,
many of the results derived from the MNOs models imply an R-G factor of over 2 (in some
cases significantly greater than 2). Most of the models also distribute the mark-up over termination and other services, attributable to the externality, on the basis of Ramsey pricing.
Therefore, given, first, that we regard an R-G factor significantly above 2 as implausible;
second, that we have reservations regarding the estimates of the elasticities used in the models
(which in many of the models are the way in which the externality is encapsulated); and third,
that we do not believe that the MNOs would set their other prices on the basis of Ramsey
pricing if the termination charge alone were regulated at Ramsey levels, we do not view such
models as an effective means of calculating the mark-up over termination attributable to the
externality.
2.373. The second approach involves capping the surcharge at a level that corresponds to
the amount of subsidy which, targeted at marginal customers for whom it would make the
87

difference between joining and not joining a mobile network, brings about at least as much
external benefit as the amount of the subsidy. This approach, too, presents some problems but
we decided to adopt it, because it avoids reliance on elasticity estimates, and adopts a more
plausible R-G factor than the first approach. In doing so, we used various survey and other
published data, including a survey carried out for us by BMRB (see Appendix 8.5). The range
of assumptions and calculations we used, described in paragraphs 8.199 to 8.242 are set out in
Appendix 8.1.
2.374. We estimated that the externality surcharge that would be justified according to the
external benefits it would generatewere the MNOs able to target these customers and were
they to have the incentive to do sowas 0.11 ppm, for a value of the R-G factor of 1.5. (In fact,
we think that 1.5 is likely to be an upper estimate of the R-G factor, but we have decided that
we should err on the side of caution with respect to the MNOs finances by allowing an R-G
factor as high as 1.5.) This (0.11 ppm) is the surcharge required to provide the level of subsidy
sufficient to induce on to the network all those marginal existing and non-subscribers for whom
the external benefits that would thereby be generated exceed this subsidy. We calculated that
this level of subsidy was 23.33 per marginal subscriber (see paragraph 8.208 and
Appendix 8.1 for the calculation). Our survey results suggested that 12 per cent of non-marginal subscribers would purchase their own handset in any event (although they would not be
prevented from receiving the subsidy); on the assumption that marginal subscribers valuations
of network membership are evenly distributed from zero to 70, the 23.33 uniform subsidy
would be sufficient to attract 33 per cent of all remaining marginal subscribers (both current
and non-current) on to the network or induce them to stay on the network. Table 8.14 (see
paragraph 8.210) shows the estimated surcharge under a range of different input assumptions.
2.375. Underlying the calculations were a number of assumptions, set out in Appendix 8.1.
These included an estimate of the mobile saturation level, based on the proportion of those
people aged 12 or over who are currently without a mobile phone (around 16 million) and who
would consider getting one, as estimated on the basis of the results of our first BMRB survey
(see Appendix 6.2). The calculations also included an estimate of the number of customers
already on the network who might be considered as marginal, in that they would need some
inducement to remain on the network were their handset to be lost, stolen or broken; our estimate of the numbers of such customers was made using data from our second BMRB survey
(see Appendix 8.5).
2.376. The calculations also employed an estimate (of four years) for the length of time
that handsets of marginal customers last before being lost, stolen or broken, and an estimate of
the minimum price of a handset (with SIM card). The MNOs objected that the minimum handset price (taken by us to be 69.99) incorporated subsidy which would be put at risk by reduced
termination charges. For example, Orange told us that the commission on its most basic handset
was a little over []. However, we note that the minimum handset price is currently 49.99
(see paragraphs 8.220 to 8.223).
2.377. It should be noted that, although a subsidy of 23.33 would need to be targeted at
marginal customers, it would not require perfect price discrimination (that is, precise targeting
of the subsidy). If perfect targeting were possiblethat is, if each marginal customer paid an
amount exactly equal to his valuation of network membershiponly half the 0.11 ppm surcharge would be requiredthat is, 0.05 ppm.
2.378. The MNOs made a number of criticisms of the methodology we had adopted in
arriving at these estimates, in particular the assumptions that they were able to target marginal
customers and had the incentive to do so, which all the MNOs challenged. Their comments are
set out in paragraphs 8.214 to 8.224. Oftel told us that work commissioned from Dr Rohlfs had
obtained similar levels of surcharge when he had modelled substantial targeting of subsidies.
As Dr Rohlfss and our methodologies were somewhat different, Oftel thought that each provided a useful cross-check on the results of the other.
88

2.379. The calculations in paragraphs 2.380 to 2.382 can be varied to take account of the
amount of targeting of subsidy that might actually be possible. In the case of current subscribers we note that MNOs know the level of spend of all their customers, so could effectively
target the marginal customers individually because their level of spend can be taken as an indicator of their private valuation of mobile subscription. Thus, a customer whose handset was
lost, stolen or broken might require a discount of almost the full price of a handset if his or her
spend was very low; on the other hand, customers with very high spend might require little or
no subsidy. Assuming as before that marginal customers are evenly distributed between 0 and
70 according to their personal valuations of network membership, the average discount that
the MNOs would have to offer is 35. Provision of a 35 (average) discount to the 3 million
current subscribers who would be marginal every year because their handset was lost, stolen or
broken would require a surcharge of 0.34 ppm (see paragraph 8.234).
2.380. So far as marginal current non-subscribers are concerned, we estimated (on the
same basis as before) that a uniform subsidy of 23.33 should be offered (as described in paragraph 2.380) which would add a further 0.03 ppm to the surcharge (see paragraph 8.235). As
before, this would induce 41 per cent of all marginal non-subscribers to join the network.
2.381. As noted in paragraph 2.385, marginal existing subscribers would be offered an
average subsidy of 35. Those marginal current subscribers who would receive a subsidy of
less than 23.33 might be more influenced by the offer to non-subscribers. This effectively
means that the minimum subsidy that could be offered to existing subscribers is 23.33, which
adds 0.04 ppm to the amount which must be raised by the surcharge (this calculation is shown
in paragraph 8.236).
2.382. Thus the total surcharge justified on the basis that some targeting is possible is
0.41 ppm. This is derived from the amounts referred to in the previous three paragraphs as
shown in Table 2.10:
TABLE 2.10 Externality surcharge justified according to the amount of targeted subsidy (ppm)
Subsidy to existing subscribers (see paragraph 2.385)
Subsidy to non-current subscribers (see paragraph 2.386)
Extra subsidy to existing subscribers for whom
Subsidy to non-current subscribers is more attractive (see paragraph 2.387)
Total

0.34
0.03
0.04
0.41

Source: CC.

2.383. Finally, we investigated the sensitivity of this result to changes in some of the parameters used to calculate it. We considered the effect of reducing handset life for marginal current subscribers from four to three and a half years and increasing the price of the cheapest
handset to 75. Changing these assumptions would take the total surcharge, on our estimation,
from 0.41 ppm to 0.5 ppm.
2.384. However, Oftel objected to the amended calculation of the externality surcharge
because it involved directing a subsidy at marginal existing subscribers for whom the marginal
external benefit was less than the marginal resource cost (ie the subsidy). Oftel said that the
correct amount of surcharge was 0.11 ppm as we discussed earlier in paragraph 2.380.

Conclusion on the externality surcharge

2.385. We conclude that the mark-up on the termination charge that should be allowed in
order to take account of externalities is 0.45 ppm, because this lies mid-way between 0.41
and 0.5.
89

2.386. As a cross-check on our externalities calculation, we compared the result with the
amount which the MNOs are currently spending to acquire each new mobile customer. Because
the various components of our calculation of spend on acquiring new customers were necessarily imprecise (for example, the amount which retailers spend on acquiring customers out of
the incentive payments made to them by the MNOs is impossible to ascertain with precision)
we sought to estimate a range rather than a single figure. We calculated that the MNOs were
spending between 0.30 ppm and 0.41 ppm to acquire each customer that was new to the mobile
network altogether (see Appendix 8.4).

Costs of call terminationconclusion

2.387. We now draw together all our findings on costs. Table 2.11 brings together our
estimates of network cost LRIC, non-network common cost, and externalities. These are calculated at our cost of capital of 11.25 per cent. These results show that termination charges are
currently above total costs of call termination by between 2.5p and 3.0p a minute. We note that
if future termination charges were set at either RPI9 or RPI12, they would still be above our
estimate of total costs in each year to 2006.
TABLE 2.11 Summary of cost of call termination (at 200/01 prices)
11.25% cost of capital, ppm
Mar 03

Year ending
Mar 04
Mar 05

Mar 06

Combined
900/1800 MHz

Adjusted LRIC*
Non-network costs
Externalities
Total

5.41
0.30
0.45
6.16

4.83
0.30
0.45
5.58

4.35
0.30
0.45
5.10

3.98
0.30
0.45
4.73

1800 MHz

Adjusted LRIC*
Non-network costs
Externalities
Total

6.26
0.30
0.45
7.01

5.54
0.30
0.45
6.29

4.96
0.30
0.45
5.71

4.51
0.30
0.45
5.26

2002/03

2003/04

2004/05

2005/06

900/1800 MHz
(2000/01 prices)
RPI9
RPI12

8.99
8.99

8.20
7.93

7.48
7.01

6.82
6.59

1800 MHz
(2000/01 prices)
RPI9
RPI12

9.57
9.57

8.73
8.45

7.96
7.46

7.26
6.59

Source: CC.
*As per Table 2.8.
Provided by Vodafone.
CC estimate.

Arguments for and against high termination charges

2.388. We now turn to consider a number of arguments concerning whether high termination charges have beneficial or detrimental effects in terms of the public interest. These arguments were as follows:
(a) Even if the MNOs were, in the absence of a price control, to set high termination

charges in order to subsidize customer acquisition at the retail level, that would not produce adverse effects because, as most people have a mobile phone, whatever they lose in
above-cost termination charges they gain whenever they make a call as a mobile customer, since access is subsidized and some outbound charges are low. We consider this
distributional argument in paragraphs 2.390 to 2.400.
90

(b) The MNOs use of revenue from termination charges to fund (or subsidize) customer

acquisition at the retail level is beneficial to consumers and society at large and its
removal would be detrimental to consumer welfare, because mobile subscribers, who
both value calling and receiving calls and whom others (including fixed-line customers)
value calling and receiving calls from, would leave the network. We address these consumer welfare issues in paragraphs 2.401 and 2.402.
(c) A reduction in termination charges as proposed by the DGT would have a detrimental

impact on other businesses in the mobile sector, for example, retailers, while benefiting
the FNOs; we discuss this possibility in paragraphs 2.403 to 2.412.
(d) A reduction in termination charges as proposed by the DGT would prejudice the MNOs

ability to finance their businesses (see paragraphs 2.413 to 2.416).

(e) A reduction in termination charges as proposed by the DGT would have a detrimental

impact on 3G services (see paragraphs 2.417 to 2.422).

(f) High fixed-to-mobile termination charges distort consumers choices (see para-

graphs 2.423 to 2.427).

2.389. We consider each of these arguments in turn, beginning with the distributional argument.

Distributional argument
2.390. The MNOs advanced the argument that, because most people had a mobile phone,
what they lost in high termination charges they gained in low access and outgoing call charges.
2.391. Oftel said that the distributional issue was one of fairness. It was not fair that callers
to mobiles should unduly subsidize mobile customers. It was quite true that, in some cases,
these were the same people. However, there was a significant number for whom the effect
could be important. Oftel said that, even if the proportion of the population affected adversely
was quite modest, for example, 10 per cent, that was a large number of peoplearound
5 million. Further, in the absence of a price control, the termination charge could be much
higher than at present and that would exacerbate the distributional problem brought about by an
undue subsidy. The people most adversely affected, Oftel believed, were those without a
mobile phoneroughly 15 per cent of the populationas these people called mobile phones
but did not themselves own one. This group had a disproportionate number of elderly and lowincome people in it.
2.392. We obtained from Oftel estimates of the number of people who had a fixed-line
telephone but were without a mobile phone. Oftel told us that (according to results from a survey taken in February 2002) there were some 3 million households, representing some
8 million people, in that position. Of the 8 million, about 2 million made at least one-quarter of
their total fixed calls to mobiles. Oftel told us that its research showed that two-thirds of fixedline-only households earned less than 11,500, compared with just over half of mobile-only
households (although the older retired fixed-line-only households might have more assetsfor
example, ownership of their home).
2.393. O2 put it to us that, with a high UK mobile ownership penetration rate, those receiving incoming calls were predominantly the same as those benefiting from lower call charges.
The vast majority of customers of fixed-line networks were also themselves customers of
mobile networks; and there was a close alignment of the interests of customers who had both a
fixed line and a mobile phone, and customers with only a mobile phone. Individuals with both
types of phone who used them in approximately equal measure could not be said to be subsidizing themselves. There were few losers: the number of fixed-line customers who subsidized
91

mobile phone users was very small and declining, and the total level of subsidy that they provided was also very small and not sufficient to justify public interest concerns.
2.394. Vodafone made several points. First, it said that, if mobile termination charges were
too low and access charges were, as a consequence, high, then these 3 million fixed-line-only
homes were much less likely to take up mobile subscriptions. If mobile subscriptions were
efficiently priced (according to Ramsey principles), however, some of the 3 million homes
could be expected to take up mobile subscriptions over the next few years. Second, Vodafone
said that the evidence clearly showed that fixed-line-only households made fewer calls to
mobiles than did those who also had a mobile subscription. Its NOP research indicated, it said,
that approximately 60 per cent of fixed-line-only subscribers made no more than one call a
week to a mobile phone, in contrast with approximately 30 per cent of people who were mobile
subscribers. Third, Vodafone said that, in relation to those who were unlikely to take up a
mobile subscription but did make calls to mobile phones, it could well be that these consumers
would be losers if mobile call termination charges were to be efficiently priced, but that
would simply be the consequence of having efficiently-priced termination charges. It would be
wrong for the CC to conclude that mobile call termination charges should generally be set at
inefficiently low levels to protect these consumers; it would be better to allow mobile call termination charges to be set at efficient levels and make special provision for the groups
adversely affected.
2.395. Orange, O2 and T-Mobile argued that, if termination charges were price-capped and
subsidies had to be reduced or removed, it would be mobile-only customers, rather than fixedline-only customers, who would be disadvantaged. Orange said that the mobile-only group was
in fact the poorest group of customers. Lexecon, on behalf of Orange, challenged Oftels data
on the relative sizes of the fixed- and mobile-only constituencies and on the income of mobileonly households. It said that the proposed regulation of termination charges would be to the
detriment of the mobile-only households, who were significantly worse off than the main beneficiaries, the fixed-line-only households. T-Mobile made similar points.
2.396. Oftel in turn submitted evidence to us that the group of mobile-only customers
represented only 5 per cent of households or about 1.25 million people, and that this number
was, moreover, falling. Its research showed that, although about half of the mobile-only group
were among the most disadvantaged in society, the other half were significantly more wealthy
and had chosen not to have a fixed-line telephone because of their lifestyle rather than because
of financial constraints. Oftel considered, moreover, that Orange had overstated the loss to
mobile-only consumers in terms of the higher outbound charges that would arise from a
reduction in termination charges. This was because Orange had, in Oftels view, failed to
recognize that the retail mobile market was imperfectly competitive and because it had also
ignored possible increases in fixed-to-mobile call volumes following price reductions for those
calls as the result of the regulation of mobile termination charges.
2.397. No agreement was reached between Oftel and Orange on these matters during our
inquiry and we were unable to establish to our satisfaction precisely which groupthe fixedline-only or the mobile-only householdswas the more disadvantaged financially. We consider
that all these comparisons need to be treated with caution, although we are satisfied that the
fixed-line-only group is the larger of the two.
2.398. There is another, smaller group, which contains people who do not own a mobile
phone, but who make calls to mobiles. Oftels market research (from November 2000) into
users of payphones showed that at least 12 per cent of British adults used payphones at least
once a month. Of these, 50 per cent used payphones to call mobiles and of this 50 per cent,
31 per cent did not own a mobile phone. This equated to 850,000 adults who frequently used
payphones to call mobiles but did not own one, thus in Oftels view paying an excessive termination charge without any benefit from lower prices at the retail level. Frequent payphone
users tended to be younger than the general population profile, Oftel said, to be from lower
income groups, to live in rented accommodation and to have no fixed line at home.
92

2.399. We acknowledge that fixed-to-mobile callers benefit from there being as many
people with mobile phones whom they can call as possible. But in our view, the appropriate
extent to which that benefit should be reflected in the termination charge is the externality surcharge (see paragraphs 2.334 to 2.386). Beyond that surcharge, we do not accept that it is fair
that fixed-line customers who call mobiles should subsidize mobile phone subscribers. It is not,
in our view, necessary to decide whether fixed line-only or mobile-only customers would be the
more advantaged or disadvantaged by a price control on termination charges. It is not only
these groups that suffer loss by virtue of high termination charges on calls to mobile phones.
All those who have both a fixed line and a mobile phone will suffer a detriment to the extent
that they use their fixed line instead of their mobile phone to call a mobile phone, or make more
off-net calls to mobile than they receive. We believe that the group of payphone users who call
mobiles but do not themselves own a mobile phone also unfairly contributes to funding the
MNOs customers through high termination charges.
2.400. For these reasons, we reject the argument that high termination charges used to subsidize customer acquisition costs and the price of outbound mobile phone calls do not produce
adverse distributional effects. In our view, this is not only a distortion that particularly
disadvantages certain groups of fixed-line-only and payphone customers but also operates as a
detriment to all customers who use fixed-line telephones more than mobile phones when
making calls to mobiles, or who make more off-net calls to mobile than they receive.

Consumer welfare
2.401. The MNOs said that the funding of subsidies for customer acquisition from termination charges was justified because it brought about the maximization of aggregate consumer
benefit. Vodafone said that the MNOs were not profligate in offering subsidies to mobile customers. All the MNOs had begun by charging high prices for handsets (O2 and Vodafone in
1985 and Orange and T-Mobile in 1993/94), Vodafone said, and had found that customers were
particularly sensitive to handset prices; as a result of this they had reduced their prices. The
availability of lower-priced handsets in the period 1998 to 2001 had allowed millions of consumers to acquire mobile phones, it said, bringing benefits to them and society as a whole.
Vodafone told us that, typically, consumers replaced their handset every three years (other
MNOs thought every 18 months or two years); this was not surprising for a young technological industry where product innovation had led to rapid technological development and
significant improvements in call quality. Moreover, in Vodafones submission, the MNOs had
no incentive to offer subsidized handsets to an unnecessary extent. In short, the subsidy maximized aggregate consumer welfare. The other MNOs made much the same points.
2.402. We do not accept the MNOs arguments that, in effect, consumer welfare is maximized only through the current structure of prices prevailing in the mobile sector. This structure
of pricing may, in theory, be an efficient way to recover fixed and common costs, but there are
other considerations which we believe should be taken into account in assessing whether a particular pricing structure operates in the public interest. These other considerations include, in
particular, whether that pricing structure is equitable as among different telecommunications
users, whether it encourages or impedes competition, and whether it encourages patterns of
behaviour which are undesirable, such as over-frequent upgrading of handsets, or too little use
of handsets in relation to the investment made in them. We believe that the structure which has
existed over the past four years has produced an inequitable outcome so far as the customers of
the FNOs are concerned. In particular, costs are loaded (via high termination charges for fixedto-mobile and off-net calls) on to consumers who are not customers of the MNOs who, because
each MNO has a monopoly on call termination on its network, are able to impose excessive
charges on them. These customers bear (through their call charges to mobile phones) the costs
of high termination charges (above those appropriate to correct for network externalities) that
are used to fund mobile customer acquisition and retention. This has encouraged a frequent
replacement of handsets while at the same time distorting the volume and pattern of use of
fixed and mobile telephony, for example, by discouraging fixed-to-mobile calls.
93

Reduction or removal of the subsidy would be detrimental to mobile sector and

benefit FNOs
2.403. The MNOs argued that reduction of their subsidies aimed at customer acquisition
would have a detrimental impact on others in the mobile sector while benefiting the FNOs. In
particular, the reduction would deter the take-up and use of mobile phones; it would prejudice
the financial position of the MNOs and the handset manufacturers and retailers; and it would
unjustly benefit the FNOs.
2.404. We discuss later (see paragraph 2.573) whether a reduction in termination charges
might bring about a reduction in mobile phone subscriptions; we conclude that we do not
expect any significant reduction in the number of subscribers. However, we do not accept the
MNOs arguments that our objective should be to maximize the number of mobile subscribers,
or the volume of calls involving mobiles, if that brings significant disbenefits to fixed-line customers. We must take account of the fixed sector as well. We have received no evidence that
mobile phone use is an activity particularly deserving of subsidy (other than to the extent of the
network externality, for which we make allowance) and we note that there is no universal
service obligation on the MNOs. However, we think that the effect of a reduction of the subsidy
on mobile subscription is less clear-cut than the MNOs have represented it to us. We consider it
likely that, in the absence of subsidy, handsets would be upgraded less often and would be kept
for longer. There may well be some small fall-off in subscriptions but we believe that such falloff is likely to be rather less than the MNOs have suggested because, once people have
acquired a mobile phone, they will not readily give it up.
2.405. The effect on others in the industry who benefit from the current subsidy, in particular retailers selling mobile packages and handset manufacturers, would be to reduce somewhat
the demand for their services and products, as people would be likely to keep their handsets for
longer. To the extent that retailers currently use the funding provided by the MNOs not to offer
subsidies to customers but to improve their own margins, then they would be worse off. However, in a saturated market, where much of the retailers and handset manufacturers business is
concerned with handset upgrading rather than with servicing new subscribers, this outcome
might be regarded as a reasonable price to pay for a more equitable price structure overall, and
hence not against the public interest. That is indeed our view. In any event, the introduction of
3G will, if there is sufficient consumer demand for it, to some extent compensate for any falloff in business which the removal of the subsidy on 2G handsets might bring about.
2.406. The MNOs also put it to us that removal of the subsidy would enrich the FNOs at
the expense of the MNOs. This objection was advanced in the strongest terms by Orange,
which argued that a price cap on mobile call termination charges should be accompanied by a
pass-through requirement, whereby the reduction in mobile termination charges would have to
be passed through directly by the FNOs into fixed-to-mobile charges. Otherwise, Orange
implied, the FNOs would gain a windfall. Orange said that its concern arose particularly from
the recent decision of the DGT that, with effect from 1 August 2002, BTs retention would be
regulated as part of a basket of retail prices, rather than individually regulated, as it had been
previously.
2.407. We do not accept this argument. In so far as the largest FNO, BT, is regulated by
reference to a basket of retail prices, the benefit must be passed to BTs customers rather than
being retained by BT itself, albeit not necessarily proportionately to the extent to which its
customers make fixed-to-mobile calls (see also paragraphs 2.474 to 2.477). To the extent that
the other FNOs which, unlike BT, are not regulated, are obliged by commercial pressures to
pass on the benefit to their customers, then they too could not become unfairly enriched. We
therefore expect all customers to benefit from a reduction in termination charges, either directly
or indirectly. It would be for Oftel to intervene if it appeared that a lack of competition was
enabling the unregulated FNOs to benefit from the removal of the subsidy to the MNOs instead
of passing it on to their customers.
94

2.408. BT drew our attention to the adverse effect which the current price structure had on
competition between the FNOs and the MNOs, and the difficulties that the FNOs experienced
as a result in securing call business. Oftel said that, so far as the impact of high termination
charges on other operators was concerned, some degree of competition existed between fixedto-mobile and on-net calls. The FNOs, however, found it very difficult to compete in this area,
Oftel said, and were seen by their customers as responsible for the high prices they charged,
which derived from the artificially high termination rates they had been charged by the MNOs.
Accordingly, the FNOs found it difficult to increase their call volumes.
2.409. Vodafone submitted that fixed and mobile subscriptions did not compete in the
same market, at least to any appreciable extent. The evidence suggested, it said, that consumers
purchased a mobile subscription to complement a fixed subscription, not as a substitute for a
fixed subscription. Therefore, to the extent that fixed subscribers contributed to subsidizing
mobile subscriptions, that would not lead to any appreciable distortion of competition between
FNOs and MNOs in selling subscriptions.
2.410. As to call origination, Vodafone said that, according to the evidence, mobile technology was more expensive than fixed technology and this fact was widely recognized. This
meant that subscribers tended to use mobile phones only when they could not use a fixed telephone. 75 per cent of mobile-originated calls were, Vodafone said, made by people on the
move. Vodafone also drew our attention to the fact that mobile customers did not fully use up
their bundles of free minutes; typically, a user on a Leisure tariff would use [] per cent of
available minutes. If mobile subscribers were tending to make mobile-originated calls in place
of fixed-originated calls, consumers would be expected to use up more of their bundled
minutes. Finally, Vodafone said that there was no evidence that mobile customers were making
more on-net calls at times of day when such calls were cheaper than fixed-to-mobile calls. All
in all, the evidence did not support an argument that there was appreciable on-net substitution
for fixed-to-mobile calls.
2.411. O2 said that it was implausible to believe that fixed and mobile operators competed
for calls and yet that this did not somehow constrain mobile operators in their approach to
pricing. There was evidence that fixed and mobile operators did compete for at least a part of
the consumer demand for telecommunications services, as its NOP survey showed. But, in any
event, the public interest would not be prejudiced by asymmetric regulation of the fixed and
mobile markets. The underlying situation of the two sectors was very different, with mobile
networks still in the process of establishing themselves (despite increasing penetration), and the
effect of not allowing recovery from general revenues was much more serious for the MNOs
than for the FNOs. Orange told us that the argument that FNOs were adversely affected
depended critically upon the presumption that MNOs and FNOs competed in the same market,
a matter which the CC had not, it said, established.
2.412. We do not accept the MNOs arguments. We think that there is at least a degree of
substitution at the level of call origination between fixed and mobile calls, even if they are not
formally in the same market. As noted above, high termination charges for fixed-to-mobile
calls skew customers choices towards making on-net rather than fixed-line calls (whenever
they are able to do so and are aware of the price differences). To the extent that the high termination charges levied by the MNOs limit the degree of competition that the FNOs can present to the MNOs, this reduces the overall level of competition for telecommunications services
and hence, indirectly, results in an uneven playing field. Moreover, even if every fixed and
mobile phone customer used both products in the same ratio, the existence of the subsidy
extended to mobile phone customers would still tend to skew usage from the inherently lowercost (fixed line) to the higher-cost (mobile) technology, to the detriment of consumers both
individually and in the aggregate.
95

Financial position of the MNOs

2.413. We were also told that the financial viability of one or more MNOs would be put at
risk by an enforced reduction in termination charges.
2.414. T-Mobile drew our attention to what it called the clear obligation of the DGT and
the CC under section 3(1) of the Act to ensure that the MNOs ability to finance their activities
was given primary weight in relation to price controls. It told us that if Oftels proposals for
price control of termination charges were implemented, its improving cash-flow trend would be
reversed [
Details omitted. See note on page iv.
]. Other operators, in particular Vodafone and to a lesser degree Orange, had
more flexibility in responding to the revenue losses that would result from a reduction in termination charges, it said. T-Mobile said that it would be forced to cut back on investment plans,
[Details omitted. See note on page iv.] undermining the future level of competitiveness in the
industry.
2.415. BT argued that our inquiry concerned the structure of prices levied by the MNOs,
that is, the impact that any regulatory action applied at the wholesale level might have at the
retail level. The inquiry was not, it said, about controlling or capping the MNOs overall profitability or driving down their total revenues. BT said it expected that, if call termination charges
were required to be brought closer to costs, there might well be an impact on other prices
charged by the MNOs and possibly on commissions for retailers, as there was no reason to
think that the MNOs would bear all of the reduction themselves.
2.416. We discuss the financial impact of a price cap on each of the MNOs more fully later
(see paragraphs 2.565 to 2.569). Here, we note that, to the extent that competition has been
effective in the sector, as the MNOs have been arguing that it is, then the benefits of the subsidy have been enjoyed not by the MNOs themselves but by their customers. To that extent, at
least, the removal of the subsidy should not adversely affect the financial standing of the
MNOs.

Roll-out of 3G
2.417. The MNOs told us that high termination charges were needed to finance 3G.
Vodafone, for example, said that if termination charges were set below efficient levels, MNOs
would need to reduce handset subsidies as a result of lower incoming revenue streams. This
would apply to the new and relatively expensive multimode 2G/3G handsets. The sale of these
handsets would accordingly be slower than would otherwise have been the case and the widespread availability of 3G services would be retarded. Marginal customers would be likely, it
said, to drop their mobile subscriptions and total call volumes would be likely to fall, leaving
the MNOs with stranded 2G assets. If there was spare 2G capacity, Vodafone said that it would
have the incentive to use it and delay the installation of 3G coverage. Vodafone asserted that
any reduction in call termination rates in the next 12 months would coincide with the launch of
3G, and consumers facing higher charges for subscription and voice calls would be less likely
to be prepared further to increase their monthly mobile expenditure on 3G services.
2.418. mmO2 plc, O2s parent company, told us that it was not a question of whether voice
callers to 2G mobile phones would fund the roll-out of 3G. The issue was whether the financial
profile of mmO2, taking into account all revenues and costs, was such as to enable the company
to attract investment that would allow it to continue to compete effectively in the mobile market, through 2G and shortly through 3G services. Any price control should therefore be set at
such a level, and over such a period, as to ensure that 2G and 3G licensees remained able to
finance investment sufficient to enable them to compete effectively in the mobile market.
2.419. T-Mobile told us that a price control involving a reduction in termination charges
would have the effect of reducing directly the required level of its cash flow and hence limiting
96

the extent of roll-out of the 2G and 3G network, [

Details omitted. See note on page iv.
]. 3G was, it said, a crucial opportunity for T-Mobile [ Details
omitted see note on page iv.
] and to create a new level playing field
between all the mobile operators. This, however, required cash which T-Mobile said it simply
would not have under the proposed price control.
2.420. Oftel told us that it did not think 3G should be subsidized but should stand or fall on
its own merits. 3G was an incremental investment from which a degree of profitability was
anticipated; if such profitability was sufficient to cover its expected cost of capital, then 3G
should be capable of attracting funding. There was no linkage between that and any reduction
in 2G termination charges. We put it to Oftel that if the lowering of termination charges
brought about a reduction in (2G) handset subsidies or increased call prices, with the result that
some people left the mobile network and subscriber numbers fell, there would be fewer people
able or willing to take up 3G services and that could affect the roll-out of 3G. Oftel took the
view that a loss of subscribers in these circumstances would be an efficient outcome, as such
customers would be valuing the existing service at below its resource cost.
2.421. BT also told us that the investment case for 3G should stand or fall on the basis of
the functionality of the services provided; it should not depend on excessive charges being
permitted on voice call termination using another, existing technology. Indeed, if this were to
be the case, with 3G investment being dependent on excessive 2G interconnection charges,
fixed-line customers with a demand only for voice call termination to mobile would be contributing to the cost of a new service for which they had no demand. BT also submitted that the
new EC Directives (see Chapter 4) did not permit a subsidy to be raised on interconnection
charges paid by interconnecting operators simply to encourage MNOs to take on more customers than they would otherwise choose to do.
2.422. We broadly share Oftels and BTs views on this matter. We believe as a matter of
principle that the MNOs wish to invest in 3G does not justify termination charges that are in
excess of a reasonable estimate of their cost, particularly if those charges are ultimately derived
from customers of FNOs.

Distortion of consumer choice

2.423. We considered the argument put to us by Oftel that excessive termination charges
for fixed-to-mobile calls distorted consumers choice between fixed-to-mobile and on-net calls.
2.424. Oftel said that current regulated termination charges were already significantly
above the costs of termination, O2s and Vodafones by 80 per cent and Oranges and TMobiles by almost 70 per cent. While the choice between fixed-to-mobile and off-net calls was
not distorted, since both reflected the same excessive termination charge, the effect of excessive
termination charges for fixed-to-mobile calls was to deter consumers artificially from using a
fixed line to call a mobile when they could make an on-net call instead. It was generally
acknowledged that mobile technology used more resources than fixed technology. The result
was economic inefficiency, Oftel said, because the call would most efficiently be provided over
the fixed line, in the sense of maximizing consumer utility relative to the resource costs
involved in providing the service. This could lead to an inefficient allocation of resources, since
the price signals would induce consumers to use the higher cost service (mobile-to-mobile).
2.425. Oftel added that, if mobile-originated and fixed-originated calls to mobiles were in
the same market, this distortion might lead, in addition to inefficient resource allocation, to a
distortion in competition. Whether or not this was the case was unclear to Oftel. However, in
97

Oftels view, there was certainly the potential for the current level of termination charges to
distort competition as well as to impair economic efficiency.
2.426. The MNOs challenged the notion that the use of high termination charges to fund
mobile customer acquisition brought about a distortion of the structure of prices. Vodafone
argued that termination charges were not above the level at which an economically efficient
balance of subscription, outgoing and inbound call charges would be achieved. A pricing
structure based on, or reflecting, Ramsey principles brought about a pattern of selling prices
that maximized aggregate consumer welfare. O2 said that, far from being a distorted price
structure, the current structure was consistent with principles of efficient recovery of fixed and
common costs across a number of services. The whole concept of subsidy was inappropriate
where a consumer was buying a bundle of services, where there were externalities and where
there were fixed and common costs to be recovered. Orange said that it was incumbent on the
CC to provide an economic analysis of the socially optimal prices before it could conclude that
there was a price distortion, and also to provide evidence that such a price distortion was giving
rise to significant consumer detriment. T-Mobile also said that there were good efficiency
reasons for the current vertically integrated structure of pricing.
2.427. For all the reasons we have set out in the preceding discussion, we do not accept
that the current structure of prices set by the MNOs is either economically efficient or operates
to the benefit of consumers or society at large. It distorts consumer choice by encouraging consumers to choose a higher cost service than they otherwise would. Even if the retail market
were fully competitive, the setting of high termination charges in order to fund low access and
call origination prices would produce a heavily skewed structure of tariffs which would both
distort consumer choices, particularly between fixed-to-mobile and on-net calls, and have
adverse distributional effects of the kind already considered (see paragraphs 2.396 to 2.406).
Reductions in termination charges would, in our view, be likely to bring about a different and
less distorted pricing structure, which would provide fewer incentives for consumers to acquire
new handsets and greater incentives for them to make calls. If a reduction in termination
charges meant higher handset prices and subscriptions for mobile customers, this would not
necessarily in our view be against the public interest. We do not, however, believe that reductions in termination changes will be fully recovered by the MNOs by offsetting access or call
origination charges. Indeed, we think that reductions in termination charges might induce a
temporary pause in the general decline of those other prices (see paragraphs 2.564 to 2.566).

Conclusion on arguments for and against high termination charges

2.428. We have a duty to consider the distributional and other arguments discussed in the
previous paragraphs in the context of section 3(2) of the Act. So far as the distributional arguments are concerned (see paragraphs 2.390 and 2.400), the data is insufficiently robust by itself
to support conclusions leading to action; and we note that Oftel said in its review statement of
September 2001 that distribution arguments by themselves did not justify intervention. However, we have found that the FNOs customers unfairly bear the costs of the distorted pricing
structure that we have identified, and this is an important element of our argument that action
should be taken to reduce termination charges.

Termination charges in the absence of a price control

2.429. Our terms of reference require us to consider whether the termination charges of the
four MNOs that are the subject of our inquiry would, in the absence of a charge control on
them, be likely to operate against the public interest. We have already concluded (see paragraph 2.147) that each MNO has a monopoly of call termination on its own network and that
insufficient constraints exist at either the wholesale or retail levels to keep call termination
charges down to cost (see paragraph 2.211). We have also concluded that call termination
98

charges are substantially above a reasonable estimate of their costs (see paragraph 2.387). This
confirms our view that competitive pressures are not as strong as they need to be to curb the
MNOs market power in call termination. We have already seen that the MNOs expend considerable sums in acquiring and retaining customers and that the structure of their prices
reflects the importance they attach to increasing the size of their respective customer bases. In
effect, the MNOs are prepared to subsidize customers to bring them on to the network and to
fund such subsidization through above-cost termination charges, leading to above-cost off-net
and fixed-to-mobile prices.
2.430. We sought to establish whether these features would persist in the absence of any
regulation of termination charges. To help us answer this question, we began by considering
whether the MNOs current pricing strategies were either based on Ramsey principles or were
not dissimilar in structure to prices that had been set according to Ramsey principles, as these
were put to us by the MNOs. If they were based on Ramsey principles, then the MNOs would
have the incentive to keep termination charges above cost in order to fund subsidies designed to
bring new customers on to the network and retain existing customers. We considered three
aspects of the MNOs businesses. These were: the reliability of elasticities of demand in the
mobiles sector; the wide differences between on-net and off-net prices in the context of a retail
market that was not fully competitive; and levels of handset subsidies.
2.431. First, we were not satisfied that there was any way of establishing reliable estimates
of elasticities of demand in the mobiles sector with enough precision to inform pricing
decisions. Hence, we believe that there are problems in calculating reliable Ramsey priceswe
discuss this issue in more detail in Chapter 8.
2.432. Second, we consider that the wide disparity in retail pricing between on-net and offnet calls (see paragraphs 2.122 to 2.131) cannot obviously or easily be accounted for by
demand or cost differences for each of these types of call. There is therefore no reason to
believe that, in the absence of regulation, this price disparity would disappear. Vodafone, while
arguing that the Ramsey structure of prices broadly prevails now and would prevail in the
absence of regulation, because the market for the supply of mobile services was generally
competitive, agreed that the differential between on-net and off-net prices was not Ramseybased. Moreover, as we have already concluded (see paragraph 2.211), the retail market is not
fully competitive. We would therefore not expect Ramsey prices to exist in that market.
2.433. Third, the withdrawal or reduction of handset subsidies on prepay phones over the
past year or 18 months, in the absence of any evidence of changes in demand for, or the cost of,
providing the subsidy, indicates that those subsidies were either not Ramsey-reflective before
the withdrawal or reduction, or are not Ramsey-reflective now; or it may be that in neither case
have the prices been Ramsey-optimal. Further, the MNOs have largely removed handset subsidies from that part of the market where they say that consumers are the most price-sensitive.
2.434. Taking the considerations in paragraphs 2.431 to 2.433 into account, we conclude
that the MNOs do not currently set Ramsey prices, except possibly for the time-of-day variations around the regulated (or quasi-regulated) average termination charge. We accept that the
MNOs decisions on the day, evening and weekend rates to set, given the target average charge,
are based on at least an informal understanding of the respective price-elasticities and are therefore fairly similar to Ramsey prices (they would be true Ramsey prices only if the average
termination charge had been set at a level which ensured that termination profits were no higher
than a normal level). The objection to Ramsey pricing which we mention earlier does not apply
so strongly here (on equity, we accept that charging more for calls made at peak times is in
principle desirable on cost grounds; on entry, we have already said that each MNO has a
monopoly of termination on its own network, so the question of possible entry is irrelevant).
2.435. In the light of our finding, we now consider the MNOs arguments that they would
have the incentive to set Ramsey prices in the absence of regulatory control and, moreover, that
99

Ramsey prices would produce maximum allocative efficiency and maximum consumer welfare
in the mobile sector. (We discuss the question of whether any regulated price for termination
should be derived from Ramsey principles later, in the context of our discussion of remedies
(see paragraphs 2.510 to 2.519).)
2.436. As we noted above, the MNOs regard their customers as revenue streams and they
compete with each other to capture these streams by offering discounts and incentives to
retailers and handset subsidies to potential subscribers to encourage them to join their networks.
Having acquired customers, the MNOs price call origination and call termination so as to
maximize their revenues.
2.437. Oftel told us that in its view the incentives faced by the MNOs would lead them, if
unregulated, to depart substantially from the structure of Ramsey prices. It was important to
understand, first, that prices were only Ramsey prices, which maximized economic efficiency,
if they reflected the market elasticities of demand rather than the elasticities faced by the MNOs
as individual operators (which are referred to as firm-specific elasticities). A second key
point, Oftel said, was that the size of the mark-up on each service under Ramsey pricing
reflected its elasticity of demand relative to the elasticities of the other services, a corollary of
the fact that Ramsey pricing described a structure of prices, not a single price. In Oftels view,
the MNOs had no incentive to set a structure of Ramsey prices, because the relative values of
their individual firm-specific elasticities were different from the relative values of the market
elasticities for the different services they sold. This was because there was a disparity in the
degrees of competition which the MNOs faced in the two main markets in which they operated,
the termination market and the retail market. Given that each MNO was a monopolist in its own
termination market, it faced the market demand and the market elasticity in termination; but, in
the retail market, retail competition (albeit not fully efficient competition) drove a substantial
wedge between the firm-specific elasticity faced by each MNO and the market elasticity.
Therefore, if the MNOs were to set their retail prices on the basis of their individual firmspecific elasticities in the retail marketas they would if they wished to maximize their
profitsthey would not set Ramsey prices.
2.438. Orange put it to us that, because the MNOs competed across a range of mobile services, the degree of competition between them might be expected to be similar in each case,
and that therefore the firm-specific and market elasticities would be similar. We do not accept
this argument. We have already established that each of the MNOs has a monopoly of termination on its own network. It therefore faces the same market and firm-specific elasticities in
respect of that service. In the retail market, however, we have also established that there is
active competition in access to mobile services and, although to a lesser extent, in call origination. Therefore, the relationship between the firm-specific and the market elasticities in the
retail market cannot be the same as that in the termination market.
2.439. O2 argued that, for a given level of termination charge, other mobile prices would be
similar to Ramsey prices in that they maximized subscriber numbers. Even if O2s argument
were valid, it is not clear to us that the maximization of subscriber numbers is the only factor
relevant to welfare maximization or a sufficient condition for its achievement. There are other
matters to be taken into account in assessing welfare maximization, including considerations of
fairness among different groups of consumers and efficiency in the use of the network, so that
calling patterns are not distorted (see paragraphs 2.423 to 2.427).
2.440. We asked Oftel and the four MNOs what they thought would happen to the level of
termination charges if there were no charge controls after March 2003. Oftel said that, in the
absence of a price control, the MNOs would have the incentive to set a profit-maximizing price
and it was quite possible that the profit-maximizing price would be substantially above the
current level, possibly exceeding 20 ppm. Oftel stressed that the termination charges currently
set by O2 and Vodafone were already at the absolute maximum allowed under the price cap and
indeed had been at that level during the whole of the current charge control period.
100

2.441. Vodafone told us that, in a market where the MNOs competed to sell mobile subscriptions and outgoing calls, each MNO would be subject to pressures to maintain termination
charges for calls from FNOs above efficient levels, with a view to using the excess revenues
thereby earned to fund competition to win subscribers. It said that, at a weighted average of
9.3 ppm, current termination charges were capped at too low a level to deliver optimal
allocative efficiency. The optimal price, by which Vodafone meant the price that reflected the
maximum consumer benefit, would lie in a range of 11 to 15 ppm. Vodafone told us that the
optimal level of call termination charges was calculated taking into account the fact that society
generally benefited from a customers decision to take up or retain a mobile subscription,
because other members of society valued making calls to, and receiving calls from, the mobile
subscriber (the so-called network externality). In the absence of regulation, Vodafone believed
that the MNOs could be expected to set termination charges above efficient levels. However,
there was a level above which MNOs would be unlikely to raise termination charges, owing to
the constraints, at such levels, from other services, including outbound call charges. Vodafone
suggested that that level was 17 ppm, and possibly even up to, but no higher than, 20 ppm.
2.442. The other MNOs did not agree with Vodafone. They said that they neither could,
nor would, raise termination charges in the absence of a charge control because of the competitive environment in which they operated. And even if they tried to raise termination charges,
they believed that BT or another MNO would complain to the DGT, who would intervene and
determine a termination charge which would certainly be no higher than current levels.
2.443. Vodafone told us that the MNOs had recently had to withdraw or reduce handset
and other subscription subsidies, since it had become clear that the subsidies needed to bring
new subscribers on to the network could not be justified by reference to the revenues which
those subscribers could be expected to generate. This provided further evidence, in Vodafones
view, that termination charges were too low and would need to be higher to bring new subscribers on to the network and retain existing subscribers. The recent withdrawal or reduction
of subsidies had brought about a reduction in the overall level of mobile subscription, yet it was
widely recognized, Vodafone said, that it would be efficient to generate more subscriptions for
the benefit of society at large.
2.444. In considering the commercial incentives of the MNOs to increase or decrease termination charges, we note that Orange had set its termination charges substantially below those
of O2 and Vodafone up to February 1997 as a competitive strategy, but without success. During
the MMCs inquiry into the termination charges of O2 and Vodafone in 1998 (see paragraph
2.7) Orange had said that:
through analysis around that time of call volumes and market research, it had
become increasingly clear that significantly undercutting other networks termination charges would not provide incentives to most customer groups to select
Orange over other operators. But it believed that insensitivity to termination
charges might no longer be so widespread and might not apply to certain user
groups.1
If the new fifth operator attempted the same strategy, seeking to gain subscribers by setting
lower termination charges than the incumbent MNOs, in our view the evidence suggests that it
would be no more successful than Orange had been.
2.445. We think it is clear that, in the absence of any price control, the MNOs would have
the incentive to raise termination charges above their current levels. As we have seen,
Vodafone thought that charges could rise to up to 20 ppm compared to an efficient level in the
range 11 to 15 ppm. Oftel told us that termination charges might rise to more than 20 ppm in
the absence of regulation. The MNOs justify the raising of termination charges on grounds of
1

See paragraph 2.91 of MMC 1998 report.

101

both economic efficiency and benefit to society at large, since higher termination charges, used
for example to subsidize handsets, generate more subscribers for mobile networks, but we have
already said that we do not accept that this is necessarily an efficient outcome. By raising termination charges the MNOs would, in accordance with their pricing strategies, enable subscription prices to be kept at levels capable of attracting and retaining subscribers. However, if
the mobile sector approached saturation and new subscribers required even larger subsidies to
induce them to become mobile subscribers, the MNOs call termination charges would need to
be increased commensurately (although clearly subject to some upper limit, defined by the
continued willingness of fixed-line customers to call mobile phones). In these circumstances, it
is possible that termination charges could exceed the 20 ppm maximum level suggested by
Vodafone, if there were no regulation in place to contain them.
2.446. For all the reasons set out above, we think it likely that the MNOs would not adopt
Ramsey prices whatever the degree of control of termination charges. It seems to us more likely
that they would price above the Ramsey level on call termination while setting some retail
prices, particularly those relating to customer acquisition, at prices below the Ramsey level.
This is because of the different degrees of competition that they face in the call termination and
retail markets respectively. The MNOs have the incentive to set very high termination charges
in the call termination market, where each operator has a monopoly of termination and faces
the market elasticity. They have the incentive to set prices very much lower than Ramsey levels
in the retail market (which is not fully competitive), particularly the access market, where
competition for subscribers is strongest.

Conclusions on the public interest

2.447. In the references, the DGT has asked us whether, in the absence of charge control
mechanisms, the charges made by O2, Vodafone, Orange and T-Mobile respectively to operators of fixed or mobile public telecommunications systems for calls to GSM handsets connected to the public telecommunications systems of O2, Vodafone, Orange and T-Mobile
operate, or may be expected to operate, against the public interest. The termination charges of
O2 and Vodafone are currently subject to a charge control. We find that these termination
charges operate against the public interest and that, were the charge controls to be removed, the
termination charges of O2 and Vodafone may be expected to operate against the public interest.
The termination charges of Orange and T-Mobile are currently subject to no charge control. We
find that these termination charges operate against the public interest and that, in the absence of
any charge control on them, the termination charges of Orange and T-Mobile may be expected
to operate against the public interest.
2.448. We make these findings because, in the absence of a charge control, we believe that
termination charges will be set above levels that reflect a reasonable estimate of their costs. As
shown in Table 2.11, a reasonable estimate of their costs is LRIC, including allowances for
relevant network and non-network fixed and common costs, plus an allowance for externalities.
We term the level of termination charges that reflects such costs the fair charge. The fair
charge in the year to March 2003 (in 2001 prices) is 6.2 ppm for O2 and Vodafone and 7.0 ppm
for Orange and T-Mobile. The current average termination charges in 2001 prices are between
9 ppm and 10 ppm . Accordingly, we find that, in December 2002, termination charges exceed
the fair charge by between 30 and 40 per cent. The fair charge for the year to March 2006 (in
2001 prices) will be, we estimate, 4.7 ppm for O2 and Vodafone and 5.3 ppm for Orange and TMobile. Therefore, if retained at their current real levels, termination charges could be up to
double the fair charge by 2006. Our expectation that termination charges will, in the absence of
a charge control, be set at levels in excess of the fair charge arises because:
(a) there is a relevant market in the termination of calls on mobile networks and O2,

Vodafone, Orange and T-Mobile each has a monopoly on the termination of calls on its
network (see paragraph 2.147);
102

(b) there are no adequate demand-side substitutes for call termination on the network of the

called party at the retail level, no ready alternatives to making a call to a mobile (such as
calling a fixed line or using text messaging), and no adequate consumer strategies that
are easy to use and effective enough to constrain the level of termination charges (see
paragraph 2.147);
(c) termination charges are already set at levels substantially above the fair charge (see

paragraph 2.387);
(d) each of O2, Vodafone, Orange and T-Mobile has a strong disincentive to reduce its

termination charges from their present levels to rates that are lower than those of its
competitors, or to the fair charge, or at all; and each has an incentive to increase its termination charges from their present levels (especially if another MNO had done so); and
thus, in summary, given that current termination charges exceed the fair charge, each
has an incentive to maintain termination charges above the level of the fair charge (see
paragraph 2.52); and
(e) there is insufficient knowledge of, and concern about, incoming call charges and the

prices of calls to mobiles on the part of consumers to induce them to change their
behaviour to such a degree as to put pressure on call termination charges and force these
down to the level of the fair charge (see paragraph 2.146).
2.449. In reaching our finding on the public interest we are required by section 13(8) of the
Act to have regard to the matters in respect of which duties are imposed on the Secretary of
State by section 3 of the Act. In our view, the amount by which termination charges will, in the
absence of a price control on them, exceed the fair charge will be contrary to the interests of
consumers in the UK in respect of the prices charged for telecommunication services. This is
because termination charges at the wholesale level are reflected in the level of charges incurred
by consumers who make fixed-to-mobile and off-net calls. The adverse effects that result from
this excess of termination charges over the fair charge are, and may be expected to be, that:
(a) the costs incurred by the FNOs and the MNOs through paying mobile call termination

charges that are in excess of the fair charge are wholly or mainly passed through by
them into their customer tariffs, with the result that consumers pay too much for fixedto-mobile and off-net calls (see paragraph 2.210);
(b) the high price of fixed-to-mobile calls discourages such calls and, as a result, too few

such calls are made, thereby distorting patterns of telephone use (see paragraphs 2.390
to 2.428);
(c) consumers who make more fixed-to-mobile or off-net calls than on-net calls unfairly

subsidize those who mainly receive calls on their mobile phones or who mainly make
on-net calls, or who make little use of their mobile phones (see paragraphs 2.390 to
2.428);
(d) the excess charges for termination have the further effect that they serve to encourage or

facilitate significant distortions in competition because MNOs are not obliged to charge
and subscribers are not obliged to pay the economic cost of handsets. This leads to the
undervaluation of mobile phone handsets by the MNOs customers combined with a
greater turnover (churn) than would take place if customers paid charges which reflected
the proper valuation of such handsets. This leads to yet further distortions in greater
expenditure on mobile customer acquisition than would have taken place if termination
charges reflected costs more closely and if handset prices reflected the costs of handsets
more closely (see paragraphs 2.174 to 2.197); and
(e) the higher prices of calls from fixed to mobile phones and the lower price of on-net

mobile calls encourage greater use of the higher-cost (mobile) technology at the expense
of the lower-cost (fixed) alternative (see paragraphs 2.390 to 2.428).
103

2.450. We do not find that there is, or that there may be expected to be, any offsetting public interest benefit arising from termination charges set by the four MNOs being in excess of
the fair charge which could justify their continuation. We therefore believe that the interests of
consumers within the UK, in respect of the prices that they pay for telecommunications services, are prejudiced.
2.451. We conclude that the adverse effects that have to be remedied are those set out in
paragraph 2.449, which flow from termination charges being in excess of the fair charge. As
already noted, currently (year to March 2003), such charges are 30 to 40 per cent in excess of
the fair charge. In the absence of a charge control on them, we expect that termination charges
could be up to double the fair charge by 31 March 2006.

Remedies
2.452. In view of the conclusions set out above, we are required to answer the question
whether the adverse effects that we have identified could be remedied or prevented by modifications to the licences granted to O2, Vodafone, Orange or T-Mobile.
2.453. In this section, we first set out the DGTs proposals for a remedy in the shape of an
RPIX charge control on the MNOs termination charges, by way of modifications of the conditions of each of the MNOs licences (see paragraphs 2.454 to 2.456). We next consider the
MNOs objections to regulation (see paragraphs 2.457 to 2.471) and then discuss some other
possible remedies, including those which we canvassed in our Remedies Statement (see paragraphs 2.472 to 2.504). We then turn to our own proposals for a price cap on the termination
charges of the MNOs (see paragraphs 2.505 to 2.509) and an alternative (Ramsey) approach
suggested by the MNOs (see paragraphs 2.510 to 2.523). We then turn to issues surrounding
implementation of a charge control (see paragraphs 2.524 to 2.547). We next examine the
financial impact of our proposals on, respectively, consumers and the MNOs (see paragraphs
2.548 to 2.577). Finally, we give our formal conclusions on licence modifications (see paragraphs 2.578 to 2.580).

The DGTs proposals for charge controls

2.454. In his terms of reference to us, the DGT set out proposals for remedying the adverse
effects he had identified (see Annex B of the terms of reference in Appendix 1.1 to this report).
He proposed licence modifications which would require the four MNOs to reduce their termination charges by RPI12 per cent each year for four years until March 2006. In order to
arrive at the appropriate charge control, the DGT identified a target charge in the final year of
the control (2005/06) of between 5.8 ppm and 6.3 ppm for combined 900/1800 MHz operators
and between 6.2 ppm and 6.8 ppm for 1800 MHz operators. The target charge was based on the
LRIC of termination, to which he added two mark-ups, the first an equal proportionate mark-up
(EPMU) for the recovery of any common costs and the second an allowance for the value of the
network externality. This target charge produced a range of X of between 11 and 13. The
DGT proposed an X in the middle of that range, namely of 12.
2.455. The DGT told us that he had considered the possibility of setting a new starting
charge for the controls, rather than initiating the controls from the existing levels. This would
have involved an initial step change (known as a P0 adjustment) in the level of the average
termination charge from 2001/02 to 2002/03 and, as a consequence, the value of X would have
been lower. A one-off adjustment might have been considered appropriate, he said, on the
grounds that the current level of termination charge was substantially above cost. However, the
DGT took the view that bringing prices down to cost over the period of the price control (which
104

he termed a glide path) was a more appropriate route. Among other things, the DGT said, it
provided greater incentives for cost reduction, since the regulated company would be able to
keep the benefits of efficiency gains for a period before consumers captured these benefits
through lower prices.
2.456. The DGT said that he recognized that regulatory intervention was appropriate only
when there was a reasonable expectation that its benefits would exceed its costs. Oftels
approach to the assessment of the appropriate level of call termination charges had, he said,
involved a cost-benefit analysis, involving an estimation of the set of charges that maximized
the welfare of consumers, subject to ensuring that suppliers were able to earn a reasonable
return on their investment. This assessment took account of the benefits to all customers,
including those calling mobile phones as well as mobile phone customers themselves. The
DGT said that he had made conservative assumptions in his cost-benefit analysis. This had
generated a conservative estimate of the overall benefits of the control, since the benefits identified in the cost benefit analysis were those associated with a better balance and structure of
prices, rather than lower overall prices, and did not include the benefits associated with the
driving out of excess profits or increases in efficiency.

The MNOs views on a charge control

2.457. O2 told us that the DGTs proposal for a control mechanism on call termination
charges was based on two mistaken views, the first that current charges were not at socially
optimal levels because they were above cost, and the second that termination charges could rise
in the absence of regulation. On the contrary, it said, because none of the carriers with whom it
interconnected would agree to pay a higher price, termination charges would not rise in the
absence of regulation and would be likely to remain at socially optimal levels. There were
indeed strong reasons for MNOs to reduce termination charges in line with and in response to
competitive pressures, it said, in order to preserve a balance between incoming and outgoing
call prices. O2 had considered how it would respond to a reduction in termination charges as the
result of regulatory intervention; clearly, it said, how it adjusted its other tariffs depended on
how other mobile operators responded to a termination revenue shortfall. For example, if one or
more operators decided not to adjust any of their outgoing call or subscription charges (including handset subsidies) O2 would not be in a position to make such changes without seriously
jeopardizing its commercial position.
2.458. O2 also said that the proposal for regulation of termination charges raised serious
issues about its capacity to remain viable. Although it would seek ways in which to rebalance
its tariffs by raising prices for other services, it said that this readjustment would be unable to
absorb the major shock that a sharp reduction in the level of termination charges would bring
about; this would not be a minor price adjustment around a set of balanced tariffs, but a significant structural change in the mobile market. Apart from the implications for the tariff structure,
O2 said that the MNOs would look for ways to cut costs, but competitive pressures had already
forced them to cut as many costs as possible. Savings might also be sought by cut-backs in
innovation and product development. This would be very damaging to the public interest in the
long term. More generally, regulation would put at risk both the existing benefits of huge
increases in mobile penetration levels and future benefits, most significantly the roll-out of 3G
services. O2 proposed an alternative remedy (call-back), which we discuss in paragraphs 2.500
to 2.503.
2.459. Vodafone told us that any significant intervention in the regulation of mobile call
termination rates would be seen by the capital markets as a sign of heightened regulatory risk,
thereby contributing to an increase in the MNOs cost of capital. Investment in new services
would therefore be retarded, as the cost of such new investment, and the returns required to be
earned from it, increased. In Vodafones view, the reduction in handset subsidies, as the result
105

of lower incoming revenue streams, would slow the sales of multimode 2G/3G handsets, and
the widespread availability of 3G services would be retarded. Vodafone said that, as marginal
customers dropped their mobile subscriptions and total call volumes fell, the MNOs could well
be left with stranded assets. The reduction in termination charges would also, Vodafone said,
coincide with the launch of 3G services but customers were unlikely to be prepared to increase
their monthly mobile expenditure on 3G services if subscriptions and call charges had
increased.
2.460. Vodafone told us that it was unnecessary for the CC to resort to an RPIX charge
control to regulate mobile-to-mobile termination charges, since these could be constrained by
the imposition on MNOs of a requirement to conclude bilateral agreements for such charges. At
most, an RPIX price control for mobile-to-mobile termination charges could be justified as a
safeguard, to reassure callers to mobiles that the bilateral agreements would make them no
worse off; but it was to be expected that the rates arrived at in bilateral agreements would be
below a mobile-to-mobile cap set at the fixed-to-mobile level. We discuss this proposal in paragraphs 2.473 to 2.479.
2.461. Vodafone accepted, however, that it might be appropriate for the CC to propose an
RPIX charge control on fixed-to-mobile call termination charges. At present, Vodafone said,
charges were not excessive, and the only function of such a charge control would be to ensure
that future efficiency gains were passed on through charge reductions. Vodafone said that termination charges were currently too low and for optimum efficiency should be in the region of
11 to 15 ppm, but might increase to between 17 and 20 ppm in the absence of a price control.
Vodafone thought the charge should be set so as to allow each MNO to recover LRIC, plus a
Ramsey mark-up to recover common costs plus the network externality. It should take account
of all relevant non-network costs and be based on the costs of an efficient operator.
2.462. Orange told us that it did not favour a price cap on mobile termination charges
because this would fail to address any of the adverse effects that the CC had purported to find
in the mobiles sector. If there was indeed a detriment of high consumer prices for fixed-tomobile calls, outweighing the positive social benefits of higher mobile penetration, then price
control would only be effective if coupled with pass-through of any changes in termination
charges to the prices charged by FNOs to their customers. A reduction in call termination
charges by itself would not guarantee any commensurate reduction in the retail prices charged
by the FNOs. Until recently, pass-through had been achieved through specific regulation of
BTs retention on fixed-to-mobile calls, Orange said, as a result of which BT was obliged to
pass on any reduction in mobile termination charges to its customers in the form of reduced
retail prices. Other FNOs were to some extent constrained to accept the BT price. However, the
recent decision by Oftel to remove this specific regulation from BT with effect from 1 August
2002 would allow BT to redirect the additional profits it could make on fixed-to-mobile calls
(in the circumstances of a price control) to other retail services within the broad retail price
basket now allowed by Oftel. The extent to which FNOs would pass through any reductions in
the mobile termination rate to their retail fixed-to-mobile prices would in future, therefore,
depend on the price-sensitivity of fixed customers in calling mobile phones. As the CC
appeared to believe that such price-sensitivity was low, the result would be a massive
regulation-driven transfer of funds from the MNOs to the FNOs.
2.463. Orange also submitted that any rebalancing by the MNOs to offset the effects of
regulated reductions in termination charges, by removing subsidies or increasing outgoing call
charges, could not take effect immediately, yet any delay would have a significant impact on
the MNOs overall revenues and levels of return. This would be especially damaging to the
financially weaker MNOs, while the financially strongest MNO, Vodafone, would be better
able to delay rebalancing, thus increasing its market share and over time its relative financial
strength. Hence a one-off reduction would significantly reduce the intensity of competition in
the mobile market. Orange pointed, in addition, to the adverse impact of a one-off reduction in
106

mobile termination rates on the stability of traffic patterns and costs (as fixed-to-mobile retail
prices decreased and outgoing call rates from mobiles increased), and on future investment in
mobile network infrastructure and 3G services in the UK.
2.464. Finally, Orange told us that the proposed remedy of a charge control tended to foreclose the prospect of competition in call termination and relied instead on perpetual regulation.
It was surely preferable, it suggested, to seek a remedy that enhanced the impact of market
forces on call termination and which did not need constant review and replication. Apart from
price control with a pass-through obligation, Orange thought that non-discrimination provisions
and greater price transparency were remedies (both of which we consider in the next section)
that would address the detriments we had identified. If there was inefficient competition in the
mobiles sector, leading to excessive focus on subsidies and excessive churn, then a prohibition
on handset subsidies might also be appropriate. Towards the end of our inquiry, Orange told us
that, although in its view no price control was justified, it would be willing to give an undertaking to continue to reduce its interconnection charges by RPI9. If Orange were permitted to
reduce its termination charges by RPI9 each year, that would, it said, alleviate the problems
associated with rebalancing its various charges and prices, and it would not expect to see the
need for further rebalancing [
Details omitted. See note on page iv
].
2.465. T-Mobile said that regulation of termination charges, if on the lines proposed by
Oftel, would be likely to lead to an inefficient and highly distorted price structure with fewer
calls being made and dramatic imbalances in traffic. Any resulting increase in call origination
charges coupled with a decrease in call termination charges would reduce overall call volumes,
given the relatively more price-elastic nature of call origination. An increase in subscription
charges would be likely to reverse the subscriber growth of the last few years, driven by the
availability of prepay packages which had provided access to mobile services by socioeconomic groups that were previously unable to take advantage of them. A substantial increase
in handset prices would reduce mobile affordability and hinder the provision of new 3G phones
and services to UK consumers. T-Mobile also told us that charge controls would be likely to
plunge loss-making operators into further losses, threatening their long-term viability. Other
MNOs, particularly Vodafone, had more flexibility in how they responded to the revenue losses
from reduced termination charges. As to the detriments to the customers of the FNOs, if the
FNOs retention was not directly regulated, there was no guarantee that charge controls would
deliver the savings to consumers that regulation was intended to secure. At a late stage in our
inquiry, T-Mobile told us that it would be prepared to hold its termination charges at their
current levels. We asked T-Mobile whether it would maintain that position even if Vodafone
were to raise its termination charges to 17 ppm in an unregulated environment. T-Mobile told
us that, in its view, there was no possibility that Vodafone would act in this way and so it was
quite willing to give us its undertaking.
2.466. The DGT told us that he had considered the impact of any regulation on the financial viability of the MNOs before deciding on his proposed price cap. He said that he had calculated a target charge for the MNOs, upon which the control was based, consisting of the cost
of providing termination, the cost of capital, a mark-up for common costs and an externality
surcharge. He had also proposed a glide path for four years, so that the regulated charge was
substantially above cost until the last year of the cap. Therefore, capping charges could not be
the reason behind any financial difficulties of the MNOs.
2.467. We considered carefully each of the matters put to us by the MNOs (as set out in the
previous paragraphs) in response to the proposal for a price control and reduction in termination charges. In doing so, we took note of the business plans of each of the four MNOs, in the
case of O2 and Vodafone, for the period to the end of March 2006; in the case of Orange, for
the period to 31 December 2005; and in the case of T-Mobile, the period to 31 December 2006.
We noted that these strategy documents have to varying degrees anticipated the impact of
Oftels proposals for a charge control; for example, they assume that retail prices will continue
107

to fall, that termination charges will be reduced and that there will be a significant levelling off
of new subscriber numbers. We set out below the main issues raised by the MNOs and our
response:
(a) that any increase in subscription charges as the result of a reduction in termination

charges would reverse the subscriber growth of the past few years. We address this
matter when considering our calculation of the impact of our price cap proposals on the
MNOs (see paragraph 2.567);
(b) that a price cap will result in perpetual regulation. We have rehearsed earlier in this

chapter (see paragraphs 2.94 to 2.100) the possibility of technological developments that
might at some time in the future obviate the need for regulation of termination charges.
There is also the possibility that the gradual move to 3G technology will produce a
similar result. In the meantime, we believe that a charge control by way of an RPIX
price cap is the only effective method of remedying the adverse public interest effects
and that none of the other remedies discussed would be adequate for this purpose;
(c) that a reduction in termination charges will lead to an upward adjustment of other tariffs

and therefore a decline in subscriber numbers. The MNOs business plans project a continued downward trend in retail prices. There is scope for the MNOs to offset the
reduction in revenue from termination charges by maintaining retail prices at their
current levels rather than decreasing them at the same rate as has occurred over recent
years (we discuss this matter in paragraphs 2.564 to 2.566);
(d) that the roll-out of 3G, and the investment in other new services, would be threatened.

We discussed this matter earliersee paragraphs 2.417 to 2.422and concluded that

there is no justification for existing customers of the FNOs to pay for new services
through above-cost 2G termination charges. We also discuss the effect of regulated termination charges on the financial position of the MNOs later in this chapter (see paragraphs 2.548 to 2.563);
(e) that the capital markets will see any reduction in termination charges as a heightened

regulatory risk. We think this is unlikely. There has been widespread discussion of the
possibility of a control on termination charges and the evidence suggests that investors
have already taken account of this possibility (for example, see paragraph 15.240);
(f) that a reduction in handset subsidies will slow the sales of dual 2G/3G handsets and put

in jeopardy the roll-out of 3G; and that this will leave the MNOs with stranded assets.
We have already discussed this possibility in paragraphs 2.417 to 2.422, in connection
with the effects of the removal of excessive termination charges on the mobile sector;
(g) that the other MNOs will suffer disproportionately in relation to Vodafone, as the

strongest MNO. We address this matter in paragraphs 2.573 and 2.574; and
(h) that any reduction in termination charges needs to be coupled with a requirement on BT

to pass through such reductions to its retail prices. We discuss this matter in the following paragraphs.
2.468. We considered the concern (see paragraph 2.467(h), expressed in particular by
Orange, that a cap on the MNOs call termination charges would be effective only if coupled
with pass-through by the FNOs of charge reductions into the prices they charge their own customers. BT told us that it currently faced two main types of competition in the provision of
calls to mobile phones: cable companies and Indirect Access Suppliers, the latter using BTs
lines but routeing their calls through another network. Both types of competitor would benefit,
along with BT, from a reduction in termination charges, it said, and BT would therefore need to
pass through the vast bulk of the reductions in those charges in order not to get out of step with
108

its competitors and risk losing its competitive position. Moreover, the prices of calls to mobiles
were a major source of dissatisfaction among its customers, BT said, and thus a matter which it
had every incentive to address.
2.469. Even if it did not pass through the reductions in termination charges penny-forpenny, customers would still not lose out, BT said. This was because, when mobile call termination call rates fell, BTs retention element on these calls increased unless BT cut the prices of
calls to mobile phones penny-for-penny. A retention increase was treated as a price increase in
the basket of prices. Under its current price control, BT had to lower other prices by an amount
equal to any failure to pass through call termination reductions on a penny-for-penny basis. BT
had thus to reflect in its prices the totality of the cost savings from lower call termination rates
to customers. In short, BT said, although it had more flexibility under the current arrangements
than formerly, it had a competitive incentive to pass through the vast bulk of reductions in termination charges to its fixed-to-mobile prices, and could not appropriate any of the reductions
if it did not do so.
2.470. We put it to BT that, even though its retention was capped, it had not hitherto taken
the opportunity to reduce its retention below the cap, so its cap was by no means a floor price.
BT made two points in response. First, it said that other FNOs competing with it for fixed-tomobile calls would be receiving any reduction in termination charges to the same extent as BT.
In the increasingly competitive market in which it operated, BT would be forced to pass the
reduction through in its prices to its own customers or lose out to those competitors. Second,
BT said that there was currently very little incentive for it to reduce the prices of fixed-tomobile calls or its own retention, because the size of the termination charge in relation to the
total cost of the call was such that any reduction would hardly be noticed by customers; for
example, if BT brought down its retention from 3 ppm to 2 ppm, the cost of a daytime fixed-tomobile call would be reduced only from 30 ppm to 29 ppm. BT said that, if termination charges
came down, then there would be much more incentive to focus on the price of fixed-to-mobile
calls and bring those prices down.
2.471. We are satisfied that the reduction in the call termination charges of the MNOs that
would follow a charge control on them would be substantially if not wholly passed through by
BT and the other FNOs to their customers, although not necessarily in the price of fixed-tomobile calls. This is because of the competitive pressure that would be brought on BT by the
other FNOs and the incentive of all the FNOs to increase their call volumes by offering more
attractive prices. Even if the reduction in mobile termination charges was not wholly reflected
in lower fixed-to-mobile prices, customers of the FNOs would benefit from price reduction in
other areas.

Other possible remedies

2.472. In view of the inherently intrusive nature of charge controls and bearing in mind our
duty to consider the proportionality of any remedy we might recommend in relation to the
detriment to be removed or prevented, we explored other possible ways in which the public
interest detriments we had identified could be addressed. Accordingly, in our Remedies
Statement, we canvassed a number of other possible remedies, in addition to the remedy of a
charge control through an RPIX type price cap. In addition, towards the end of our inquiry,
O2, Orange and T-Mobile suggested further alternative remedies. We now consider each of
these possibilities, beginning with bilateral agreements between MNOs.

Bilateral agreements
2.473. Vodafone proposed that, as there would be no incentive for MNOs to maintain termination charges for mobile-to-mobile calls above efficient levels if they could be sure that no
109

competitor could do so, mobile-to-mobile termination charges could be constrained by way of

bilateral agreements. The incentive to price above efficient levels could be neutralized,
Vodafone argued, if all MNOs were required by regulation to negotiate and conclude bilateral
interconnection agreements, under which each MNO agreed a specified, reciprocal call termination rate, and not to alter that rate without giving the counterparty MNO an opportunity to
do so at the same time. Such an arrangement would prevent either MNO from stealing a
march on the other by raising its call termination charges to a higher level. Each MNO would
instead have an incentive to set its own charges at the efficient level, so as to maximize consumption of its total services. Vodafone subsequently suggested that, if the CC thought it was
necessary, there could also be a safeguard charge control under which each MNO would be
prohibited from charging more for mobile-to-mobile termination than it was permitted to
charge for fixed-to-mobile termination under a separate fixed-to-mobile termination charge
control.
2.474. In reply to concerns raised by other MNOs, Vodafone also argued that such bilateral
agreements could contain provisions to prevent FNOs from benefiting from the mobile-tomobile call termination rate (the optimal level of which would, in Vodafones view, be lower
than the optimal fixed-to-mobile call termination charge because of the higher super-elasiticity
of demand for mobile-to-mobile calls) where they transited traffic across another mobile network. Vodafone suggested that we should make clear in our findings that the inclusion of such
provisions in bilateral agreements would promote the public interest, by allowing MNOs effectively to maintain separate call termination charges for mobile-to-mobile calls, and fixed-tomobile calls, each being set at its efficient level. Conversely, said Vodafone, if MNOs could not
prevent FNOs from benefiting from the mobile-to-mobile call termination rate , then the very
fact of that arbitrage would ensure that, in setting its rate to other MNOs, each MNO would be
constrained to set its mobile-to-mobile termination rate by reference to the applicable fixed-tomobile rate.
2.475. Vodafone submitted that the bilateral agreements should provide that, if one MNO
delivered fixed-originated traffic to the other, it should disclose the origin of such traffic to the
other MNO and pay, in respect of such traffic, the relevant FNOs termination charge.
Vodafone further submitted that our recommendation should reflect the fact that MNOs could
continue to have recourse to Oftel if they were unable to reach agreement. In addition, noting
that both Vodafone and O2, being presently designated as having significant market power
(SMP) for the purposes of the Interconnection Directive, were subject to non-discrimination
obligations (which would, however, be removed with the implementation of the new EC
Directives), Vodafone submitted that our recommendation should expressly acknowledge that it
should not be regarded as being unduly discriminatory for MNOs to agree lower mobile-tomobile termination rates than they offered to FNOs.
2.476. The DGT commented that, in his view, the argument that bilateral agreements
would avoid the need to regulate mobile-to-mobile termination charges was not robust, and
noted that Vodafone itself (by proposing a safeguard cap) apparently recognized this. The
MNOs incentives were to set charges for mobile-to-mobile calls in a way that would act to
weaken retail competition and this was not conducive to the public interest. Traffic imbalances
would be likely to inhibit a downward movement of prices, because (as between two MNOs
potentially entering into a bilateral agreement) the MNO with the balance of inbound traffic in
its favour would prefer the status quo of high termination charges. Moreover, MNOs might find
it mutually beneficial to keep off-net termination charges high in order to create an entry barrier
for Hutchison 3G, given that, compared with the MNOs customers calls, in the early stages of
Hutchison 3Gs operation a much larger proportion of its customers mobile-to-mobile calls
would be off-net (rather than on-net).
2.477. The DGT considered that the termination charges for mobile-to-mobile calls should
be regulated so that they were no higher than those for fixed-to-mobile calls. This could be
110

achieved in one of two ways. First, by including both fixed-to-mobile and mobile-to-mobile
termination charges in the same charge cap but, to the extent that MNOs set a lower weighted
average level of charges for mobile-to-mobile compared with fixed-to-mobile termination, this
would not count towards compliance with the cap (the converse would apply, however).
Alternatively, there could be a distinct safeguard cap on termination charges for mobile-tomobile calls, separate from, but at the same level as, the cap on termination charges for fixedto-mobile calls. If the argument put forward by the MNOs was correct, then the safeguard cap
would be a non-binding constraint; but it would provide protection against the incentives that
the MNOs might have to set excessive termination charges for mobile-to-mobile calls. On the
basis that the MNOs were able to distinguish the origin of calls that they terminated (between
fixed-originated and mobile-originated), the DGT considered the two-cap solution more appropriate because the weights in each cap would reflect the traffic mix of the relevant call type.
The DGT also argued that, with a safeguard cap in place, mandated bilateral interconnection
agreements were neither required nor appropriate.
2.478. None of the other MNOs lent positive support to Vodafones proposals for bilateral
agreements. T-Mobile believed they might in theory deliver efficient termination charges, but
feared that, as overseen by Oftel, they would lead to inefficient pricing since net purchasers of
call termination services would have the incentive of opting for an inefficiently low price,
covering only network costs, rather than reach a commercial agreement. O2 considered that
bilateral negotiation of agreements did not differ from the current institutional constraints
already operating as an insurmountable bar on increased termination charges; nor did Orange
see the proposals as an effective remedy, since they did not address what it regarded as the
fundamental issue of calls from fixed lines to mobiles. Of the FNOs, BT emphasized that
binding bilateral agreements could not be used to set fixed-to-mobile termination charges if one
of the parties (the regulated FNO) was not free to depart from preagreed termination rates for
its services; Telewest Communications plc (Telewest) expressed concern that bilateral
agreements could lead to collusive behaviour; and the COG did not believe that subjecting the
MNOs to an obligation to negotiate and enter into bilateral interconnection agreements would
lead to efficient pricing, or remove the need for a price cap on termination charges. The
International Telecommunications User Group (INTUG) did not see bilateral agreements as a
viable solution, whilst both the CA and the National Consumer Council (NCC) worried about
the potential for bilateral agreements between MNOs to have anti-competitive effects, the CA
believing that MNOs would have an incentive to act against new entrants, who might come into
the market with a more transparent pricing structure.
2.479. In our view, the concerns expressed by interested third parties and the lack of support, even from the other MNOs, for Vodafones proposals for a system of mandated bilateral
interconnection agreements represent a fair reflection of their lack of merit as a possible
remedy for excessive mobile-to-mobile termination charges, with or without an accompanying
safeguard cap. We consider below the question of whether there should be separate caps,
respectively, for mobile-to-mobile and for fixed-to-mobile call termination charges.

Yardstick regulation
2.480. Another possible remedy that we canvassed in our Remedies Statement was to tie
year-on-year changes in termination charges to average price changes in more competitive
markets, such as the market for retail mobile services. Broadly speaking, this was what was
recently implemented in Australia, by the Australian Competition and Consumer Commission
(ACCC) (see Appendix 2.2, paragraph 14(c), and paragraph 15.235).
2.481. O2 believed that a yardstick remedy, as favoured by the ACCC, passed all the legal
tests for an appropriate remedy in our inquiry, provided that it was not coupled with an
111

immediate downward P0 adjustment. Moreover, said O2, such a remedy would provide all the
benefits of switching from CPP to RPP with none of the detriments. Contrary to objections,
there was no risk of distortion of the competitive outgoing sector, or of fostering collusion
among operators; furthermore, the ACCC experience showed that a retail benchmarking
approach was workable. A yardstick remedy would, said O2, provide a market-driven, proportionate and potentially permanent solution without the need to revisit the issues periodically.
If, for some reason, the CC were to conclude that O2s preferred, voicemail remedy would be
inappropriate or ineffective (contrary to O2s belief), then the CC should recommend a retail
benchmarking remedy along ACCC lines, requiring reductions in termination charges at the
same rate as the decline in the rate of mobile retail prices.
2.482. The DGT maintained that the ACCC remedy was defective because it did nothing to
redress the distorted pricing structure perceived as the key adverse public interest effect. At
best, the remedy would reduce termination charges by as much as mobile retail prices but,
given the public interest detriment identified, a necessary condition for any remedy was that it
should reduce termination charges by more than that.
2.483. Vodafone told us that it did not favour a remedy whereby charges for call termination were regulated by reference to movements in retail prices, believing it attractive in principle but, in practice, difficult to specify accurately and leading to distortions in the setting of
the comparator retail prices. Orange believed there would be a distorting effect in relation to
mobile retail services, and that we should seek remedies that were more clearly targeted at the
relevant public interest issue. T-Mobile believed this remedy would be less intrusive than a
one-off adjustment to call termination charges, but would be complex to implement and would
still have a significant financial impact on T-Mobile. The CA commented that it would be concerned if a solution more complex than capping were proposed, as this would make an
unintended impact more likely; and the NCC thought that such a remedy made little sense and
would distort other markets.
2.484. The COG argued that the tying of call termination to call origination charges would
not be effective in the absence of a fully competitive call origination market and, moreover, that
it would not be a viable remedy even if call origination was fully competitive. When making
pricing decisions for call origination, MNOs would be constrained from making competitive
reductions by the effect that such reductions would also have in the (less competitive) termination market. In any case, the COG believed, it was in practice very difficult to calculate an
average origination charge to be used as a benchmark, given the complexity of tariff packages
and the practice of tying free origination minutes to the line rental charge.
2.485. In our Remedies Statement, we gave it as our current view that it would be inappropriate for call termination charges to be tied to (retail) competitive services because of the risk
of distorting a more competitive market, and we stand by that view. Another possible method
of yardstick regulation might be to tie termination charges of each MNO to the average price of
originating calls of all the other MNOs. However, this is problematic in network markets, since
the termination charge of each MNO affects the outgoing prices of the other MNOs. Thus, a
further possibility might be to tie the termination charge of each MNO to the average price of
on-net calls of the other MNOs (times an engineering-based fraction to reflect the share of an
on-net call accounted for by the termination leg). This mechanism has a number of the desirable properties. First, an MNO that lowers its on-net price gains more customers directly; and
since termination is set as the average of all other MNOs on-net prices, it does not lower its
termination charge, and hence gains the termination revenue from these new customers. Hence,
all MNOs have an incentive to lower their origination prices, and this will also lower all termination charges. Thus the mechanism should effectively introduce competition into the termination market without distorting competitive markets. Second, MNOs with cheaper networks
can charge lower origination charges, reducing termination charges on rival networks and so
112

providing an incentive for all parties to lower termination charges. Third, the scheme is simple
and does away with the need for building a LRIC model and allocating fixed and common
costs, and hence constitutes very light regulation. However, in a market with a small number of
players interacting repeatedly even if not collusively, estimating something like a LRIC model,
which the scheme attempts to avoid, might be considered essential for monitoring purposes. In
sum, we were concerned at the scope for conscious parallelism (even if not other forms of
collusion) if such a scheme were to be implemented, and about the consequential requirement
to deal with that.

Technological solutions
2.486. We considered earlier how technological solutions including dual SIM cards,
delivery of calls via other wireless technologies and VoIP might lead to supply-side substitution, and how technical solutions based around optional RPP might lead to an increase in
competitive pressure on call termination charges. We did not see satisfactory evidence that any
of these possibilities would be likely to deal with the adverse public interest effects brought
about by an absence of regulation of termination charges in the period to 2006. Moreover, we
were not satisfied that, in the absence of regulation, there would be adequate incentives for the
MNOs to introduce any such potential solutions; this was evident from the fact that, although a
number have been and continue to appear feasible, none has yet been implemented.
2.487. We also considered whether there was a practical form of regulatory intervention to
encourage the MNOs to implement technological solutions that might either lead to supply-side
substitution or increased competitive pressure. Given that any solution would need time to be
developed, and that its success would be far from certain, we concluded that we could not rely
on such a measure within the timescale considered in this inquiry. We return to these matters
below.

Receiving party pays

2.488. We considered whether a system of RPP whereby the billing systems of the MNOs
would be modified so that called parties paid for the termination leg (but not the outbound leg)
of an inbound call might represent an alternative to the regulation of termination charges for
remedying the adverse public interest effects that we had identified. Such a system could be
expected to bring about a greater concern on the part of mobile customers about call termination charges and encourage competition between the MNOs in the setting of termination
charges in order to gain customers. An additional incentive to the MNOs to keep termination
charges low would be the desirability of encouraging customers to keep their mobiles switched
on, so that the MNOs would not lose termination business.
2.489. The DGT told us that he accepted that a system of RPP was likely to remove the
competition problems associated with CPP, by increasing competitive pressure on prices for
inbound calls, but he considered that the benefits of RPP were likely to be outweighed by
adverse effects on economic efficiency, and also in terms of consumer resistance, and the initial
costs of implementation. The DGTs view, therefore, was that the case for the introduction of
RPP in the UK was weak.
2.490. O2 expressed similar views and identified, in particular, a significant risk that under
RPP there would be detrimental effects (particularly on low-income members of society), as
people would turn off their phones to avoid paying for incoming calls. Vodafone believed that a
system of RPP would lead to a sub-optimal usage of mobile services and fail to secure the
regulatory objectives of section 3 of the Act. Orange strongly disagreed that RPP would be in
113

the public interest, arguing that it would have significant distorting effects, would be exceptionally disruptive for consumers and MNOs and would be totally unjustified. T-Mobile
believed compulsory RPP would be highly disruptive to the continued development of the
mobile industry but requested us to recommend that MNOs should develop, make available,
and promote, optional tariff package(s) for mobile customers that would enable customers of
fixed networks to call those mobile consumers using a freephone and/or local rate number.
2.491. BT considered that RPP would, if it could be introduced, make call termination
charge control unnecessary, since it would remedy the underlying causeCPPof the problem; however, it accepted that RPP might be costly to implement and might lead to mobile
phone users turning off their phones to avoid paying for unwanted incoming calls. The INTUG
believed that the introduction of RPP would be disruptive, confusing and expensive. The CA
said that the benefits of RPP had to be balanced against the evidence from other markets, such
as the USA, where RPP had constrained market growth through encouraging consumers not to
keep their phones on, severely undermining the network benefits offered by all MNOs. The
NCC believed that reversing the CPP principle would be hugely disruptive both to consumers
and operators.
2.492. The evidence we received suggested that the MNOs might be able to change their
systems to RPP at reasonable cost. However, we consider that, although going to the root
causeCPPof the lack of competitive constraints on termination charges, a mandatory system of RPP would entail too many significant disadvantages for consumers for us to
recommend it as an appropriate and proportionate remedy for the adverse public interest effects
that we have identified, not least because it might lead to significant numbers of users switching off their mobile phones. We also note that both T-Mobile and, indeed, Orange already offer
tariff options whereby customers may pay to receive incoming calls if they so choose, but it
appears that neither has experienced more than modest demand for these services to date.

Pass-through and price transparency measures

2.493. Orange argued that pass-through regulation and improved price transparency
offered significant possibilities of improvements in the competitive dynamics of call termination. We have seen (see paragraph 2.467(h)) that Orange believed a price control on fixed-tomobile calls would only be effective if conjoined with pass-through of reduced termination
charges into FNOs retail prices for such calls. More generally, Orange considered that the
incentive for MNOs to compete on fixed-to-mobile calls by reducing their termination charges
was critically weakened by the fact that the FNOs set the retail price and this was no longer
required to reflect the mobile termination rate. Orange therefore thought it essential that passthrough regulations were imposed on FNOs (or at least on BT). Alternatively, or in addition,
similar effects could be achieved, said Orange, if FNOs (or BT, at least) were required to show
on their customers retail bills a breakdown between the termination charge and the amount of
the FNOs retail retention. T-Mobile, for its part, favoured additional transparency measures,
provided that they were extended to the fixed retail market, and included specific proposals
among its suggested alternative remedies (see paragraph 2.503). O2 disputed that insufficient
information was available to enable customers to make rational purchasing decisions, pointing
to our first BMRB survey showing that only 18 per cent of respondents underestimated the
price of a 2-minute call to a mobile; however, O2 said that, if we were concerned about price
transparency, it was prepared to work with Oftel and retailers to develop greater transparency.
The CA and the NCC argued that both improved transparency and a price cap on termination
charges were required.
2.494. In our Remedies Statement we mentioned that we shared Oftels view that, while
improved transparency was desirable in bringing about increased consumer awareness, it would
be insufficient by itself to remedy the adverse effects we had identified. We return to these
matters below.
114

Non-discrimination
2.495. A number of FNOs put it to us that a non-discrimination provision (possibly supported by a specific price-squeeze test along the lines referred to in Appendix 2.2, paragraph 14(h)) was required, along with a price cap, in order to prevent competitive distortions
brought about by the MNOs termination charges: the MNOs should be required, they said,
implicitly to charge themselves the same amount for terminating on-net calls as they charged to
terminate incoming calls-to-mobiles from other operators. This should apply, in particular, in
cases where the MNOs provided retail fixed-to-mobile services using mobile virtual private
networks (MVPN) (see Figure 3.7 and paragraph 3.89).
2.496. The COG believed that the fact that the MNOs were apparently able to offer 5 ppm
retail charges on MVPNs was one indication of market failure in mobile call termination. In
offering MVPN retail rates at less than 50 per cent of the fixed-to-mobile termination charge,
the MNOs operated a classic price squeeze which effectively excluded all FNOs from a significant part of the UK market for calls-to-mobiles in the corporate sector. As the volume of
calls-to-mobiles as a proportion of all calls had risen, this problem was becoming increasingly
significant. The COG believed this market failure operated against the public interest, in that it
led to distortion of demand and resulted in a reduction in public welfare and a sub-optimal
allocation of economic resources. The COG considered it a matter of urgency that we should
deal with this situation and, in its view, the only solution was a non-discrimination condition
allied with a price-squeeze test. In addition, because MNOs had not reduced daytime charging
rates (as would have been expected following recent reductions in weekend and evening rates),
resulting in sub-optimal reductions in the relative price of evening calls, the COG believed we
should recommend a one-off reduction to the daytime termination rate (see Chapter 15). THUS
plc wrote in support of the COGs proposal for a non-discrimination condition along with a
price cap, and also pointed out that the present situation resulted in duplication of infrastructure
and hence inefficiency costs. Energis asked us to investigate and to recommend a nondiscrimination requirement or other effective solution for any anti-competitive practices
discovered. In Telewests view, the use by MNOs of their market power in call termination as a
basis to build market power in call origination was the key competition issue: Telewest therefore believed there were strong grounds for both a price cap and a non-discrimination rule. BT,
however, expressed scepticism that non-discrimination and price squeeze tests could offer
much in terms of effectiveness, cost or practicality.
2.497. Commenting on our Remedies Statement, the DGT said that the FNOs main concern appeared to relate to the MNOs private wire tariffs, on the grounds that the FNOs were at
present unable to compete with the low charges made available by the MNOs to corporate customers with large numbers of handsets. The DGT had not previously examined this situation,
which would require a separate investigation. The DGT believed that, if termination charges
were reduced as proposed, this would enable the FNOs better to compete in this area. However,
the DGT suggested that, if we were to consider the FNOs case to warrant early assistance, a
one-off reduction (as opposed to a glide path) would be preferable to suggesting that the DGT
should consider extending the scope of the inquiry into peripheral areas of the retail mobile
market.
2.498. O2 commented that a non-discrimination remedy would fail to address any concern
about the underlying level of termination charges, and that no remedy concerned with addressing the balance of charges at the retail level was justified. In circumstances where the majority
of calls crossed network boundaries, operators simply could not capture consumers on to their
network with the allure of cheap on-net calls. Nor would a price-squeeze test for fixed-tomobile services be justified: MNOs competed with FNOs in a competitive market where each
offered an attractive proposition to consumers, reflecting their respective advantages, and FNOs
could always offer private wire products as service providers of the MNOs if they so chose.
Nor did Vodafone believe that a non-discrimination requirement, whether or not accompanied
by a price-squeeze test, apparently intended to apply to MVPNs, was relevant, since the
115

incidence of the charges generated was de minimis, amounting to only a tiny percentage of
Vodafones revenues. Orange believed a price-squeeze test was unnecessary, unjustified and
would be unduly onerous, but thought that a non-discrimination provision could be an effective
remedy, were we to find an imbalance in the competitive position between FNOs and MNOs
which was an adverse public interest effect, arising from unregulated termination charges
(which Orange disputed). T-Mobile said that it did not favour non-discrimination remedies
since they would lead to unnecessary distortions in pricing, and believed a price squeeze
remedy would have similar disadvantages.
2.499. Our inquiry has focused on the level of termination charges for calls-to-mobiles in
the wholesale market and we have not, therefore, investigated in detail the FNOs specific
complaints of discrimination against them in the corporate sector of the mobile retail market.
However, we think that such complaints (to the extent justified) are likely to result, at least
partially, from the currently high level of the MNOs termination charges. Consequently, if
these charges are brought down to the levels considered below, it may well be that no specific
measures to address the FNOs complaints will be required. Should that not prove to be the
case, it would be open to complainants to pursue appropriate action under the Competition Act
1998, or to the DGT to launch an investigation into the issue.

Further alternative remedies suggested by MNOs

2.500. As already noted, towards the end of our inquiry, O2, Orange and T-Mobile
suggested further alternative remedies.
2.501. O2 believed there was only one remedy that addressed the identified failures in
competition and went no further than was necessary to achieve that objective: the introduction
of voicemail access at fixed-line rates as an alternative means for callers to mobiles to communicate with those whom they were calling (O2s voicemail remedy, see paragraphs 2.111 to
2.112). Rather than seeking to address the symptom of the perceived problem through a price
control, O2 considered that this remedy sought to address any underlying regulatory problem
arising from there being insufficient competitive pressure on call termination charges. Under
the remedy, callers to a mobile would be given the option either to reach the mobile subscriber
in the normal way and to pay termination charges, or to leave a voicemail message at a fixed
national rate: this would bring more pressure on termination charges the more it was used, as
the wider the disparity between those charges and fixed-line rates, the more likely the caller
was to use the voicemail option. O2 shared with us the results of a survey demonstrating, it said,
that callers would welcome the option to pay the termination charge or request a call-back, and
that the remedy would prove attractive enough to consumers to act as a significant constraint on
the setting of termination charges (see paragraph 11.92); O2 also pointed out that callers were
familiar with the concept from BTs voice messaging service. O2 highlighted that the remedy
operated within the competitive dynamic by giving callers an option to pay a termination rate or
to request call-back; that it could also be combined with certain RPP options, so that those
mobile subscribers who wanted incoming calls to be routed to them directly could pay a premium for this facility; that it would be practical, viable, sustainable and, most importantly,
relatively easy to use and to implement; and that it would provide a market-driven, proportionate and potentially permanent solution without the need to revisit the issues periodically. In
short, said O2, this remedy would provide a new and strong incentive on MNOs to ensure that
incoming and outgoing retail charges were balanced and would therefore provide an effective
constraint on termination charges.
2.502. O2 believed, however, that, if we were not prepared to recommend such a solution,
we should recommend a retail benchmarking remedy along ACCC lines (see paragraph 2.481).
2.503. Orange told us that it would be willing to undertake to continue to reduce its termination charges by RPI9 per cent: a step which it envisaged might be accompanied by transparency measures, including the provision of additional information on customers bills.
116

T-Mobile, for its part, told us that it was confident that its prices for termination would not rise
even in the absence of regulation and that it was therefore willing to commit itself to cap its
rates at current levels for a period of three years. That said, T-Mobile agreed that, as a way of
addressing any continued concern on our part about potential increases in termination rates in
an unregulated environment, O2s voicemail remedy would address the root cause of the
perceived problem by imposing additional constraints on termination rates. T-Mobile added
that it would, however, be happy to see one or more further alternative proportionate remedies
adopted, with or without O2s voicemail remedy. These could include a public education
campaign to increase awareness of retail prices of fixed-to-mobile calls; discounted rates for
calls to mobiles for socially disadvantaged fixed-line-only customers; and the offer and
promotion of optional freephone/local rate tariff packages for fixed-to-mobile calls (see
paragraphs 13.65 to 13.67).

Conclusion on other possible remedies

2.504. In considering the possible alternative remedies described above, we have carefully
weighed the obligations on us to ensure that our recommended remedy is proportionate to the
nature and severity of the detriments we have identified. In our judgement, none of these
remedies properly addresses the central issues which we have identified, namely that termination charges are currently excessive in relation to cost, that they are not subject to sufficient
competitive constraints, that the MNOs have the incentive to raise them above current levels,
that they are unlikely to be driven down towards a more cost-reflective level in the absence of a
charge control on them and that the matter is important enough to warrant considering recommending such a control. Given that the termination charges of O2 and Vodafone have persistently been at the maximum permitted levels since the current charge controls were imposed,
and those of Orange and T-Mobile persistently slightly above those levels, we see no prospect
that any of the above alternative remedies (whether individually or in combination) would be
sufficient to bring about a reduction of termination charges to acceptable levels over the period
to April 2006. We therefore now turn to the proposals for a charge control by way of a price
cap, which we consider to be the only remedy likely to address adequately the adverse effects
we have identified. We note that a charge cap would not pre-empt bilateral agreements, price
transparency measures, O2s voicemail remedy or T-Mobiles suggested alternative remedies.
The MNOs put it to us that some of these remedies would be more effective and might even
reduce prices below the cap that we propose below. We are sceptical of these claims but note
that nothing in our proposals would prevent the MNOs from putting their preferred remedies
into practice, should they choose to do so. If and to the extent that they do so, then future regulatory reviews will usefully be informed by hard evidence.

Remedying the adverse effects

Our proposals for a charge control
2.505. We now turn to our own proposal for a charge control. In our view, the most effective licence modification to prevent the adverse effects that we have identified is one imposing
a price cap on termination charges. The design of the cap would be an RPIX formula applied
to the weighted average termination charge of the MNOs. This proposal would directly remedy
the adverse effects we have identified by addressing their root cause. The purpose of a cap on
the level of call termination charges would be to bring termination charges down to the fair
charge, that is, the level of a reasonable estimate of LRIC, plus an allowance for network fixed
and common costs and relevant non-network costs, plus a mark-up for the network externality.
2.506. Fundamental to our approach to assessing the costs of call termination has been the
principle that the party calling a mobile phone should incur only those costs that he himself
causes in making that call. We term this the cost-causation principle. If a consumer wants to
make a fixed-to-mobile or off-net call, then he has to avail himself of the service of termination
117

provided by the operator of the network whose customer he wishes to call and it is fair that
such a consumer should bear the appropriate cost of doing so. The issue of who benefits from
the call can also be relevant, but we consider that this should be reflected in the externality
adjustment, not via the cost causation principle. The cost-causation approach has been an
important part of our identification of the public interest issues and of the adverse effects. After
long and careful consideration of the alternatives, we have not identified any better or fairer
approach, or any reason not to pursue the same cost-causation approach in the selection of a
remedy.
2.507. As we have already seen, the MNOs told us that this approach was an unacceptable
alternative to the correct approach, namely Ramsey. They put it to us that the optimal level of
call termination charges would be achieved by applying Ramsey principles. Ramsey pricing
dictated that the recovery of fixed and common costs would reflect what the MNOs said was
the lower elasticity of inbound calls compared with the elasticity of other mobile services. They
argued that we ought to concern ourselves with efficient solutions to the problem of allocating
common costs, and fairness ought to be a secondary consideration.
2.508. We have described our approach to the LRIC of call termination, which we have
calculated in accordance with the cost-causation principle, earlier in this chapter (see paragraphs 2.244 to 2.333 and paragraph 2.506). Our estimate of the relevant LRIC cost per minute
of call termination of a combined 900/1800 MHz MNO is 5.4p in 2002/03 and 4.0p in 2005/06;
and of an 1800 MHz MNO, 6.3p in 2002/03 and 4.5p in 2005/06 (see Table 2.11). To the LRIC
of call termination, we then add allowances for network fixed and common costs and relevant
non-network common costs, and for externalities. We decided that the mark-up for nonnetwork (that is, administration) costs should be 0.3 ppm throughout the period concerned. To
the extent that customer acquisition costs lead to new mobile subscribers coming onto the
network, thus generating a greater overall volume of calls, we have said (see paragraph 2.331 to
2.333) that this is better captured by way of the externality adjustment than the cost-causation
principle. So far as the network externality is concerned, we have already concluded (see
paragraph 2.385) that a mark-up of 0.45 ppm should be allowed, this also to apply throughout
the period. This is a justified addition in arriving at the fair charge, in our view, because the
caller benefits from having a large pool of subscribers to call and be called by, and hence
should contribute to their recruitment on to, and retention on, the mobile network as a whole.
2.509. In our view, this approach, which starts from LRIC, is both right and fair, because it
attributes costs on the basis of who causes, or benefits from, them. In particular, and having
regard to the adverse effects set out at paragraph 2.449, it should mean that:
(a) consumers do not pay too much for fixed-to-mobile or off-net calls;
(b) consumers who make more fixed-to-mobile or off-net calls than on-net calls, or who

make more off-net calls than they receive, will not unfairly subsidize other consumers;
(c) cost-reflective call charges should minimize distortion in the volumes and patterns of

calling;
(d) there should be no displacement from less resource-intensive to more resource-intensive

technology; and
(e) there will be less incentive for the MNOs to subsidize handset acquisition, which should

reduce the rate of replacement of handsets.

Alternative approaches to calculating a regulated price control

2.510. We discussed earlier (see paragraphs 2.429 to 2.446) whether it was likely that, in
the absence of any charge control on them, the call termination charges of the MNOs would be
118

set at Ramsey levels. We concluded that the MNOs would set neither the structure nor the level
of prices in accordance with Ramsey principles, but rather that they would have the incentive to
set call termination charges that were significantly above Ramsey levels, while setting prices in
the retail market (particularly those related to customer acquisition) that were significantly
different from (and probably below) the Ramsey level. We also concluded earlier (see paragraphs 2.388 and 2.400) that a Ramsey-based approach to pricing might be thought to lead to
distributional unfairness, in that fixed-to-mobile and off-net callers, whose demand is thought
to be relatively price-inelastic, would pay prices for mobile call termination well in excess of
the costs which their calling activity causes the MNOs to incur.
2.511. We received widely differing views on the use of Ramsey from parties who submitted evidence to us. Thus, Telewest put it to us that the Ramsey pricing analyses that had
been presented to the CC (and disclosed to the main parties and some FNOs) were based upon
incomplete models, flawed modelling assumptions and questionable empirical estimates of key
parameters, such that they were probably unusable for any practical regulatory purpose. The
Independent Regulators Group (the group of EC telecommunications regulators, of which Oftel
is a member) said that, although in theory Ramsey pricing would minimize distortion in the
recovery of common costs, the method required robust and detailed information on elasticities,
which was often difficult to obtain. We note that, so far as we are aware, no telecommunications regulator in any other country has fixed prices on Ramsey principles, although
regulators typically allow regulated companies to use Ramsey principles to set the prices of
individual services where a basket of regulated services is subject to an overall price cap.
2.512. Orange said that the work that had been carried out on its behalf showed that the
results for termination charges were relatively insensitive to changes in the demand elasticities,
and O2 said that precise estimates of elasticities were not essential to determining an appropriate structure of prices. O2 said that the absolute size of the elasticities did not matter; what was
important was that the relativities were correct. T-Mobile told us that we should not prefer an
approach just because it appeared to offer simplicity if the Ramsey approach was the best available, albeit based on estimates.
2.513. Vodafone and T-Mobile submitted that the duties imposed on us under section 3 of
the Act were such that Ramsey pricing, with a network externality was, prima facie, the only
correct way to proceed, because this was the means by which to ensure that all reasonable
demands for mobile services were met, and also because this would be the efficient outcome
which was, Vodafone said, required by the words best calculated. It was not, Vodafone
argued, for the CC to decide that, in its judgement, certain demands for services which could be
met at Ramsey prices were unreasonable and should not be met. Vodafone said that there was
no objection in principle to applying a Ramsey mark-up and an externality mark-up to call termination charges, if that was the best way to attain the optimal level of consumption of mobile
services. In Vodafones view, this meant that the CC was not justified in abandoning what
would be the most efficient method of pricing for the benefit of consumers generally, in the
interests of some particular class of consumer, in particular a class of fixed-line-only users. If
the CC were to do so, Vodafone argued, it would elevate the section 3(2) duties above the section 3(1) duties, which would be contrary to the provisions of the Act. Even if the CC were
justified in favouring a particular class of consumer, it would be obliged to pay only a proportionate regard to that class of consumer, and should not jeopardize the attainment of an
optimal level of consumption for the benefit of users generally in order to protect a small group
suffering some limited detriment. The right way for the CC to protect the interests of any such
small group was not to distort the telecommunications market generally by imposing the wrong
structure of prices but by requiring some specific assistance to be provided to that group.
2.514. In our view, Vodafones arguments in relation to the operation of section 3 and to
Ramsey pricing are not supported by the wording of the relevant statutory provisions. We do
not accept that we are putting the section 3(2) duty above the section 3(1) duty or that the
section 3(1) principle (that all reasonable demands for services should be met) necessitates a
119

Ramsey approach. Nor do we accept Vodafones view that best calculated means that we
have to elevate efficiency considerations above those of equity. It is clear that the Act requires
us to pursue the primary duties laid down in section 3(1) in whatever rational way we see fit,
taking all the relevant factors into account, of which efficiency may be one; and that once we
are satisfied that we have, we can consider the section 3(2) duties. We are satisfied that our own
recommendation not only satisfies all the requirements of section 3, but may be expected to
fulfil the duties in section 3(1) while bringing about a more equitable distribution of benefits
among fixed and mobile callers than would be attained under a Ramsey system of pricing. Oftel
also rejected Vodafones argument. It told us that, in its view, there were no proper reasons to
support Vodafones argument that the section 3 duties implied that Ramsey pricing was the
only correct way to proceed. It said that, as arbiter of the means by which to ensure that all
reasonable demands for mobile services were met, the DGT had rejected Ramsey pricing as the
best means to the desired end.
2.515. We have set out above our conclusion on the correct approach to establishing the
costs of call termination. In our view, a Ramsey approach is not consistent with the costcausation principle, that callers to mobile phones should pay only those costs that they cause by
virtue of making calls. But, in addition, we identified three further objections to the Ramsey
approach, which we set out in the following paragraphs.
2.516. First, we believe that if we were to set a Ramsey-based termination charge, we
would have to be confident that other relevant prices would also be set at Ramsey levels. In
other words, we would have to be confident either that it would be appropriate for us to intervene to set not just termination charges but all other relevant prices, or that if we set termination charges at Ramsey levels those other relevant prices would be set at Ramsey levels by the
MNOs. This is because the Ramsey approach is concerned with the relativities between different prices, for example, subscription, origination and termination. In other words, Ramsey
pricing is about a structure of pricing. However, we do not believe it would be appropriate for
us to try to set levels for all the relevant prices. Further, if termination charges were set at the
level commensurate with Ramsey pricing, there would, in our view, be no guarantee that the
MNOs would set their different retail prices at Ramsey levels which maximized overall consumer surplus, subject to the recovery of fixed and common costs. A regulated Ramsey-based
termination charge would be likely to ensure an efficient, Ramsey-type outcome in the mobile
sector as a whole only if competition at the retail level was sufficient to constrain the MNOs to
set their overall structure of prices at Ramsey levels. We have, however, already concluded that
the retail market is not fully competitive and, as a result, the MNOs would both wish, and be
able, to set some prices at levels different from those of Ramsey in that market. Moreover, it is
clear that at least some prices at the retail level (for example, on-net and off-net calls) are even
now not set at levels consistent with a Ramsey pattern of prices. Furthermore, as profitmaximizing firms, the MNOs will set prices using firm-specific elasticities of demand and their
relative levels. We think these will clearly differ from the analogous market elasticities, not
least because the retail market is not fully competitive. Consequently, even if (which is not the
case) we thought that Ramsey pricing was in principle superior to the fair charge approach
which we advocate, we would nevertheless find it impossible to recommend Ramsey pricing on
this occasion.
2.517. Our second objection to Ramsey pricing arises from the formidable problems
associated with obtaining reliable estimates of the elasticities of demand which are the basis of
the Ramsey approach. It is clear from the evidence submitted to us during our inquiry (discussed in detail in Chapter 8) that no unanimityindeed, considerable disagreementexists
among the main parties and their economic advisers as to the correct values for the various
different elasticities that are involved. Moreover, a large number of elasticity estimates are
required to establish Ramsey prices in the mobile sector as a whole (including those for text
messaging and mobile internet), which compounds the problems inherent in obtaining a reliable
set of estimates. This means that, if Ramsey were to be used as the basis of a regulated price,
we believe that the DGT would experience considerable difficulty in achieving consensus on
120

reliable estimates of the relevant elasticities forming the basis of the Ramsey price structure. No
consensus was reached among the parties submitting evidence on this matter during our
inquiry. There would in our view be considerable scope for disagreement if these exercises
were to become part of the regulatory process, giving rise to prolonged disputes between the
DGT and the MNOs and resulting in an increase in the overall costs of regulation and delay in
the realization of the regulatory benefits to consumers. Oftel, indeed, told us that the informational requirements were too onerous for Ramsey pricing to provide a reliable basis for
regulated termination charges.
2.518. Third, we believe that a regulated price based on Ramsey principles would lead to
the sort of distributional inequities which we have discussed earlier in this chapter (see paragraphs 2.388 to 2.400) and which result in some consumers unfairly subsidizing other consumers.
2.519. The considerations set out in paragraphs 2.510 to 2.518, together with evidence discussed earlier in this chapter and in Chapters 6 and 8, lead us to believe that the objections to
using Ramsey principles to set an average regulated mobile termination charge are conclusive.
In summary, they are:
(a) the absence of consensus regarding absolute and relative price-elasticities, determination

of which is necessary for the computation of Ramsey prices (see Chapters 8 and 9);
(b) the extreme disagreement among the parties as to the nature and extent of fixed and

common costs, determination of which is also necessary for the computation of Ramsey
prices (see Chapter 7);
(c) the implausible outputs of some of the models claiming to compute Ramsey prices (see

Chapter 9 and Appendix 9.1);

(d) the inconsistency between actual retail price patterns observed and the retail price

patterns claimed as naturally resulting, in accordance with Ramsey principles, from

competition in the absence of regulatory constraint (see paragraphs 2.122 to 2.131);
(e) the uncertainty as to the actual level of prices, having regard to the difficulty of allocat-

ing subscription charges between handsets and calls (see paragraphs 2.177 to 2.187);
(f) our doubts as to the complete effectiveness of retail competition necessary for the

emergence of Ramsey pricing patterns even in theory (see paragraphs 2.210 and 2.211);
and
(g) our inability, in any event, to recommend Ramsey prices in the retail market, if our pro-

posals for regulation relate to one element only, namely termination charges (see paragraph 2.516).
2.520. We should say that the price cap that we advocate is based on an average of termination charges. It is not our recommendation that all termination charges should be set at the
same rate irrespective of, for example, time of day; neither do we seek to set individual differentiated prices by time of day or week. Consequently, our own proposals for regulation of termination charges would allow the MNOs, if they wished to do so, to set Ramsey-type prices in
the wholesale (termination) market. The cap effectively ensures that, on average, callers will
pay the costs they cause the MNOs to incur. Further, we believe that to impose individual time
of day charges would be unduly intrusive. This approach is consistent with that taken by the
DGT in his proposed licence modification. Thus, the MNOs would be free to vary termination
charges by time of day, day of the week etc, according to the various relevant elasticities,
subject to an overall average charge cap that will keep the MNOs from earning excessive
profits on call termination.
121

2.521. We also considered EPMU as an alternative basis on which to allocate the fixed and
common costs of the MNOs, as proposed by the DGT. Under EPMU, prices are set so as to
recover fixed and common costs by marking up the fixed and common costs in equal proportion to the LRICs of each service.
2.522. The MNOs argued strongly that use of EPMU would lead to a less efficient outcome than the application of Ramsey principles, because, they said, mobile services at the retail
level were more price-elastic than call termination and, for that reason, ought to carry a lower
proportion of fixed and common costs. They said that there was no indication that LRIC plus
EPMU would represent any less of a regulatory burden than a Ramsey approach; that all regulation was costly, but arbitrary and unnecessary regulation would produce welfare consequences that damaged rather than benefited consumers. Vodafone said that the Act required us
to propose a form of regulation, if any were necessary, which was best suited to achieving
efficient prices and to ensuring that all reasonable demands for mobile services were met.
Orange warned of the detrimental impact on the industry when regulation was inappropriately
applied or wrongly set; sub-optimal pricing would, it said, give rise to both allocative and
dynamic inefficiency, as investment might be inhibited or distorted.
2.523. EPMU is equivalent to the special case of Ramsey pricing when all the superelasticities of the goods or services in question are equal, since EPMU and Ramsey would then
give the same results. Therefore, if the evidence on the elasticities of the various mobile services in this case does not show large differences, the error in using EPMU rather than in
choosing Ramsey prices would appear to be small, particularly if, as we believe, fixed and
common costs are small. However, we have already concluded that the value of the elasticities
cannot be reliably established and we could not, therefore, establish that they were not different
from each other. We also note that EPMU is not consistent with our preferred approach (that is,
that termination charges should reflect the costs caused by the calling party) and, in this sense,
it seems to us to be somewhat arbitrary. We note, however, that in this particular case, the costcausation principle yields a very similar answer to that which results from the application of
Oftels EPMU proposal. There are two reasons for this: first, the DGT has allocated no customer acquisition costs to termination charges, and nor have we; and second, the DGT has in
effect allocated network fixed and common costs to termination charges using routing factors
and we have done so explicitly.

Implementation of charge control

2.524. Before we can reach a formal conclusion on licence modifications, we have to consider the following: whether we should recommend that termination charges be reduced immediately to the fair charge or whether there should be a glide path; the period of our proposed
price control; whether there should be one or more price caps governing off-net and fixed-tomobile termination charges; and whether the same controls should apply to all the four MNOs
equally. All these matters must be considered in the light of our existing legal obligations and
with the evolving legal framework within which we are operating in mind.

Finding on whether the adverse effects identified can be remedied or prevented by a modification of the MNOs licences
2.525. In our view, this question requires us to consider whether a licence modification can
remedy or address the adverse effects both as a matter of fact and as a question of law. In principle, and as a matter of fact, the excess of termination charges over costs can be prevented by a
licence modification. The modification in question is one which caps termination charges. That
this modification can be effected is demonstrated by the feasibility of the caps currently in force
in the licences of Vodafone and O2, albeit that those caps are at levels that we consider to be
above the fair charge.
122

2.526. We have found the question of the actual design of the licence modifications difficult. Although this is a difficult question generally, the difficulty is made more acute by legal
circumstance. The model licence modification presented by the DGT sought to regulate termination charges within the period to 31 March 2006. During the course of the inquiry it has
become clear that the current licences of the MNOs will be abolished by 25 July 2003. Consequently, any licence modification that we are able to recommend will have a very short life.
Indeed, it is likely to have effect for a few months at best. Nonetheless, it is our view that a
licence modification which lasts but a few months could remedy the problem of excess termination charges for the period of the duration of the licence and that the sums involved are
sufficiently material to warrant such a modification.
2.527. In reaching this conclusion we have paid careful regard to the effect of the new EC
Directives1 and to the obligation of member states in the period for implementation of those
directives. However, and having regard to the facts that we have found, we do not believe that
there is any legal rule which must prevent any licence modification at all from being introduced
in the period to 25 July 2003. We recognize that there may be some licence modifications that
are not open to us. But we do not think that the new EC Directives rule out a price cap having
effect for the period to 25 July 2003. Indeed, we note that many of our findings are consistent
with the structure of price regulation under the new EC regime. In particular, we have identified
wholesale mobile termination as four separate markets. Because each MNO is a monopolist on
its own network, each MNO has significant market power as understood under the new regime.
Our approach to remedying the problem of excess termination charges is cost-oriented price
regulation, and this sort of price regulation is expressly envisaged by Article 13 of the Access
Directive.
2.528. At the same time we recognize that, before ex ante regulation can be imposed under
the Access Directive, other conditions have to be satisfied. We do not say that our conclusions
in this report can determine whether there will be price regulation of termination charges after
25 July 2003. Our view, as stated above, is that a cap on termination charges imposed on the
MNOs through licence conditions which must fall away is not prohibited by virtue of the new
directives.
2.529. There remains then the question of whether a suitable licence modification can be
designed and given effect as a matter of fact and law. In discharging our responsibilities under
section 14 of the Act, we have to state whether we believe that the adverse effects can be
remedied or prevented. Under the Licensing and Interconnection Directives, any modification
that we propose must be within the range of modifications specifically permitted in relation to
individual licences, and must in addition be objectively justified in relation to the services concerned, non-discriminatory, proportionate and transparent.
2.530. One possible solution to the adverse effect identified is the immediate and complete
reduction of termination charges to the level of the fair charge. We do not believe that there are
any objections in principle to such an approach. However, we are concerned about the probable
effects that such an immediate and complete adjustment might have on the mobiles sector, and
therefore on consumers. It would, in our view, be disruptive and create an unacceptable range
of adjustment costs to consumers and the MNOs. This market disruption would not be in the
interests of consumers. We acknowledge that the MNOs have embarked upon contracts with
third parties based upon the continuation of a particular level (albeit excessive) of termination
charges; we also acknowledge that there is likely to be a process of revenue adjustment from
termination charges to other sources of MNO revenue and that this is unlikely to be achieved
overnight; or, if it were attempted, it would lead to significant disruption. In our view there is a
risk in moving immediately to the level of the fair charge. The risk is that such a remedy would
not provide the greatest advantage to consumers and that in effect some of the adverse effects
we have identified would not be remedied. In short an immediate and complete reduction of
1

For a general description of the new directives, see Chapter 4.

123

termination charges would have undesirable side-effects which we are entitled to take into
account in forming a judgment as to whether a licence modification would be capable of
remedying the adverse effects found. We think that the scheme of domestic and EC legislation
is sufficiently wide to allow us to take these side effects into account in recommending a
remedy. Accordingly, in the light of this risk, we propose an immediate licence modification
which takes full account of those adverse effects which can be remedied and prevented within
the existing regime.
2.531. We noted Oftels view in the September 2001 review that a glide path provided
greater incentives to the MNOs to reduce costs than an immediate reduction of charges to costs,
since the regulated MNOs were then able to keep the benefits of efficiency gains for a period
before consumers captured these gains through lower prices. Oftel also thought that a glide path
would produce less disruption in the markets. The MNOs expressed concerns about the disruption to the capital markets and the retail mobile market, the threat to their finances and the
revenue effect of a one-off price cut.
2.532. Some of the FNOs submitted views to us on this matter. In summary, these were
that (a) the MNOs had been setting call termination charges above cost for some time and it
was therefore appropriate that price distortions in call termination should be corrected with
immediate effect; (b) any market disruption would be minimal: any increase in the retail price
that was justified as a result of the loss of revenue from termination charges should not be
significant, even on a highly conservative estimate; (c) fixed network users and mobile network
users were by no means identical and it was inequitable that the former should continue to subsidize the latter; (d) mobile termination did not need to be above cost in order to drive penetration; and (e) the MNOs should be given no incentive for delay in implementing charge cuts.
2.533. BT proposed a middle course. It acknowledged the view that regulation should not
undermine incentives for operators, and believed that incentives would be diminished following
a one-off reduction down to cost. At the same time, it believed that the MNOs were earning
well in excess of costs, brought about by efficiency gains but also by the generosity of the
current price cap. BT therefore proposed an initial reduction to a level which was 50 per cent
between current charges and the cost of termination, followed by a glide path.
2.534. As we noted in paragraph 2.526, it only became the case that any licence
modification we propose would fall away by 25 July 2003 once the inquiry was under way. As
we stated in that paragraph, when the references were made, the DGT envisaged that licence
modifications would be in effect for the period until 31 March 2006. Oftel told us that a price
control should be set for the longest period for which it was possible sensibly to forecast market
conditions. It said that it did not anticipate any material change in competitive constraints on
mobile termination over the next four years. It would not, it said, be precluded from launching a
market review at an early stage and possibly modifying obligations on the MNOs in the light of
its analysis, were unanticipated developments with major competitive effects to occur in the
intervening period.
2.535. In our view, a period to 31 March 2006 is a more suitable period for regulatory
assessment than a shorter period such as that to 25 July 2003. Consequently, in carrying out our
inquiry, we have continued to use a longer period for the purposes of our analysis of the mobile
market and the likely developments in that market.
2.536. We have considered all these matters very carefully. We believe that the analysis
that we have carried out enables us to take a view of the levels at which charges should be set
for the period to 2006. Consequently, and in view of the side effects that we believe would be
brought about by an immediate reduction in termination charges to the fair charge (see paragraph 2.530), and notwithstanding that any modification that we propose for the period beginning 25 July 2003 can have little more than persuasive effect, we have decided to state our
views on the levels at which we think termination charges should be set to 2006.
124

2.537. Consequently, the modification that we propose for the period to 25 July 2003 is,
effectively, a partial elimination of the adverse effects identified, which allows adjustments to
limit the impact of undesirable side effects. The modification is a 15 per cent reduction in real
terms in the level of the average termination charges of each of the four MNOs, to take effect
within the period 1 April to 25 July 2003. This is to be followed by a reduction in the average
termination charge equivalent to RPI15 for O2 and Vodafone, and RPI14 for Orange and
T-Mobile for the period from 25 July 2003 to 31 March 2004; and then further reductions of
RPI15 for O2 and Vodafone, and RPI14 for Orange and T-Mobile, in each of the years
1 April 2004 to 31 March 2005 and 1 April 2005 to 31 March 2006. The effect of these reductions will be to bring termination charges down to the level of the fair charge by the end of the
period, as shown in Table 2.12. We believe that this approach is not inconsistent with the
specific provisions and general requirements of applicable domestic and EC law.
TABLE 2.12 The CCs proposals for termination charges to 2005/06
2002/03

2003/04
Apr-July

2003/04
July-Mar

2003/04
Full year

Initial cut of 15% and RPI15 for the remaining 3 periods

900/1800 MHz
2000/01 prices
Out-turn prices

9.0
9.4

7.5

6.5

9.6
10.0

7.9

7.1

2005/06

Unrounded 14.6
6.8

Initial cut of 15% and RPI14 for the remaining 3 periods

1800 MHz
2000/01 prices
Out-turn prices

2004/05

5.6

4.7

Unrounded 13.5
7.3

6.1

5.3

Source: CC.

One or more charge caps

2.538. There are two aspects to the question of how many price caps we should set. The
first is whether we should set individual caps for each of the four MNOs or whether they should
all be subject to the same cap. We considered a number of reasons why there should be different caps: first, because of the different costs of the combined 900/1800 MHz and the 1800 MHz
operators; second, because of the different market shares of the MNOs; and third, because of
the different costs of capital of the MNOs. We discussed these matters earlier (see paragraphs
2.241, 2.275 to 2.280 and 2.301 to 2.307) and have decided that the only differences we should
allow are those that are absolutely outside the companies control: that is, the differences due to
the allocation of spectrum. Accordingly, we set two price caps, one for O2 and Vodafone and
another for Orange and T-Mobile.
2.539. The other aspect of the question relates to whether we should set separate caps on
termination charges for fixed-to-mobile and off-net calls. As we have already noted, the price
control we are recommending does not seek to regulate the level at which the MNOs set call
termination charges according to the time of day or day of the week, or according to where the
call originates. However, we considered that it was desirable to preclude the possibility that
very different termination charges would be set, respectively, for fixed-to-mobile and off-net
calls (for example, through most of the charges being loaded on to one or other type of call),
albeit that the average charge lay within the constraints of an overall price ceiling.
2.540. We accordingly recommend that two charge caps be set (in each case, at different
levels for 900/1800 MHz operators and for 1800 MHz operators), one to control fixed-tomobile, and the other to control off-net, call termination charges. Such an arrangement would
provide reassurance to consumers. Moreover, as already noted, it would not prevent the MNOs
125

from entering into bilateral agreements with each other to set lower prices, but if they did so,
those prices would be beneath an overall price ceiling. We do not believe that the MNOs could
reasonably object to a separate off-net cap, since such a cap places no pressure on them over
and above the fixed-to-mobile cap.

Other matters relevant to the implementation of a charge control

Ported numbers
2.541. Two further matters were raised by Oftel. The first matter concerns the handling of
ported numbers in a price control. For numbers ported into a network from another operator,
the recipient MNO receives the termination charge of the original MNO minus a small transit
charge. If the termination charges of all the networks were the same, the revenue that the
MNOs would receive from ported-in calls would be, at all times, less than the standard termination rate. For an MNO with a higher termination rate than another MNO, it is further
disadvantaged in that, for some ported-in calls, it will receive the lower termination rate, minus
the transit charge. For an MNO with a lower termination rate than another, it may actually
receive more for a ported-in call than its standard termination rate, if the termination charge of
the original MNO was more than its own termination rate plus the transit charge. O2 and
Vodafone have excluded ported calls from the calculation of the termination charge on the
grounds that they have no control over them, leaving them with higher revenue than is allowed
in the cap whenever calls are ported in from an MNO with higher charges.
2.542. There are a number of possible mechanisms for taking ported calls into account as
part of a price cap. The two of most relevance for our purposes in this inquiry are:
(a) that all calls ported out to another MNOs network should be included in the calculation

of the average termination rate. This is the current regime, although not implemented by
O2 and Vodafone; and
(b) that where any MNO sets a termination charge for calls to a former customer of the

MNO who has ported to another MNO, the donor MNO should charge no more for termination of such a call than it would charge for a similar call to a customer who has not
ported.
2.543. Given that the price cap for 1800 MHz operators is higher than that for combined
900/1800 MHz operators, there will be a systematic bias against the 1800 MHz operators if the
current regime of counting only calls delivered to a particular MNO is continued. Oftel told us
that there were practical difficulties with option (b) above, and proposed that the current position be formalized, but with the option for the DGT to impose (b) if he found evidence of
manipulation. Orange told us that it saw no valid reason to exclude terminating calls to ported
numbers from a termination charge control. O2 said that ported numbers should be excluded, as
is the current practice. Vodafone made the point in paragraph 2.542(b). We take the view that,
given the change of regulatory regime that must come into effect as from 25 July 2003, the
DGT has an opportunity to re-examine the matter in that context and that we do not need to
reach a conclusion on this matter here.

Target average charge

2.544. The second matter concerns a proposal, set out in Annex 6 to the DGTs review
Statement of 26 September 2001, for a new method of calculating the target average charge.
Oftel told us that it had concerns about maintaining the current charge control mechanism,
which it said provided scope for the MNOs to avoid charge reductions. The MNOs however
126

told us that their flexibility would be reduced if it was changed. Having considered this matter,
we have come to the view that, as with ported numbers, this matter should be left to Oftel to
consider further when the current licence modifications expire in July 2003.

2G/3G calls
2.545. The four MNOs and Hutchison 3G expressed concern to us that, because voice calls
over 2G technology will be indistinguishable from those over 3G, any charge controls on 2G
termination charges would necessarily regulate 3G pricing. T-Mobile argued that regulation of
3G pricing was not permitted under EC law and was not even contemplated by the EC draft
Recommendation on markets. Therefore, any charge control which de facto regulated 3G
termination charges would be unlawful. Hutchison 3G expressed concern about the possibility
that its charges would become subject to control, notwithstanding that 3G calls were not formally included in such controls. As we have already noted (see paragraph 2.40), Oftel
announced in its September 2001 Statement that it did not intend to include 3G voice calls in its
current proposals for regulation.
2.546. Oftel told us that it expected O2, Vodafone, Orange and T-Mobile to charge a
melded rate, based on the actual volumes of each type of call termination, which we note that
these MNOs said they could estimate. We asked Oftel whether the proposal to regulate 2G calls
but not 3G calls would enable the MNOs to levy very high charges on 3G voice calls which,
with the regulated charges on 2G calls, would result in an average price well above the capped
2G termination charge, thereby undermining the purpose of the charge control. Oftel told us
that it did not have sufficient information about 3G costs to enable 3G calls to be included in its
charge control proposals for 2G, but that if the level of charges for 3G call termination raised
concerns, then it would investigate.
2.547. As discussed in paragraphs 2.148 and 2.149, we are limited by our terms of reference to making recommendations concerning only 2G call termination charges. However, we
note that Hutchison 3G will have a monopoly over the termination of voice calls to its network.
To the extent that mobile numbers are ported into Hutchison 3G, then Hutchison 3G will
inevitably be subject to price limitations imposed by 2G regulation. We do not regard that as a
serious impediment to Hutchison 3Gs roll-out. During the period of some three months in
which our recommended licence modifications, if accepted, would have effect, we would not
expect them to have any significant detrimental effects on the MNOs; (in any event, the four
incumbent MNOs have announced a delay to the launch of their 3G services, so the number of
voice calls over 3G in the period to 25 July 2003 is likely to be small). The DGT has the option
of reviewing his stance on 3G voice call termination during that period and considering
whether any regulation of termination charges for voice calls to 3G would be desirable or
feasible.

Impact of price cap

2.548. We now turn to the impact which our recommendations for a price cap might have
on welfare generally and on consumers and the MNOs. In paragraphs 2.550 to 2.558, we discuss the welfare impact; in paragraphs 2.559 to 2.563, we consider the impact on MNOs; and in
paragraphs 2.564 to 2.568, we look at the impact on consumers. Our overall conclusion is at
paragraph 2.569.
2.549. It was put to us by the MNOs that our recommendations for a price cap would result
both in financial loss for them and higher retail prices for consumers. We discuss the exercises
we carried out to assess the likely impact of a reduction in termination charges on these two
groups before reaching our conclusions on licence modifications. This involved two separate
sets of calculations, summarized below, and which are described in detail in Chapter 9.
127

Impact on welfare
2.550. We analysed what the level of welfare would be if the termination charge were
regulated (that is, if a price cap were imposed), and compared this with the level of welfare that
would result with no such price cap. Any change in welfare would be the net effect of the
change in consumer surplus and the change in producer surplus. Consumer surplus is a measure
of how much consumers value the consumption of a good, over and above what they pay for
that good. Producer surplus is effectively the difference between total revenue and total costs.
2.551. CRA, DotEcon, Frontier Economics and Dr Rohlfs provided us with models which
investigated the effect on welfare of the imposition of a cap on termination charges as compared with various alternatives. A discussion of these models and their results is set out in full
in Chapter 9 and Appendix 9.1. In summary, Dr Rohlfs (on behalf of Oftel) found welfare gains
from reducing termination charges, while CRA, DotEcon and Frontier Economics (each working for an MNO) found that departures from Ramsey-optimal termination charges (which they
claimed were at or above current levels) resulted in welfare losses. We noted that the analyses
carried out by DotEcon and Frontier Economics must result in a welfare loss, as they compared
the optimum derived from their models with a constrained situation (see paragraphs 9.33 to
9.36 and 9.42 to 9.44). We noted that the way in which the CRA model was set up and the way
in which it incorporated the effects of externalities made it more likely than not that there
would be a loss in welfare as a result of a reduction in the termination charge.
2.552. We analysed the effect on welfare of a regulated termination charge in two ways:
(a) first, we used the parties models to compare the results from a regulated termination

charge with the results from two possible unregulated situations. To do this, we used the
parties models under a set of base case assumptions. However, in both these situations
(regulated and unregulated) the models effectively set retail prices according to Ramsey
principles; and,
(b) second, we looked at the effects of using the termination charge as calculated by

Rohlfss Ramsey-based model against Rohlfss model of an unregulated industry, where

prices in the unregulated situation are not necessarily Ramsey-based. In this latter comparison we used Rohlfss models only, because only Dr Rohlfs provided a model of how
MNOs might be expected to operate in a completely unregulated industry.
2.553. We do, however, have a number of reservations about using the models referred to
in the previous paragraph. First, with the exception of Rohlfss unregulated model, they are
based on Ramsey prices at the retail level and we have already noted that we do not believe that
such prices occur or would occur. Second, the models make use of elasticity estimates which
we believe are unreliable. Third, as we have already stated, we believe that costs should be
allocated on the cost-causation principle, where the costs of termination are allocated to those
users that generate the costs, unlike the MNOs models which allocate the costs of the MNOs
operations on the basis that they are mostly fixed and common costs. For these reasons, we
believe that results from the MNOs models should be viewed only as approximations of
changes in welfare.
2.554. In our first approach, we looked at two scenarios for unregulated termination
charges, which are set out in paragraphs 9.60 and 9.61. We consider the more relevant scenario
to be where unregulated termination charges remain at current levels in nominal terms. Broadly
speaking, we ensured that the R-G factor implied by the models was around 1.5 (although, as
we said earlier, we believe that 1.5 is probably something of an overestimate) and we note that
under these assumptions the DotEcon, Frontier Economics and Rohlfs models all calculate the
optimal fixed-to-mobile termination mark-up as being below current levels (see Table 9.12).
2.555. Overall, our analysis using this approach shows that, when a regulated scenario is
compared with an unregulated one, using the models of DotEcon, Frontier Economics and
128

Rohlfs, the result is an increase in welfare as a consequence of regulation. Regulating termination charges from current levels produces welfare gains (in net present value (NPV) terms)
of around 700 million (with an immediate reduction in charges) and around 325 million
(with a progressive reduction in charges) over the three-year control period. For the CRA
model, a move from the unregulated to the regulated termination charges results in a loss in
welfare. However, we and other parties have reservations about CRAs model and these are set
out in paragraphs 9.48 to 9.50 and 9.85 to 9.86.
2.556. The MNOs criticized the approach described above. Their criticisms are summarized fully in Chapter 9. Many of the criticisms concerned the methodology we employed.
Some of the MNOs suggested that we should follow the methodology employed by DotEcon
and Frontier Economics in their welfare analyses, that is, comparing the unconstrained optimum with the outcome derived by constraining the fixed-to-mobile termination charge at the
unregulated scenario level. However, as noted earlier, the unconstrained situation must, by
construction, produce a greater level of welfare than the constrained situation. Given this, use
of the DotEcon and Frontier Economics methodology would result in a gain in welfare from
regulation because the fixed-to-mobile prices used in our welfare analysis (as shown in Table
9.12) are optimal (under our base case assumptions) and moving to the higher prices considered
in the unregulated scenario must, therefore, reduce welfare (as we are moving away from the
optimum).
2.557. Our second approach was to use Rohlfss model of regulated and unregulated
behaviour to carry out the same comparative exercise. This showed large gains over the control
period from regulating termination charges (in NPV terms, 3 billion from an immediate
reduction in termination charges and about 1.4 billion from a progressive reduction in
charges). We believe that these gains are probably overestimates because we do not expect that,
in the absence of regulation, termination charges would rise to levels projected by Rohlfss
unregulated model. We do, however, believe that the basic conclusion from the Rohlfss model
analysis is correct, that is, that there will be gains not losses from regulating termination
charges.

Conclusion on welfare
2.558. For the vast majority of the simulations that we ran, it was the case that there was a
welfare gain from regulation of termination charges, and in most cases this gain was large. It is
clear that the magnitude of the gain depends very much on the comparison made and the model
(and assumptions) used. By using the MNOs models with our preferred assumptions, welfare
gains are shown of between around 325 million and around 700 million over a three-year
period. While acknowledging that this simulation cannot be regarded as precise, we consider
that the fact that most of the models produce large welfare gains under various assumptions
lends strong support to the argument that termination charges should be regulated.

Impact on MNOs
2.559. We turn next to consider the effects of price caps on the finances of the MNOs, on
the changes in the prices likely to be charged to mobile customers (in the form of changes to
average bills) and on subscriber numbers. We begin with the financial impacts on the MNOs of
charge controls on termination charges.
2.560. In order to assess the financial consequences of a price cap for MNOs and consumers, we have compared what would happen were a price cap to be imposed with what might
happen otherwise. We have considered two counter-factuals: first, that termination charges
remain constant at their current levels, and second, that termination charges rise to 17 ppm. The
129

latter reflects a level of charges that could be reached in an unregulated environment, according
to Oftel and Vodafone. However, we regard the scenario of prices remaining constant as more
realistic, and hence we have decided to concentrate on that option for our comparison.
2.561. Bearing in mind the views of the MNOs (set out in Chapter 9), it is important to
ascertain the extent, if any, to which the reduction in revenue resulting from the imposition of a
price cap on termination charges would oblige them to rebalance their other retail prices in the
ways they have described, or to try to recover the revenue in other ways. Naturally, the nature
and extent of the rebalancing will depend on the amount of revenue to be recouped (and, as discussed earlier in this chapter, the degree of competition in the retail market). We have found it
convenient to refer to this rebalancing of the loss of revenue from capping termination charges
with its retrieval through raising prices in the retail sector as the waterbed effect.
2.562. Our calculations suggest that, with no waterbed effect, the combined reductions in
revenue of the four MNOs, in NPV terms, incurred as a result of capping termination charges
would be between 1.5 billion and 2 billion (in 2001/02 prices). (These calculations, together
with others positing different glide paths, are shown in Tables 9.26 and 9.27, and are discussed
in paragraph 9.124.) These reductions in revenue account for about 6 per cent of the NPV of the
MNOs aggregate turnover.
2.563. We believe that, in practice, these reductions in revenue are overestimates, first
because they assume that the MNOs turnover will remain constant at 2001/02 levels, and
second because, in our view, there will be a waterbed effect, ie most of the reductions in
revenue from termination charges being capped will be recovered from the retail market. It is
difficult to be precise about the size of this effect but, as we have found that the retail market is
not fully competitive, we expect that the waterbed effect may not be a full 100 per cent. We
estimate that it would be possible for the MNOs to recover much, or perhaps all of, the lost
revenue which arises from capping termination charges by delaying price cuts at the retail level.
We consider this possibility more fully in the following paragraphs. The MNOs reductions in
revenue are halved with a 50 per cent waterbed effect and, obviously, there are no financial
impacts on the MNOs with a full waterbed effect.

Impact on consumers
Price impacts for mobile customers
2.564. We start by estimating the size of retail price changes which would occur if there
were a full waterbed effect. With a full waterbed effect and unregulated termination charges at
current levels, prices would rise between 4 and 7 per cent in 2003/04. We also express the price
increases as a compound percentage change over three years. This method gives an annual
average increase in prices of less than 3 per cent. With a 50 per cent waterbed effect, these
price increases would be halved, and of course there would be no impacts on the prices to
mobile customers with no waterbed effect, that is, where the MNOs did not recover lost
revenue from a fall in termination charges by increasing retail prices to mobile customers.
2.565. These possible price rises need to be set in context. We have done this by comparing them with the price changes projected by the MNOs in their business plans. Orange did
not provide comparable figures but it told us that it expected outbound prices to fall in real
terms. This view is consistent with the other MNOs business plans, which projected real
reductions in outbound calling prices. We calculated the annual average real price reductions in
the MNOs business plans for the period 2004 to 2006, taking account of Oranges view, calculated as a simple average for all the MNOs. This showed expected real price reductions of
5.5 per cent a year. Comparing this reduction with our estimate of the price increases that
would (assuming a full waterbed) follow a termination charge reduction implies that average
retail prices would still fall but by, on average, about half of the rate as shown in the MNOs
business plans, that is, by about 3 per cent a year.
130

2.566. Orange and T-Mobile told us that it was irrelevant to the issue of whether termination charge controls were warranted that outbound price rises resulting from regulation of
termination might or might not be offset by planned outbound price falls. They said that the
possible rebalancing effects on outbound call prices should be assessed on a stand-alone basis,
against the counterfactual of what would otherwise occur, absent the regulation. They stated
that this would enable the impact of the regulation itself to be assessed, so that a decision could
be made as to whether the regulation would lead to greater benefits than costs.

Volume impacts of regulation

2.567. As we believe that, overall, there will be a reduction in average retail prices, we do
not expect a price cap to lead to any significant reduction in the number of subscribers. If, however, handset subsidies were reduced or removed, we believe that, in the first instance, most
marginal subscribers would not replace their handsets so often. We also believe that the
acquired habit of using mobile phones means that such subscribers are now less marginal than
they were when they first signed on to the network and that people already owning mobile
phones are unlikely to leave the network unless their handset is lost, stolen or broken. The
MNOs have the option to offer marginal subscribers cheaper packages to induce them to stay
on the network once that happens. Thus, even if handset subsidies were reduced, we do not
expect a large reduction in the number of subscribers on the four networks.

Conclusion on impact on consumers

2.568. We conclude that it will be possible in principle for the MNOs to more or less fully
rebalance their prices, following our proposed reduction in termination charges, without instituting average retail prices increases. Thus, we believe that any increase in retail prices
following the implementation of our recommendations would be a commercial decision on the
part of one or more of the MNOs, and not the result of our recommendations. This is because
we believe that the theoretical increase in retail prices ostensibly necessitated by the reductions
in termination charges that we are proposing will be more than offset by reductions in other
prices as set out in the MNOs business plans. It follows that we believe that our decision to
regulate termination charges will have no major impact on the number of mobile phone customers.

Overall conclusion
2.569. In summary, we conclude that a price cap on termination charges to bring them
down to the fair charge will benefit consumers by reducing the price of fixed-to-mobile and offnet calls and, depending on how the MNOs respond, could be broadly neutral so far as concerns
the effect on the MNOs financial viability. As their own business plans indicate, it is sufficient
that the MNOs slow the decline of retail prices in order to recover the revenue loss from
reduced termination charges that we are recommending, but average retail prices should still
fall.

Proportionality and effectiveness

2.570. Some of the MNOs submitted that our proposed charge cap remedy was disproportionate in relation to the alleged detriment. We do not accept this. We are obliged by virtue
of various provisions of the Licensing and Interconnection Directives1 to act in accordance with
the principle of proportionality. This is closely related to our duty to decide whether the
1

See, for example, Article 3(2) of the Licensing Directive.

131

existing licences, if left unamended, operate or may be expected to operate contrary to the
public interest. It would have been open to us to have assessed that the element of excess we
identified was so small as to render any intervention disproportionate and thus the continuation
of the licences unamended would not operate or be expected to operate contrary to the public
interest. On this, we are satisfied that the element of excess we identified is sufficiently high to
warrant a licence modification requiring an initial reduction in charges.
2.571. Proportionality arises again when assessing the appropriate licence modification to
recommend to the DGT. The object of such a proposed modification is to prevent or remedy the
adverse effects identified by the CC. Such modification can go no further than that statutory
objective; any recommended modification, albeit only advisory, must accord with the principle
of proportionality, that is, it must be the least intrusive means of preventing or modifying the
adverse effects. We considered this very carefully in our choice of modification and in relation
to the extent and timing of the reductions in charges we proposed and their likely impact upon
each party.
2.572. We believe that none of the other proposed remedies, whether alone or in combination, would have been effective to remedy the adverse effects associated with above-cost termination charges. In our view, only a price cap on termination charges is adequate for this
purpose. As we have shown in the discussion of the impact of our proposals on consumers and
the MNOs in paragraphs 2.559 to 2.568, we have fulfilled our obligations under section 3(2) of
the Act by addressing the detriment to consumers which above-cost termination charges have
brought about, while ensuring that the matters to which we are to have regard under section
3(1) have been secured.
2.573. O2, Orange and T-Mobile put it to us that our recommendations would prejudice
them in relation to Vodafone, whom they considered to be the strongest player, with both a
greater capacity to absorb the effects of the reductions in revenue brought about by the price
cap than the other three MNOs and the incentive to exploit that advantage at those companies
expense. For example, Vodafone might, they suggested, maintain its retail prices at current
levels at a time when the other MNOs were forced, by virtue of the reduction in termination
revenue, to raise theirs.
2.574. We accept that Vodafone may be in a more advantageous competitive position;
however, to the extent that it is already the strongest player on a number of measures, Vodafone
could have exploited this strength with a view to gaining a competitive advantage at any time;
we have not observed that it has chosen to do so. We note also that the two more recent entrants
have succeeded in gaining market share at the expense of the established operators and we
would expect that, in a fairly competitive market, each MNO would continue to find ways to
win (and retain) customers at the expense of its competitors. We do not, therefore, accept that
our proposals for bringing termination charges down to the fair charge will adversely affect the
competitive position of O2, Orange or T-Mobile in relation to Vodafone. Further, as noted in
paragraph 2.565, we do not believe that any re-balancing of prices consequent on our proposals
will necessitate increases in average retail prices; on this basis, the concerns of the MNOs are
misconceived.
2.575. The MNOs have also claimed that the commercial effects of our conclusions and
remedies in this inquiry are particularly far-reaching, and that the detriments if we capped call
termination charges at too low a level would be much greater than if we capped them at too
high a level. In our view neither claim is valid.
2.576. In contrast to the outcome of other regulatory inquiries involving monopoly considerations, we place no constraints on the prices that the MNOs charge to their customers, or
on their profits. Further, because of the continuing decline in retail prices that is observed in the
recent past and planned for in the future, the reduction in termination charges we recommend
will necessitate little or no increase in the MNOs retail prices in order to preserve their
132

financial position. And our conclusions and remedies do not cover the 3G business, which all
the MNOs have told us is where their long-term future lies. Thus, the effects of our conclusions
and remedies are much less far-reaching than in other inquiries.
2.577. The main effect and purpose of our remedy is to remove an unjustified subsidy currently paid by FNOs or their customers to the benefit of the MNOs or their customers. If we
were to cap termination charges at too high a level, all or part of this subsidy would continue. If
we were to cap them at too low a level (which we do not believe we have done), the effect
would be to create an unjustified subsidy in the reverse direction. As noted, we do not seek to
do this; neither do we believe we have done so.

Formal recommendations on licence modifications

2.578. We recommend that the four MNOs should each be subject to two termination
charge caps, set at the same level, one to control the level of GSM call termination charges for
fixed-to-mobile calls and the other to control the level of GSM call termination charges for offnet calls.
2.579. We recommend (see Table 2.12) that the GSM call termination charges of O2 and
Vodafone:
(a) be reduced by 15 per cent in real terms before 25 July 2003; and
(b) be subject to further reductions of RPI15 from 25 July 2003 to 31 March 2004, RPI15

from 1 April 2004 to 31 March 2005, and RPI15 from 1 April 2005 to 31 March 2006.
2.580. We recommend (see Table 2.12) that the GSM call termination charges of Orange
and T-Mobile:
(a) be reduced by 15 per cent in real terms before 25 July 2003; and
(b) be subject to further reductions of RPI14 from 25 July 2003 to 31 March 2004, RPI14

from 1 April 2004 to 31 March 2005, and RPI14 from 1 April 2005 to 31 March 2006.

133